CN108021856A - Light for vehicle recognition methods, device and vehicle - Google Patents
Light for vehicle recognition methods, device and vehicle Download PDFInfo
- Publication number
- CN108021856A CN108021856A CN201610978015.4A CN201610978015A CN108021856A CN 108021856 A CN108021856 A CN 108021856A CN 201610978015 A CN201610978015 A CN 201610978015A CN 108021856 A CN108021856 A CN 108021856A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- taillight
- image
- bright block
- herbaceous peony
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/13—Sensors therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Lighting Device Outwards From Vehicle And Optical Signal (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of light for vehicle recognition methods, device and vehicle, the described method includes：According to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is identified；According to identification as a result, determining historical position region of the taillight of the first object vehicle in the history Chinese herbaceous peony image；According to the historical position region, using preset model, the position of the taillight of the first object vehicle before the current vehicle collected in image is predicted, obtains object detection area；Before the current vehicle in the object detection area of image, the taillight of the first object vehicle is identified, to obtain the current information of the taillight of the first object vehicle.The disclosure obtains object detection area using the light for vehicle band of position of image before the historical position regional prediction current vehicle of light for vehicle, and light for vehicle identification is only carried out in the object detection area, having reached reduces operand, improves detection efficiency and improves the effect of real-time.
Description
Technical field
This disclosure relates to intelligent transport system field, and in particular, to a kind of light for vehicle recognition methods, device and car
.
Background technology
With the rapid development of science and technology, in intelligent transport system field, carried by being detected to front vehicles
The identification capability of high road pavement, is especially identified front vehicles in the night running of road traffic environment complexity, gives
The more security warnings of driver and auxiliary, so as to improve the active safety of vehicle, it has also become the important side of field development
To.
At present, tail-light is realized before identifying front vehicles generally by identification under driving at night environment.It is related
In technology, the identification to light for vehicle, usually using the corresponding extraction of feature formulation of light for vehicle and matched rule, and according to
Extraction and matched rule are identified taillight from the every frame Chinese herbaceous peony image collected.
Taillight recognition methods in correlation technique is, it is necessary to which all areas of each frame Chinese herbaceous peony image to collecting all carry out
Detection, to identify light for vehicle.Therefore, the recognition methods of correlation technique, detection calculations amount is big, and detection efficiency is low, and detection is real
When property is poor.
The content of the invention
The purpose of the disclosure is to provide a kind of light for vehicle recognition methods, device and vehicle, can solve in correlation technique
Need all areas of each frame Chinese herbaceous peony image to collecting all to be detected and make it that detection calculations amount is big, detection efficiency
Low, the problem of detection real-time is poor, reach the effect for improving detection efficiency and real-time.
To achieve these goals, the disclosure provides a kind of light for vehicle recognition methods, the described method includes：
According to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is identified；
According to identification as a result, determining history bit of the taillight of the first object vehicle in the history Chinese herbaceous peony image
Put region；
According to the historical position region, using preset model, the taillight of the first object vehicle is being collected
Position before current vehicle in image is predicted, and obtains object detection area；
Before the current vehicle in the object detection area of image, the taillight of the first object vehicle is identified,
To obtain the current information of the taillight of the first object vehicle.
Alternatively, the preset model is AR models；
It is described according to the historical position region, using preset model, the taillight of the first object vehicle is being gathered
To current vehicle before position in image be predicted, obtain object detection area the step of include：
According to the minimum enclosed rectangle frame boundary position information in the historical position region and the historical position region institute
Chinese herbaceous peony image acquisition time, obtain the first Random time sequence；
According to first Random time sequence, the taillight of the first object vehicle is being gathered using the AR models
To current vehicle before position in image be predicted.
Alternatively, the method further includes：
At interval of preset time, the region before the current vehicle collected beyond the object detection area of image, into
End of line lamp identifies；
If region recognition before the current vehicle collected beyond the object detection area of image is to the second target carriage
Taillight, then according to identification as a result, determining the band of position of the taillight in Chinese herbaceous peony image of second target vehicle；
By the minimum enclosed rectangle frame side of the band of position according to the taillight of second target vehicle in Chinese herbaceous peony image
Boundary position and the acquisition time of the Chinese herbaceous peony image, the second Random time sequence got add the first random time sequence
In row.
Alternatively, step described in the object detection area, that the taillight of the first object vehicle is identified
Suddenly further include：
If before the current vehicle of the default frame number of collection in image, do not recognize described the in the object detection area
The taillight of one target vehicle, then delete first Random time sequence.
Alternatively, step described in the object detection area, that the taillight of the first object vehicle is identified
Suddenly include：
According to preset rules, the object detection area is extended；
The taillight of the first object vehicle is identified in object detection area after expansion.
Alternatively, the history Chinese herbaceous peony image that the basis collects, the step that the taillight of first object vehicle is identified
Suddenly include：
The area that color characteristic meets the first preset condition is screened out in the default area-of-interest of the history Chinese herbaceous peony image
Domain, determines target area；
Adaptive threshold fuzziness is carried out to the image of the target area using maximum variance between clusters, obtains binary picture
Picture；
According to the fisrt feature information of bright block in the binary image, candidate's bright block is extracted；
The second feature information of the bright block pair obtained according to multiple candidate's bright block combination of two, extracts target bright block
To the taillight as the first object vehicle.
Alternatively, the fisrt feature information includes：Area, the height of minimum enclosed rectangle frame and minimum enclosed rectangle frame
Width；
The characteristic information according to bright block in the binary image, extracts candidate's bright block, including：
The binary image is connected and marked, obtains the bright block containing mark；
The area is extracted from the bright block less than preset area threshold value and the height of the minimum enclosed rectangle frame
Meet the bright block of the first preset range as candidate's bright block with the width of the minimum enclosed rectangle frame.
Alternatively, the second feature information includes：Area, central point, symmetry and the institute of the bright block centering bright block
State the minimum enclosed rectangle frame of bright block pair；
The second feature information of the bright block pair obtained according to multiple candidate's bright block combination of two, extracts target
The step of bright block is to taillight as the first object vehicle, including：
The area is extracted from the bright block centering and the central point meets the bright block of the second preset range to as time
Select bright block pair；
The minimum enclosed rectangle frame, which is extracted, from candidate's bright block centering meets the 3rd preset range and described symmetrical
The target bright block of the second preset condition of sexual satisfaction is to the taillight as the first object vehicle.
The disclosure also provides a kind of light for vehicle identification device, and described device includes：
First identification module, for according to the history Chinese herbaceous peony image collected, knowing to the taillight of first object vehicle
Not；
First determining module, for according to identification as a result, determine the first object vehicle taillight in the history
Historical position region in Chinese herbaceous peony image；
Prediction module, for according to the historical position region, using preset model, to the tail of the first object vehicle
Position of the lamp before the current vehicle collected in image is predicted, and obtains object detection area；
Second identification module, for before the current vehicle in the object detection area of image, to the first object car
Taillight be identified, to obtain the current information of the taillight of the first object vehicle.
Alternatively, the preset model is AR models；
The prediction module includes：
Acquisition submodule, for the minimum enclosed rectangle frame boundary position information according to the historical position region and described
The acquisition time of Chinese herbaceous peony image where historical position region, obtains the first Random time sequence；
Submodule is predicted, for according to first Random time sequence, using the AR models to the first object
Position of the taillight of vehicle before the current vehicle collected in image is predicted.
