JP2008065756A - Driving assist system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a driving assist system for further stably detecting a pedestrian when a vehicle is traveling at night. <P>SOLUTION: The driving assist system corrects a pickup image acquired by an image pickup means for imaging the periphery of a vehicle including a road according to the amounts of vibration of a vehicle (S103), and detects a road area equivalent to a road from the corrected pickup image (S104), and decides whether or not the detected pedestrian candidate is a pedestrian based on a positional relationship between a pedestrian candidate detected from the pickup image and the detected road area (S107, S108). <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a driving support device that supports driving of a vehicle, and more particularly to a driving support device that can detect pedestrians around the vehicle.

  In recent years, with the development of information processing technology, a driving support device that acquires information related to traveling of a vehicle such as a two-wheeled vehicle or a four-wheeled vehicle and uses the information to support driving of the vehicle has been studied. As such a driving assistance device, for example, a device that detects a pedestrian in a surrounding area while a vehicle is running and notifies the driver of the result is attracting attention. In particular, when driving at night, it may be difficult for the driver to visually recognize a pedestrian, so the need for such a driving assistance device is high.

  As a driving support device that detects a pedestrian, for example, in an infrared image captured by two infrared cameras, an area corresponding to the head of the pedestrian and an area corresponding to the space above the shoulder of the pedestrian are set. And what recognizes a pedestrian based on those set area | regions is proposed (for example, refer patent document 1).

According to this driving support device, since the region corresponding to the pedestrian's head and shoulder captured in the infrared image can be recognized, the pedestrian's Detection can be performed more stably.
JP 2004-303219 A

  However, in the above conventional driving support apparatus, two cameras are required to use stereo vision. In that case, there is no problem when mounted on a four-wheeled vehicle, but the width of a two-wheeled vehicle is narrow. Therefore, there arises a problem that it is difficult to mount.

  In addition, there is a case where a stable image cannot be obtained because vibration is generated in a running vehicle. In the conventional driving support device, a non-pedestrian is mistakenly recognized as a pedestrian or a pedestrian. Sometimes things cannot be recognized as pedestrians. In particular, when a driving support device is mounted on a two-wheeled vehicle, the two-wheeled vehicle has a greater vibration than a four-wheeled vehicle, and thus it tends to be difficult to obtain a stable image.

  This invention is made | formed in view of such a situation, The objective is to provide the driving assistance apparatus which can perform a pedestrian's detection more stably.

  In order to solve the above-described problems, a driving support apparatus according to the present invention is mounted on a vehicle, and in the driving support apparatus that supports driving of the vehicle, an imaging unit that images the surroundings of the vehicle including a road, and the imaging Road area detection means for detecting a road area corresponding to a road from the captured image obtained by the means, pedestrian candidate detection means for detecting a pedestrian candidate from the captured image, and the road area detection means. Pedestrian determination means for determining whether or not the detected pedestrian candidate is a pedestrian based on the positional relationship between the road area and the pedestrian candidate detected by the pedestrian candidate detection means. Prepare.

  In this way, by determining whether the pedestrian candidate is a pedestrian based on the positional relationship between the pedestrian candidate and the road area, the pedestrian to be watched by the driver of the vehicle can be detected more stably. It becomes possible.

  In the driving support apparatus according to the above invention, the pedestrian determination means determines that the detected pedestrian candidate is a pedestrian when a lower part of the detected pedestrian candidate is in contact with the detected road area. It may be configured to determine that

  In the driving support device according to the invention, the road area detection unit determines whether the specific area is a road area based on the size of the area of the specific area in the captured image and its position. It may be configured to determine.

  Further, in the driving support device according to the above invention, the road area detection unit is configured to detect the specific area when the area of the specific area is larger than that of the other areas and the specific area is positioned below. May be determined to be a road area.

  In the driving support apparatus according to the invention, the imaging unit is a far-infrared camera, and the road region detection unit is configured to determine whether the specific region is a road region based on a luminance change of the specific region in the captured image. It may be configured to determine whether or not.

  When a far-infrared camera is used as the imaging unit, a small change in luminance in a captured image obtained as the imaging unit means that a temperature change is small. In the nighttime, the temperature on the road is usually substantially constant, and thus it is possible to detect the road more stably by paying attention to the luminance change.

  Further, in the driving support device according to the above invention, the road area detection unit is configured to determine the specific area as a road area when a change in luminance of the specific area is small compared to other areas. It may be.

  Further, in the driving support device according to the above invention, the imaging unit is a far-infrared camera, and the pedestrian candidate detection unit is configured to determine a region having a higher luminance than other regions as a pedestrian candidate. May be.

  In the driving support apparatus according to the above invention, the imaging unit is a far-infrared camera, and the pedestrian candidate detection unit includes a region having a higher luminance than other regions and a human head shape pattern. Based on the comparison result, it is determined whether or not the high brightness area is a head area corresponding to a pedestrian candidate's head, and the high brightness area is a head area. When it is determined, the region including the region with the high luminance may be determined as a pedestrian candidate.

  Further, in the driving support device according to the above invention, the pedestrian candidate detection means compares a region located below the high luminance region with a shape pattern other than the human head and based on the comparison result. And determining whether the region located below is a region corresponding to a portion other than the pedestrian candidate's head, and the region positioned below is a region corresponding to a portion other than the pedestrian candidate's head. When it is determined, the region including the high luminance region and the region located below may be determined as a pedestrian candidate.

  In the driving support device according to the invention, the imaging unit is a far-infrared camera, and the pedestrian candidate detecting unit is configured to calculate the luminance based on an aspect ratio of a region having a higher luminance than other regions. You may be comprised so that it may determine whether a high area | region is a pedestrian candidate.

  Further, in the driving support device according to the invention, the pedestrian candidate detection means compares the high luminance region with a shape pattern of a person's whole body, and based on the comparison result, the high luminance region is You may be comprised so that it may determine whether it is a pedestrian candidate.

  Moreover, the driving assistance apparatus according to the above invention may further include image correction means for correcting the captured image in accordance with the vibration amount of the vehicle.

  By comprising in this way, even if it is a case where disorder arises in a captured image by the vibration of a vehicle, a pedestrian can be detected more stably.

  Furthermore, in the driving support apparatus according to the above invention, the imaging unit is configured to acquire a plurality of captured images at a predetermined time interval, and the image correction unit is preceded by the imaging unit. The captured image and the subsequent captured image may be compared, and the subsequent captured image may be corrected based on the comparison result.

  Further, a driving support device according to the present invention is mounted on a vehicle, and in a driving support device that supports driving of the vehicle, from an imaging unit that images the periphery of the vehicle, and a captured image obtained by the imaging unit, Pedestrian detection means for detecting a pedestrian, the imaging means is a far-infrared camera, the pedestrian detection means includes a region having a higher brightness than other regions and a shape pattern of a human head. And based on the comparison result, it is determined whether or not the high brightness area is a head area corresponding to a pedestrian's head, and the high brightness area is a head area. When determined, the region including the high luminance region is determined to be a pedestrian.

  When a far-infrared camera is used as the image pickup means, the head of a person who is not affected by normal clothes is often expressed with high luminance, and therefore the detection can be performed easily and more stably. Therefore, by comprising as mentioned above, it becomes possible to detect a pedestrian more stably.

  In the driving support apparatus according to the above invention, the pedestrian detection means compares a region located below the high luminance region with a shape pattern other than a human head, and based on the comparison result, When it is determined whether or not the region located below is a region corresponding to other than the pedestrian's head, and the region located below is determined to be a region corresponding to other than the pedestrian's head The region including the high luminance region and