Alternatively, described device further includes：
3rd identification module, at interval of preset time, the target detection of image before the current vehicle collected
Region beyond region, carries out taillight identification；
Second determining module, if for the region beyond the object detection area of image before the current vehicle collected
The taillight of the second target vehicle is recognized, then according to identification as a result, determining the taillight of second target vehicle in Chinese herbaceous peony figure
The band of position as in；
Increase module, for by the minimum of the band of position according to the taillight of second target vehicle in Chinese herbaceous peony image
The acquisition time of boundary rectangle frame boundary position and the Chinese herbaceous peony image, the second Random time sequence for getting add described the
In one Random time sequence.
Alternatively, second identification module includes：
Submodule is deleted, if for before the current vehicle of the default frame number gathered in image, not in the target detection area
The taillight of the first object vehicle is recognized in domain, then is deleted first Random time sequence.
Alternatively, second identification module further includes：
Submodule is extended, for according to preset rules, the object detection area to be extended；
Submodule is identified, for knowing in object detection area after expansion to the taillight of the first object vehicle
Not.
Alternatively, first identification module includes：
Submodule is screened out, meets for screening out color characteristic in the default area-of-interest of the history Chinese herbaceous peony image
The region of one preset condition, determines target area；
Split submodule, for carrying out adaptive threshold point to the image of the target area using maximum variance between clusters
Cut, obtain binary image；
First extracting sub-module, for the fisrt feature information according to bright block in the binary image, extracts candidate
Bright block；
Second extracting sub-module, for the second feature of the bright block pair obtained according to multiple candidate's bright block combination of two
Information, extracts target bright block to the taillight as the first object vehicle.
Alternatively, the fisrt feature information includes：Area, the height of minimum enclosed rectangle frame and minimum enclosed rectangle frame
Width；
First extracting sub-module, is used for：
The binary image is connected and marked, obtains the bright block containing mark；
The area is extracted from the bright block less than preset area threshold value and the height of the minimum enclosed rectangle frame
Meet the bright block of the first preset range as candidate's bright block with the width of the minimum enclosed rectangle frame.
Alternatively, the second feature information includes：Area, central point, symmetry and the institute of the bright block centering bright block
State the minimum enclosed rectangle frame of bright block pair；
Second extracting sub-module, is used for：
The area is extracted from the bright block centering and the central point meets the bright block of the second preset range to as time
Select bright block pair；
The minimum enclosed rectangle frame, which is extracted, from candidate's bright block centering meets the 3rd preset range and described symmetrical
The target bright block of the second preset condition of sexual satisfaction is to the taillight as the first object vehicle.
The disclosure also provides a kind of vehicle, including above-mentioned light for vehicle identification device.
Through the above technical solutions, according to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is known
Not；According to identification as a result, determining historical position area of the taillight of the first object vehicle in the history Chinese herbaceous peony image
Domain；According to the historical position region, using preset model, to the taillight of the first object vehicle in the current vehicle collected
Position in preceding image is predicted, and obtains object detection area；It is right before the current vehicle in the object detection area of image
The taillight of the first object vehicle is identified, and to obtain the current information of the taillight of the first object vehicle, can solve
Certainly the taillight recognition methods in correlation technique needs all areas of each frame Chinese herbaceous peony image to collecting all to be detected
Identify the problem of detection calculations amount is big, detection efficiency is low and real-time is poor caused by light for vehicle, reach reduction computing
Amount, improve detection efficiency and improve the effect of real-time.
Other feature and advantage of the disclosure will be described in detail in subsequent specific embodiment part.
Brief description of the drawings
Attached drawing is for providing further understanding of the disclosure, and a part for constitution instruction, with following tool
Body embodiment is used to explain the disclosure together, but does not form the limitation to the disclosure.In the accompanying drawings：
Fig. 1 is the flow chart of the light for vehicle recognition methods of one embodiment of the disclosure；
Fig. 2 is the structure diagram of the light for vehicle identification device of one embodiment of the disclosure；
Fig. 3 is the flow chart of the light for vehicle recognition methods of the history Chinese herbaceous peony image of one embodiment of the disclosure；
Fig. 4 is the implement scene figure of the light for vehicle recognition methods of one embodiment of the disclosure；
Fig. 5 is the flow chart of candidate's bright block extracting method of one embodiment of the disclosure；
Fig. 6 is the flow chart of the light for vehicle extracting method of one embodiment of the disclosure；
Fig. 7 is the flow chart of the Chinese herbaceous peony image object detection zone acquisition methods of one embodiment of the disclosure；
Fig. 8 is the flow chart of light for vehicle recognition methods in the object detection area of one embodiment of the disclosure；
Fig. 9 is the flow chart of the object detection area acquisition methods of another embodiment of the disclosure；
Figure 10 is the flow chart of the light for vehicle recognition methods of another embodiment of the disclosure；
Figure 11 is the flow chart of the light for vehicle recognition methods of one embodiment of the disclosure；
Figure 12 is the flow chart of the light for vehicle recognition methods of one embodiment of the disclosure；
Figure 13 is the block diagram of the light for vehicle identification device of one embodiment of the disclosure；
Figure 14 is the block diagram of the device for light for vehicle recognition methods of one embodiment of the disclosure.
Embodiment
The embodiment of the disclosure is described in detail below in conjunction with attached drawing.It should be appreciated that this place is retouched
The embodiment stated is only used for describing and explaining the disclosure, is not limited to the disclosure.
It should be noted that term " first " in the specification and claims of the disclosure and above-mentioned attached drawing, "
Two ", " 3rd " etc. is for distinguishing similar object, without for describing specific order or precedence.
Fig. 1 is the flow chart of the light for vehicle recognition methods of one embodiment of the disclosure.With reference to Fig. 1, this method can include
Following steps：
In step s 11, according to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is identified.
With reference to Fig. 2, in an embodiment of the disclosure, can by installed in vehicle suitable position (for example, vehicle front windshield
After glass) image collecting device 110 (for example, forward sight camera) image before collecting vehicle in real time, processor 120 is (for example, DSP
Chip) the Chinese herbaceous peony image is received by the input interface being connected with image collecting device 110 and stores it in memory 130
In, subsequently to be analyzed and processed to the Chinese herbaceous peony image, as shown in Figure 2.
In embodiment of the disclosure, 110 real-time continuous of image collecting device carries out Chinese herbaceous peony Image Acquisition.History Chinese herbaceous peony figure
Seem the Chinese herbaceous peony image gathered in the past at current time, for example, using the 5 frame Chinese herbaceous peony images collected before current time as history
Chinese herbaceous peony image.Thus, in the 5 frame Chinese herbaceous peony image, the taillight of first object vehicle is identified respectively.
Under night running environment, since automobile tail light feature is stablized relatively, such as the brightness of light for vehicle, color, shape
The symmetry of shape and left and right vehicle wheel taillight, thus in the history Chinese herbaceous peony image collected after, using automobile tail light feature
The taillight of vehicle is identified.
In step s 12, according to identification as a result, determining taillight the going through in history Chinese herbaceous peony image of first object vehicle
The history band of position.
History Chinese herbaceous peony image for recognizing light for vehicle, can respectively obtain vehicle tail in these history Chinese herbaceous peony images
The historical position region of lamp.In embodiment of the disclosure, the historical position region of light for vehicle can be by the minimum of light for vehicle
2 points of upper left the angle point TL and lower right corner BR of boundary rectangle frame is determined.
In step s 13, according to historical position region, using preset model, the taillight of first object vehicle is being gathered
To current vehicle before position in image be predicted, obtain object detection area.
In step S14, before current vehicle in the object detection area of image, the taillight of first object vehicle is known
Not, to obtain the current information of the taillight of first object vehicle.
, can be according to preset model, according to these historical position areas after the historical position region of light for vehicle is determined
Before domain prediction current vehicle in image the light for vehicle the band of position, these bands of position can be as the target of image before current vehicle
Detection zone, is then only identified the taillight of vehicle in the object detection area, obtains the taillight in image before current vehicle
Information.
The light for vehicle recognition methods of the embodiment of the present disclosure, according to relevance of the taillight of target vehicle in successive frame,
Using the taillight position in history Chinese herbaceous peony image, the band of position occurred to taillight in next frame is predicted, and realization utilizes frame
Between the relevance of taillight realize the quick detection of light for vehicle.Thus, can be predicted using the historical position region of light for vehicle
The light for vehicle band of position of image obtains object detection area before current vehicle, and vehicle is only carried out in the object detection area
Taillight identifies that having reached reduces operand, improves detection efficiency and improve the effect of real-time.
With reference to Fig. 3, in an embodiment of the disclosure, step S11 can include：
In step S31, color characteristic is screened out in the default area-of-interest of history Chinese herbaceous peony image and meets that first is default
The region of condition, determines target area.
Since the calculation amount being detected to entire image is larger, time-consuming so that and the real-time of detection is poor, and according to right
Knowable to substantial amounts of Chinese herbaceous peony image is counted, taillight appears in position stabilization in the Chinese herbaceous peony image in certain area.Therefore,
An area-of-interest can be preset to Chinese herbaceous peony image to be detected, only preceding tail-light is examined in the area-of-interest
Survey, so as to reduce the workload of vehicle detection, shorten detection time, improve the real-time of detection.
In an embodiment of the disclosure, using the coboundary of Chinese herbaceous peony image 400 to be detected as x-axis, vertical direction is y-axis
Rectangular coordinate system is established, as shown in figure 4, boundary line in detection zone can will be set at the y-axis direction 1/3 of the Chinese herbaceous peony image 400,
It will be set to boundary line under detection zone at the maximum in y-axis direction, the region in detection zone between boundary line and lower boundary line is as pre-
If area-of-interest 410.In this manner it is possible to improve the real-time of detection and can effectively eliminate the interference in the Chinese herbaceous peony image
Light, such as street lamp and billboard are reflective.
It should be noted that in embodiment of the disclosure, boundary line and lower boundary line can be according to collections in detection zone
Image is suitably adjusted, and changes the size of the first area-of-interest according to actual conditions.
In the road image of the Chinese herbaceous peony gathered under night traffic environment, it will usually there are many highlight regions, such as before
Tail-light, street lamp, billboard are reflective etc..According to priori, in most cases, the taillight and other light sources of vehicle
Have significant difference in color, for example the taillight of vehicle shows as the R in partially red region, i.e. image in Chinese herbaceous peony image,
In tri- components of G, B, R component value is higher.Therefore, in embodiment of the disclosure, the first preset condition can be that R component value is more than
Default R component threshold value.According to the first preset condition, in Chinese herbaceous peony image, the region for meeting the first preset condition is screened out, then car
Remaining region is target area in the default area-of-interest of preceding image.
In addition, in embodiment of the disclosure, in order to improve picture quality, it is emerging that processor presets sense in Chinese herbaceous peony image
Before carrying out vehicle detection in interesting region, the Chinese herbaceous peony image can also be pre-processed.Pretreatment includes：Utilize Gaussian filter
Denoising and smoothing processing are carried out to the Chinese herbaceous peony image.
In step s 32, adaptive threshold fuzziness is carried out to the image of target area using maximum variance between clusters, obtained
Binary image.
After the target area of color characteristic the first preset condition of satisfaction is extracted, the image of target area is divided
Cut, obtain binary image.Maximum variance between clusters can be used to carry out adaptivenon-uniform sampling to Chinese herbaceous peony image, i.e., pass through statistics first
Substantial amounts of taillight brightness information obtains the initial segmentation threshold value Ts of image, and calculates initial segmentation threshold value T1 according to algorithm；So
The segmentation threshold T for making inter-class variance maximum is obtained using the method for traversal in the brightness cumulative histogram of multiple image afterwards；Most
The Chinese herbaceous peony image is split according to segmentation threshold T afterwards, the pixel more than segmentation threshold T is doubtful light for vehicle, also
It is target pixel points, labeled as white, i.e. pixel value is 255；Otherwise it is background, labeled as black, i.e. pixel value is 0, then
Obtain the binary image for only including 0 and 255 pixel values.
In step S33, according to the fisrt feature information of bright block in binary image, candidate's bright block is extracted.
In embodiment of the disclosure, fisrt feature information can include area, minimum enclosed rectangle frame height and
The width of minimum enclosed rectangle frame., can basis since the shape of light for vehicle has certain feature and meets certain proportion
Fisrt feature information, screens out the interference bright block not being consistent with the shape facility of light for vehicle, and remaining bright block is then vehicle tail
The bright block of lamp, as candidate's bright block.
With reference to Fig. 5, in an embodiment of the disclosure, above-mentioned steps S33 includes：
In step S331, binary image is connected and is marked, obtain the bright block containing mark.
What the binary image after adaptive threshold fuzziness obtained is some scattered pixels, these pixel phases
Between mutually and onrelevant, it is therefore desirable to connect the pixel for belonging to same bright block using the method for connection to be formed and right
Connect obtained bright block to be marked, to distinguish the pixel for belonging to different objects.
In an embodiment of the disclosure, the mode of communicating in 8 fields can be used and connected region is marked, also
It is that can be scanned according to default scan mode (such as from top to bottom, from left to right), when searching first aim picture
During vegetarian refreshments (pixel value is 255 pixel), assigns the target pixel points and be labeled as " 1 ", and using this point as starting point, utilize bright block
Contour tracing method, determine this bright block, and the same mark " 1 " of all pixels point distribution to same bright block, i.e., according to default
Direction (such as counterclockwise) the next target pixel points of peripherad eight fields search；If next pixel is run into be somebody's turn to do
Pixel value is 255 pixel in neighborhood, then this pixel is labeled as " 1 ", otherwise continues search for next side of its neighborhood
To pixel, finished until by the point search of the neighborhood.After a connected region has been marked, search after the same method
New connected region, mark number carries out plus " 1 " calculates, the mark of the mark namely bright block of pixel in bright block.In this way, just
The number of the bright block of doubtful preceding tail-light and the mark of each bright block are obtained.Due to features such as the size of these bright blocks, shapes
It is different, it is therefore desirable to which that the binary image is further processed.
In step S332, area is extracted from bright block less than preset area threshold value and the height of minimum enclosed rectangle frame
Meet the bright block of the first preset range as candidate's bright block with height.
Since car light, road may be included in the bright block after being marked in night traffic scene there are various interference illumination, connection
Lamp, pavement reflecting and the billboard interference bright block such as reflective, it is therefore desirable to according to the area of bright block, shape, minimum enclosed rectangle frame
Height and width characteristics, extract candidate's bright block of the taillight feature for meeting vehicle.
First, the area of bright block need to meet to be greater than or equal to preset area threshold value, to screen out the interference bright block of small area, i.e.,
A(Ci)≥THA, in formula, A (Ci) represent bright block CiArea, THAFor preset area threshold value., can in an embodiment of the disclosure
To set preset area threshold value as 3, that is, screen the bright block that area is less than 3 pixels.
Secondly, the height and width of the minimum enclosed rectangle frame of bright block are also because meeting the first preset range.In the disclosure
In one embodiment, shown in the first preset range such as formula (1).
In formula, W (Ci) and respectively bright CiMinimum enclosed rectangle frame width and height, THWFor the minimum enclosed rectangle
The predetermined width threshold value of frame, THWHR1And THWHR2Respectively default the ratio of width to height lower and upper limit of the minimum enclosed rectangle frame.
It should be noted that in embodiment of the disclosure, the default face of the limitation bright block area in taillight extracting rule
Product threshold value, the height of minimum enclosed rectangle frame for limiting bright block and the predetermined width threshold value of width and default the ratio of width to height lower limit and
The upper limit is rule of thumb set, and can be adjusted according to the Chinese herbaceous peony image collected.
Area is less than the height of preset area threshold value and minimum enclosed rectangle frame and height meets the bright of the first preset range
Block extracts, as target bright block, in order to carry out the taillight that subsequent treatment identifies vehicle.
In step S34, the second feature information of the bright block pair obtained according to multiple candidate's bright block combination of two, carries
Target bright block is taken out to the taillight as first object vehicle.
In night running, the left and right taillight of vehicle generally all has the features such as similar area, height, symmetry,
Along with left and right taillight occurs in pairs, therefore using these features of light for vehicle, to candidate's bright block for extracting group two-by-two
Obtained bright block is closed to being further analyzed processing, the taillight of vehicle is obtained, realizes the identification to light for vehicle.
With reference to Fig. 6, in an embodiment of the disclosure, above-mentioned steps S34 includes：
In step S341, meet the bright block pair of the second preset range from the area and central point of bright block centering extraction bright block
As candidate's bright block pair.
Since automobile tail light generally occurs in pairs, and two taillights of same vehicle in shape and position have it is similar
Property, the difference of the difference and bright light central point that are embodied between two taillight areas in vertical direction should be solid at one
In fixed scope.And the bright block that multiple candidate's bright block combination of two are formed not necessarily belongs to same vehicle, it is therefore desirable to according to bright
The area and central point of block centering bright block, extract the bright block for meeting the second preset range to candidate's bright block pair, to exclude not belong to
In the bright block pair of same vehicle.
In an embodiment of the disclosure, shown in the second preset range such as formula (2).
In formula, A () is the area of candidate's bright block；CY() is the central point ordinate of candidate's bright block；THdA,Respectively
For the difference in areas threshold value and taillight central point Diff N threshold value of bright block centering bright block.
In step S342, extract minimum enclosed rectangle frame from candidate's bright block centering and meet the 3rd preset range and bright block
The second preset condition of symmetrical sexual satisfaction bright block to the taillight as first object vehicle.
According to the transformational relation of world coordinates to image coordinate, the taillight of vehicle has necessarily in Chinese herbaceous peony image
Feature, the i.e. width of the minimum enclosed rectangle frame for the bright block that two taillights are formed and the ratio of width to height all meet certain condition.In this public affairs
In the embodiment opened, shown in the 3rd preset range such as formula (3).
In formula, u (), d (), l () and r () are respectively bright block centering bright block CiWith bright block CjMinimum enclosed rectangle frame
Up, down, left and right side position；THw1And THw2The respectively width threshold value lower and upper limit of the minimum enclosed rectangle frame of bright block pair；
THRatio1And THRatio2The respectively lower and upper limit of the ratio of width to height of the minimum enclosed rectangle frame of bright block pair.
In addition, under normal circumstances, two taillights of same vehicle have higher symmetry, can pass through cross-correlation function degree
The symmetry of two candidate's bright blocks is measured, i.e., the cross correlation value of two candidate's bright blocks is obtained according to cross-correlation function, if cross correlation value meets
Second preset condition, then it is believed that candidate's bright block is to the taillight for vehicle.In an embodiment of the disclosure, the second default bar
Part can be that the cross correlation value of two bright blocks in candidate's bright block is more than predetermined threshold value.
With reference to Fig. 7, in an embodiment of the disclosure, step S13 includes：
In step S71, according to the minimum enclosed rectangle frame boundary position information in historical position region and historical position area
The acquisition time of Chinese herbaceous peony image where domain, obtains the first Random time sequence.
The historical position region of the taillight of first object vehicle can be by boundary position (the i.e. upper left of its minimum enclosed rectangle frame
Angle TL and 2 points of lower right corner BR) determine, and the Chinese herbaceous peony image where per frame historical position region all corresponds to an acquisition time, because
TL and two point coordinates of BR can be regarded as two time serieses of X and Y-direction respectively by this, that is, obtain first time sequence L=[TL
(x), BR (x), TL (y), BR (y)], then the position of target taillight pair is determined by four coordinates.
In step S72, according to the first Random time sequence, the taillight of first object vehicle is being gathered using AR models
To current vehicle image in position be predicted.
For moment t, the first Random time sequence is denoted as Lt=[TL (xt),BR(xt),TL(yt),BR(yt)], for when
Between sequence t1, t2..., tk, the first Random time sequence is denoted as [Lt1,Lt2,...,Ltk].Due to determining the tail of first object vehicle
Four coordinate components movement tendencies in the historical position region of lamp are not necessarily identical, by the first Random time sequence [Lt1,
Lt2,...,Ltk] four one-dimensional time serieses are resolved into, using four coordinates of the AR models difference prediction history band of position
The coordinate at component collection moment of image before current vehicle, you can obtain the taillight of first object vehicle in the current vehicle collected
Position in preceding image.
Since the exponent number acquirement of AR models is excessive, the real-time of the increase of algorithm calculation amount, vehicle detection will be caused to be deteriorated, because
The AR models of this optional 3 rank predict the position of the taillight of first object vehicle.
In an embodiment of the disclosure, the taillight of first object vehicle can be set in the 5 frame history Chinese herbaceous peonies collected
Position in image utilizes AR model predictions first as the first initial Random time sequence according to the first Random time sequence
The band of position of the taillight of target vehicle before the current vehicle collected in image, the band of position can be used as image before current vehicle
Object detection area.If the taillight of the first object vehicle is recognized in the object detection area of image before current vehicle,
According to the minimum enclosed rectangle frame boundary bit confidence of the band of position of the taillight of the first object vehicle before current vehicle in image
The acquisition time of image is updated the first initial Random time sequence before breath and current vehicle, for first object car
Position of the taillight in next frame Chinese herbaceous peony image be predicted.
With reference to Fig. 8, in an embodiment of the disclosure, step S14 includes：
In step S73, if not identified before the current vehicle of the default frame number of collection in image in object detection area
To the taillight of first object vehicle, then the first Random time sequence is deleted.
Since in night running, some disturbing factors may cause the first object car that previous frame detects on Chinese herbaceous peony road
Taillight before current vehicle transient loss in image column, therefore, in the disclosure, if in the Chinese herbaceous peony image of the default frame number of setting
All identify the taillight less than first object vehicle, then it is assumed that the taillight of the first object vehicle has been moved out Chinese herbaceous peony image, at this time may be used
Corresponding first Random time sequence of the taillight of the first object vehicle will be deleted, no longer to the taillight of first object vehicle into
Line trace identifies.
With reference to Fig. 9, in another embodiment of the present disclosure, step S14 is further included：
In step S91, according to preset rules, object detection area is extended.
In step S92, the taillight of first object vehicle is identified in object detection area after expansion.
In an embodiment of the disclosure, preset rules can be that obtained target prediction region will be predicted to external expansion 5
Pixel, and the taillight of first object detection zone is identified in object detection area after expansion, thus can be to prediction
Slight error compensate, to improve the accurate fixed and reliability of detection, reduce flase drop and omission factor.
Figure 10 is a kind of flow chart of the light for vehicle recognition methods shown according to another exemplary embodiment.With reference to figure
10, this method can include：
In step S101, according to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is identified.
In step s 102, according to identification as a result, determining the taillight of first object vehicle in history Chinese herbaceous peony image
Historical position region.
In step s 103, according to historical position region, using preset model, the taillight of first object vehicle is being gathered
To current vehicle before position in image be predicted, obtain object detection area.
In step S104, before current vehicle in the object detection area of image, the taillight of first object vehicle is carried out
Identification, to obtain the current information of the taillight of first object vehicle.
In step S105, at interval of preset time, before the current vehicle collected beyond the object detection area of image
Region, carry out taillight identification.
Due in vehicle travel process, vehicle front may enter other vehicles, it is therefore desirable to when default
Between to beyond the object detection area of Chinese herbaceous peony image region carry out taillight identification, determine whether other vehicles enter.At this
In a disclosed embodiment, preset time can be the time for often gathering 20 frame Chinese herbaceous peony images, that is, at interval of 20 frame Chinese herbaceous peony figures
Picture, just carries out taillight identification to the region beyond the object detection area of the Chinese herbaceous peony image currently collected.
In step s 106, if region recognition before the current vehicle collected beyond the object detection area of image is to second
The taillight of target vehicle, then according to identification as a result, determining the band of position of the taillight of the second target vehicle in Chinese herbaceous peony image.
In step s 107, the band of position by the taillight according to the second target vehicle in Chinese herbaceous peony image is minimum external
When the second Random time sequence addition first that the acquisition time of rectangle frame boundary position and the Chinese herbaceous peony image is got is random
Between in sequence.
If the region recognition before current vehicle beyond the object detection area of image is obtained to the taillight of the second target vehicle
Take the minimum external square of the band of position of the taillight of the second target vehicle in Chinese herbaceous peony image, the i.e. taillight of second target vehicle
The coordinate of 2 points of the upper left corner TL and lower right corner BR of shape frame and the time for collecting image before current vehicle, obtain second it is random when
Between sequence, and by the second Random time sequence add the first Random time sequence in, thus, it can be achieved that prediction the second target vehicle
The band of position of the taillight in next frame Chinese herbaceous peony image, carry out Tracking Recognition that can be to the taillight of the second target vehicle.
Referring to Figure 11, in an embodiment of the disclosure, global detection is carried out to the Chinese herbaceous peony image collected first and (is existed
The default area-of-interest of the Chinese herbaceous peony image of collection carries out taillight identification), when occurring target vehicle (i.e. first in Chinese herbaceous peony image
Target vehicle) taillight when, using the Chinese herbaceous peony image comprising the first object vehicle collected as history Chinese herbaceous peony image, at this
In a little history Chinese herbaceous peony images, the taillight of first object vehicle is identified respectively, the taillight of acquisition first object vehicle is gone through
The history band of position.And the first Random time sequence is obtained according to historical position region, according to first Random time sequence, utilize
Trip region of the AR models to the taillight of first object vehicle is predicted to obtain object detection area, with the target detection area
Domain is detected identification to the taillight of first object vehicle.When occurring fresh target vehicle in the Chinese herbaceous peony image collected (for example,
Two target vehicles) when, the taillight of the target vehicle is identified, and the second random time sequence is obtained according to its taillight region
Row, the second random event sequence of acquisition is added in the first Random time sequence, to realize to first object vehicle and second
The taillight of target vehicle is carried out at the same time Tracking Recognition.If in the Chinese herbaceous peony image collected, and it is not detected by the taillight of any vehicle
(i.e. all target vehicles disappear), then come back to global detection.
Referring to Figure 12, in the embodiment of the present disclosure, to the Chinese herbaceous peony image collected, using AR models to each target carriage in front
The position of taillight carry out real-time estimate, in the obtained object detection area of prediction, carry out the quick of target vehicle taillight
Identification, so that the tail-lamp information quickly recognized can be realized.In certain embodiments, can be into according to the tail-lamp information identified
One step is handled tail-lamp information to obtain the situation of target vehicle, for example, the tail of the distance, target vehicle with target vehicle
The information such as lamp situation (brake lamp, steering indicating light etc.), are reminded, output safety prompt message so as to provide driver driving strategy
Deng.
By the above-mentioned technical proposal of the disclosure, the taillight of first object vehicle is known in history Chinese herbaceous peony image
After not, according to historical position region of the taillight of first object vehicle in history Chinese herbaceous peony image using preset model to this first
The position of the taillight of target vehicle carries out real-time estimate, using the band of position predicted as object detection area to first object
The taillight of vehicle is identified, can solve correlation technique need to all areas of each frame Chinese herbaceous peony image that collect all into
Row is detected to identify the problem of detection calculations amount is big, detection efficiency is low and real-time is poor caused by light for vehicle, is reached
Reduce operand, improve the effect of detection efficiency and real-time.
Secondly, carried out at interval of region of the preset time before the current vehicle collected beyond the object detection area of image
Taillight identifies, and the band of position of second target vehicle in Chinese herbaceous peony road image is added to the random time sequence of preset model
In row, to predict position of second target vehicle in follow-up Chinese herbaceous peony image, it is possible to reduce the detection time of light for vehicle with
And reduce omission factor.
In addition, during the taillight of vehicle is identified, image is carried out using maximum variance between clusters adaptive
Threshold segmentation is answered to obtain binary image, the adaptive ability of Threshold segmentation can be improved, so as to improve the reality of vehicle detection
When property and validity.
Figure 13 is the block diagram of the light for vehicle identification device of one embodiment of the disclosure.With reference to Figure 13, device 100 includes：The
One identification module 131, the first determining module 132,133 and second identification module 134 of prediction module；
First identification module 131, for according to the history Chinese herbaceous peony image collected, being carried out to the taillight of first object vehicle
Identification；
First determining module 132, for according to identification as a result, determine the first object vehicle taillight gone through described
Historical position region in history Chinese herbaceous peony image；
Prediction module 133, for according to the historical position region, using preset model, to the first object vehicle
Position of the taillight before the current vehicle collected in image be predicted, obtain object detection area；
Second identification module 134, for before the current vehicle in the object detection area of image, to the first object
The taillight of vehicle is identified, to obtain the current information of the taillight of the first object vehicle.
In an embodiment of the disclosure, prediction module 133 includes：
Acquisition submodule 1331, for the minimum enclosed rectangle frame boundary position information according to the historical position region and
The acquisition time of Chinese herbaceous peony image where the historical position region, obtains the first Random time sequence；
Submodule 1332 is predicted, for according to first Random time sequence, using the AR models to described first
Position of the taillight of target vehicle before the current vehicle collected in image is predicted.
In an embodiment of the disclosure, device 100 further includes：
3rd identification module 135, at interval of preset time, the target of image to be examined before the current vehicle collected
The region beyond region is surveyed, carries out taillight identification；
Second determining module 136, if beyond for the object detection area of image before the current vehicle collected
Region recognition to the second target vehicle taillight, then according to identification as a result, determine second target vehicle taillight in car
The band of position in preceding image；
Increase module 137, for by the band of position according to the taillight of second target vehicle in Chinese herbaceous peony image
The acquisition time of minimum enclosed rectangle frame boundary position and the Chinese herbaceous peony image, the second Random time sequence got add institute
State in the first Random time sequence.
In an embodiment of the disclosure, the second identification module 134 includes：
Submodule 1341 is deleted, if in image before the current vehicle of the default frame number gathered, not examined in the target
Survey in region and recognize the taillight of the first object vehicle, then delete first Random time sequence.
In another embodiment of the present disclosure, the second identification module 134 further includes：
Submodule 1342 is extended, for according to preset rules, the object detection area to be extended；
Identify submodule 1343, in object detection area after expansion to the taillight of the first object vehicle into
Row identification.
In an embodiment of the disclosure, the first identification module 131 includes：
Submodule 1311 is screened out, is expired for screening out color characteristic in the default area-of-interest of the history Chinese herbaceous peony image
The region of the first preset condition of foot, determines target area；
Split submodule 1312, for carrying out adaptive thresholding to the image of the target area using maximum variance between clusters
Value segmentation, obtains binary image；
First extracting sub-module 1313, for the fisrt feature information according to bright block in the binary image, extracts
Candidate's bright block；
Second extracting sub-module 1314, for obtained according to multiple candidate's bright block combination of two the second of bright block pair
Characteristic information, extracts target bright block to the taillight as the first object vehicle.
In an embodiment of the disclosure, the fisrt feature information includes：Area, minimum enclosed rectangle frame height and
The width of minimum enclosed rectangle frame；
First extracting sub-module 1313, is used for：
The binary image is connected and marked, obtains the bright block containing mark；
The area is extracted from the bright block less than preset area threshold value and the height of the minimum enclosed rectangle frame
Meet the bright block of the first preset range as candidate's bright block with the width of the minimum enclosed rectangle frame.
In an embodiment of the disclosure, the second feature information includes：The area of bright block, center in the bright block
The minimum enclosed rectangle frame of point, symmetry and the bright block pair；
Second extracting sub-module 1314, for extracting the area and central point satisfaction second from the bright block centering
The bright block of preset range is to as candidate's bright block pair；
The minimum enclosed rectangle frame, which is extracted, from candidate's bright block centering meets the 3rd preset range and described symmetrical
The target bright block of the second preset condition of sexual satisfaction is to the taillight as the first object vehicle.
On the device in above-described embodiment, wherein modules perform the concrete mode of operation in related this method
Embodiment in be described in detail, explanation will be not set forth in detail herein.
Figure 14 is a kind of block diagram for device 1400 for light for vehicle recognition methods that the disclosure provides.For example, device
1400 may be provided as vehicle.With reference to Figure 14, device 1400 includes：Electronic control unit 1401, image collecting device 1402,
Processor 1403, memory 1404, automobile brake system 1405, steering wheel angle sensor 1406, wheel speed sensors 1407, hair
Motivation system 1408 and CAN bus 1409 etc..
Wherein, image collecting device 1402 can be used for the real-time collection for performing above-mentioned Chinese herbaceous peony image.Processor 1403 can be used
In the above-mentioned light for vehicle recognition methods of execution.Memory 1404 can be used for the Chinese herbaceous peony figure that storage image collecting device 1402 collects
Picture, easy to subsequent analysis processing.
The preferred embodiment of the disclosure is described in detail above in association with attached drawing, still, the disclosure is not limited to above-mentioned reality
The detail in mode is applied, in the range of the technology design of the disclosure, a variety of letters can be carried out to the technical solution of the disclosure
Monotropic type, these simple variants belong to the protection domain of the disclosure.
It is further to note that each particular technique feature described in above-mentioned embodiment, in not lance
In the case of shield, it can be combined by any suitable means.In order to avoid unnecessary repetition, the disclosure to it is various can
The combination of energy no longer separately illustrates.
In addition, it can also be combined between a variety of embodiments of the disclosure, as long as it is without prejudice to originally
Disclosed thought, it should equally be considered as disclosure disclosure of that.
Claims (17)
- A kind of 1. light for vehicle recognition methods, it is characterised in that the described method includes：According to the history Chinese herbaceous peony image collected, the taillight of first object vehicle is identified；According to identification as a result, determining historical position area of the taillight of the first object vehicle in the history Chinese herbaceous peony image Domain；It is current what is collected to the taillight of the first object vehicle using preset model according to the historical position region Position in Chinese herbaceous peony image is predicted, and obtains object detection area；Before the current vehicle in the object detection area of image, the taillight of the first object vehicle is identified, with To the current information of the taillight of the first object vehicle.
- 2. according to the method described in claim 1, it is characterized in that, the preset model is AR models；It is described according to the historical position region, using preset model, the taillight of the first object vehicle is being collected The step of position before current vehicle in image is predicted, obtains object detection area includes：According to where the minimum enclosed rectangle frame boundary position information in the historical position region and the historical position region The acquisition time of Chinese herbaceous peony image, obtains the first Random time sequence；According to first Random time sequence, the taillight of the first object vehicle is being collected using the AR models Position before current vehicle in image is predicted.
- 3. according to the method described in claim 2, it is characterized in that, the method further includes：At interval of preset time, the region before the current vehicle collected beyond the object detection area of image, into end of line Lamp identifies；If region recognition before the current vehicle collected beyond the object detection area of image is to the second target vehicle Taillight, then according to identification as a result, determining the band of position of the taillight of second target vehicle in Chinese herbaceous peony image；By the minimum enclosed rectangle frame boundary bit of the band of position according to the taillight of second target vehicle in Chinese herbaceous peony image The second Random time sequence got with the acquisition time of the Chinese herbaceous peony image is put to add in first Random time sequence.
- 4. according to the method described in claim 2, it is characterized in that, described in the object detection area, to described first The step of taillight of target vehicle is identified further includes：If first mesh is not recognized in the object detection area in image before the current vehicle of the default frame number of collection The taillight of vehicle is marked, then is deleted first Random time sequence.
- 5. according to the method described in claim 1, it is characterized in that, described in the object detection area, to described first The step of taillight of target vehicle is identified includes：According to preset rules, the object detection area is extended；The taillight of the first object vehicle is identified in object detection area after expansion.
- 6. according to the method described in claim 1, it is characterized in that, the history Chinese herbaceous peony image that the basis collects, to first The step of taillight of target vehicle is identified includes：The region that color characteristic meets the first preset condition is screened out in the default area-of-interest of the history Chinese herbaceous peony image, really Set the goal region；Adaptive threshold fuzziness is carried out to the image of the target area using maximum variance between clusters, obtains binary image；According to the fisrt feature information of bright block in the binary image, candidate's bright block is extracted；The second feature information of the bright block pair obtained according to multiple candidate's bright block combination of two, extracts target bright block to making For the taillight of the first object vehicle.
- 7. according to the method described in claim 6, it is characterized in that, the fisrt feature information includes：Area, minimum external square The height of shape frame and the width of minimum enclosed rectangle frame；The characteristic information according to bright block in the binary image, extracts candidate's bright block, including：The binary image is connected and marked, obtains the bright block containing mark；The area is extracted from the bright block less than preset area threshold value and height and the institute of the minimum enclosed rectangle frame The width for stating minimum enclosed rectangle frame meets the bright block of the first preset range as candidate's bright block.
- 8. according to the method described in claim 6, it is characterized in that, the second feature information includes：The bright block centering is bright The area of block, central point, the minimum enclosed rectangle frame of symmetry and the bright block pair；The second feature information of the bright block pair obtained according to multiple candidate's bright block combination of two, extracts target bright block The step of to taillight as the first object vehicle, including：The area is extracted from the bright block centering and the central point meets the bright block of the second preset range to bright as candidate Block pair；Minimum enclosed rectangle frame the 3rd preset range of satisfaction is extracted from candidate's bright block centering and the symmetry expires The target bright block of the second preset condition of foot is to the taillight as the first object vehicle.
- 9. a kind of light for vehicle identification device, it is characterised in that described device includes：First identification module, for according to the history Chinese herbaceous peony image collected, the taillight of first object vehicle to be identified；First determining module, for according to identification as a result, determine the first object vehicle taillight in the history Chinese herbaceous peony Historical position region in image；Prediction module, for according to the historical position region, using preset model, existing to the taillight of the first object vehicle Position before the current vehicle collected in image is predicted, and obtains object detection area；Second identification module, for before the current vehicle in the object detection area of image, to the first object vehicle Taillight is identified, to obtain the current information of the taillight of the first object vehicle.
- 10. device according to claim 9, it is characterised in that the preset model is AR models；The prediction module includes：Acquisition submodule, for the minimum enclosed rectangle frame boundary position information according to the historical position region and the history The acquisition time of Chinese herbaceous peony image where the band of position, obtains the first Random time sequence；Submodule is predicted, for according to first Random time sequence, using the AR models to the first object vehicle Position of the taillight before the current vehicle collected in image be predicted.
- 11. device according to claim 10, it is characterised in that described device further includes：3rd identification module, at interval of preset time, the object detection area of image before the current vehicle collected Region in addition, carries out taillight identification；Second determining module, if for the region recognition beyond the object detection area of image before the current vehicle collected To the taillight of the second target vehicle, then according to identification as a result, determining the taillight of second target vehicle in Chinese herbaceous peony image The band of position；Increase module, for by the minimum external of the band of position according to the taillight of second target vehicle in Chinese herbaceous peony image The acquisition time of rectangle frame boundary position and the Chinese herbaceous peony image, the second Random time sequence for getting add described first with In machine time series.
- 12. device according to claim 10, it is characterised in that second identification module includes：Submodule is deleted, if for before the current vehicle of the default frame number gathered in image, not in the object detection area The taillight of the first object vehicle is recognized, then is deleted first Random time sequence.
- 13. device according to claim 9, it is characterised in that second identification module further includes：Submodule is extended, for according to preset rules, the object detection area to be extended；Submodule is identified, for the taillight of the first object vehicle to be identified in object detection area after expansion.
- 14. device according to claim 9, it is characterised in that first identification module includes：Submodule is screened out, meets that first is pre- for screening out color characteristic in the default area-of-interest of the history Chinese herbaceous peony image If the region of condition, determines target area；Split submodule, for carrying out adaptive threshold fuzziness to the image of the target area using maximum variance between clusters, Obtain binary image；First extracting sub-module, for the fisrt feature information according to bright block in the binary image, extracts candidate's bright block；Second extracting sub-module, for the second feature letter of the bright block pair obtained according to multiple candidate's bright block combination of two Breath, extracts target bright block to the taillight as the first object vehicle.
- 15. device according to claim 14, it is characterised in that the fisrt feature information includes：Area, minimum are external The height of rectangle frame and the width of minimum enclosed rectangle frame；First extracting sub-module, is used for：The binary image is connected and marked, obtains the bright block containing mark；The area is extracted from the bright block less than preset area threshold value and height and the institute of the minimum enclosed rectangle frame The width for stating minimum enclosed rectangle frame meets the bright block of the first preset range as candidate's bright block.
- 16. device according to claim 14, it is characterised in that the second feature information includes：The bright block centering The area of bright block, central point, the minimum enclosed rectangle frame of symmetry and the bright block pair；Second extracting sub-module, is used for：The area is extracted from the bright block centering and the central point meets the bright block of the second preset range to bright as candidate Block pair；Minimum enclosed rectangle frame the 3rd preset range of satisfaction is extracted from candidate's bright block centering and the symmetry expires The target bright block of the second preset condition of foot is to the taillight as the first object vehicle.
- 17. a kind of vehicle, it is characterised in that including the claims 9-16 any one of them light for vehicle identification devices.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610978015.4A CN108021856B (en) | 2016-10-31 | 2016-10-31 | Vehicle tail lamp identification method and device and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610978015.4A CN108021856B (en) | 2016-10-31 | 2016-10-31 | Vehicle tail lamp identification method and device and vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108021856A true CN108021856A (en) | 2018-05-11 |
| CN108021856B CN108021856B (en) | 2020-09-15 |
Family
ID=62083703
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610978015.4A Active CN108021856B (en) | 2016-10-31 | 2016-10-31 | Vehicle tail lamp identification method and device and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108021856B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764102A (en) * | 2018-05-22 | 2018-11-06 | 哈尔滨理工大学 | A kind of characteristic recognition method of the front vehicles image based on color extraction algorithm |
| CN109785385A (en) * | 2019-01-22 | 2019-05-21 | 中国科学院自动化研究所 | Visual target tracking method and system |
| CN110188645A (en) * | 2019-05-22 | 2019-08-30 | 北京百度网讯科技有限公司 | For the method for detecting human face of vehicle-mounted scene, device, vehicle and storage medium |
| CN110276742A (en) * | 2019-05-07 | 2019-09-24 | 平安科技（深圳）有限公司 | Tail light for train monitoring method, device, terminal and storage medium |
| CN110673609A (en) * | 2019-10-10 | 2020-01-10 | 北京小马慧行科技有限公司 | Vehicle running control method, device and system |
| CN110696807A (en) * | 2019-09-29 | 2020-01-17 | 一汽解放青岛汽车有限公司 | Engine shutdown control method and system under traffic jam condition, vehicle and storage medium |
| CN110758394A (en) * | 2019-09-29 | 2020-02-07 | 一汽解放青岛汽车有限公司 | Engine starting control method and system under traffic jam working condition, vehicle and storage medium |
| CN111062231A (en) * | 2018-10-17 | 2020-04-24 | 财团法人车辆研究测试中心 | Vehicle detection method, night vehicle detection method based on light intensity dynamic and system thereof |
| CN111191482A (en) * | 2018-11-14 | 2020-05-22 | 杭州海康威视数字技术股份有限公司 | Brake lamp identification method and device and electronic equipment |
| CN111652143A (en) * | 2020-06-03 | 2020-09-11 | 浙江大华技术股份有限公司 | Vehicle detection method and device and computer storage medium |
| CN112164090A (en) * | 2020-09-04 | 2021-01-01 | 杭州海康威视系统技术有限公司 | Data processing method and device, electronic equipment and machine-readable storage medium |
| CN116994514A (en) * | 2023-09-26 | 2023-11-03 | 永林电子股份有限公司 | Image recognition-based vehicle display brightness control method, device and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101727748A (en) * | 2009-11-30 | 2010-06-09 | 北京中星微电子有限公司 | Method, system and equipment for monitoring vehicles based on vehicle taillight detection |
| CN102867416A (en) * | 2012-09-13 | 2013-01-09 | 中国科学院自动化研究所 | Vehicle part feature-based vehicle detection and tracking method |
| CN103150898A (en) * | 2013-01-25 | 2013-06-12 | 大唐移动通信设备有限公司 | Method and device for detection of vehicle at night and method and device for tracking of vehicle at night |
| CN103778786A (en) * | 2013-12-17 | 2014-05-07 | 东莞中国科学院云计算产业技术创新与育成中心 | Traffic violation detection method based on significant vehicle part model |
| CN103824308A (en) * | 2014-02-11 | 2014-05-28 | 广州中国科学院先进技术研究所 | Image processing method in wireless multimedia sensor network |
| CN103984917A (en) * | 2014-04-10 | 2014-08-13 | 杭州电子科技大学 | Multi-feature nighttime vehicle detection method based on machine vision |
| CN103984950A (en) * | 2014-04-22 | 2014-08-13 | 北京联合大学 | Moving vehicle stop lamp state recognition method adaptable to day detection |
| CN104392629A (en) * | 2014-11-07 | 2015-03-04 | 深圳市中天安驰有限责任公司 | Method and device for detecting car distance |
| CN104866838A (en) * | 2015-06-02 | 2015-08-26 | 南京航空航天大学 | Vision-based automatic detection method for front vehicle |
| CN105718893A (en) * | 2016-01-22 | 2016-06-29 | 江苏大学 | Car tail light pair detecting method for night environment |
-
2016
- 2016-10-31 CN CN201610978015.4A patent/CN108021856B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101727748A (en) * | 2009-11-30 | 2010-06-09 | 北京中星微电子有限公司 | Method, system and equipment for monitoring vehicles based on vehicle taillight detection |
| CN102867416A (en) * | 2012-09-13 | 2013-01-09 | 中国科学院自动化研究所 | Vehicle part feature-based vehicle detection and tracking method |
| CN103150898A (en) * | 2013-01-25 | 2013-06-12 | 大唐移动通信设备有限公司 | Method and device for detection of vehicle at night and method and device for tracking of vehicle at night |
| CN103778786A (en) * | 2013-12-17 | 2014-05-07 | 东莞中国科学院云计算产业技术创新与育成中心 | Traffic violation detection method based on significant vehicle part model |
| CN103824308A (en) * | 2014-02-11 | 2014-05-28 | 广州中国科学院先进技术研究所 | Image processing method in wireless multimedia sensor network |
| CN103984917A (en) * | 2014-04-10 | 2014-08-13 | 杭州电子科技大学 | Multi-feature nighttime vehicle detection method based on machine vision |
| CN103984950A (en) * | 2014-04-22 | 2014-08-13 | 北京联合大学 | Moving vehicle stop lamp state recognition method adaptable to day detection |
| CN104392629A (en) * | 2014-11-07 | 2015-03-04 | 深圳市中天安驰有限责任公司 | Method and device for detecting car distance |
| CN104866838A (en) * | 2015-06-02 | 2015-08-26 | 南京航空航天大学 | Vision-based automatic detection method for front vehicle |
| CN105718893A (en) * | 2016-01-22 | 2016-06-29 | 江苏大学 | Car tail light pair detecting method for night environment |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108764102A (en) * | 2018-05-22 | 2018-11-06 | 哈尔滨理工大学 | A kind of characteristic recognition method of the front vehicles image based on color extraction algorithm |
| CN111062231A (en) * | 2018-10-17 | 2020-04-24 | 财团法人车辆研究测试中心 | Vehicle detection method, night vehicle detection method based on light intensity dynamic and system thereof |
| CN111062231B (en) * | 2018-10-17 | 2023-03-24 | 财团法人车辆研究测试中心 | Vehicle detection method, night vehicle detection method based on light intensity dynamic and system thereof |
| CN111191482B (en) * | 2018-11-14 | 2023-04-28 | 杭州海康威视数字技术股份有限公司 | Brake lamp identification method and device and electronic equipment |
| CN111191482A (en) * | 2018-11-14 | 2020-05-22 | 杭州海康威视数字技术股份有限公司 | Brake lamp identification method and device and electronic equipment |
| CN109785385B (en) * | 2019-01-22 | 2021-01-29 | 中国科学院自动化研究所 | Visual target tracking method and system |
| CN109785385A (en) * | 2019-01-22 | 2019-05-21 | 中国科学院自动化研究所 | Visual target tracking method and system |
| CN110276742B (en) * | 2019-05-07 | 2023-10-10 | 平安科技（深圳）有限公司 | Train tail lamp monitoring method, device, terminal and storage medium |
| CN110276742A (en) * | 2019-05-07 | 2019-09-24 | 平安科技（深圳）有限公司 | Tail light for train monitoring method, device, terminal and storage medium |
| CN110188645A (en) * | 2019-05-22 | 2019-08-30 | 北京百度网讯科技有限公司 | For the method for detecting human face of vehicle-mounted scene, device, vehicle and storage medium |
| CN110696807A (en) * | 2019-09-29 | 2020-01-17 | 一汽解放青岛汽车有限公司 | Engine shutdown control method and system under traffic jam condition, vehicle and storage medium |
| CN110758394A (en) * | 2019-09-29 | 2020-02-07 | 一汽解放青岛汽车有限公司 | Engine starting control method and system under traffic jam working condition, vehicle and storage medium |
| CN110758394B (en) * | 2019-09-29 | 2020-10-16 | 一汽解放青岛汽车有限公司 | Engine starting control method and system under traffic jam working condition, vehicle and storage medium |
| CN110673609A (en) * | 2019-10-10 | 2020-01-10 | 北京小马慧行科技有限公司 | Vehicle running control method, device and system |
| CN111652143A (en) * | 2020-06-03 | 2020-09-11 | 浙江大华技术股份有限公司 | Vehicle detection method and device and computer storage medium |
| CN111652143B (en) * | 2020-06-03 | 2023-09-29 | 浙江大华技术股份有限公司 | Vehicle detection method and device and computer storage medium |
| CN112164090A (en) * | 2020-09-04 | 2021-01-01 | 杭州海康威视系统技术有限公司 | Data processing method and device, electronic equipment and machine-readable storage medium |
| CN116994514A (en) * | 2023-09-26 | 2023-11-03 | 永林电子股份有限公司 | Image recognition-based vehicle display brightness control method, device and system |
| CN116994514B (en) * | 2023-09-26 | 2024-01-02 | 永林电子股份有限公司 | Image recognition-based vehicle display brightness control method, device and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108021856B (en) | 2020-09-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108021856A (en) | Light for vehicle recognition methods, device and vehicle | |
| CN106652465B (en) | Method and system for identifying abnormal driving behaviors on road | |
| Wu et al. | Lane-mark extraction for automobiles under complex conditions | |
| Kuo et al. | Vision-based vehicle detection for a driver assistance system | |
| Huang et al. | Vehicle detection and inter-vehicle distance estimation using single-lens video camera on urban/suburb roads | |
| Chen et al. | Nighttime vehicle detection for driver assistance and autonomous vehicles | |
| CN107891808B (en) | Driving reminding method and device and vehicle | |
| CN107886034B (en) | Driving reminding method and device and vehicle | |
| CN106650553A (en) | License plate recognition method and system | |
| CN104036262B (en) | A kind of method and system of LPR car plates screening identification | |
| CN104778444A (en) | Method for analyzing apparent characteristic of vehicle image in road scene | |
| CN103034843B (en) | Method for detecting vehicle at night based on monocular vision | |
| Chen | Nighttime vehicle light detection on a moving vehicle using image segmentation and analysis techniques | |
| CN102902957A (en) | Video-stream-based automatic license plate recognition method | |
| CN103366154A (en) | Reconfigurable clear path detection system | |
| CN111881832A (en) | Lane target detection method, device, equipment and computer readable storage medium | |
| CN105095869A (en) | Pedestrian detection method and apparatus | |
| CN103049788B (en) | Based on space number for the treatment of object detection system and the method for computer vision | |
| CN112464782A (en) | Pedestrian identification method and system | |
| Wen et al. | A rear-vehicle detection system for static images based on monocular vision | |
| Di et al. | Forward Collision Warning system based on vehicle detection and tracking | |
| Li et al. | A low-cost and fast vehicle detection algorithm with a monocular camera for adaptive driving beam systems | |
| CN101369312A (en) | Method and equipment for detecting intersection in image | |
| FAN et al. | Robust lane detection and tracking based on machine vision | |
| Helala et al. | Road boundary detection in challenging scenarios |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |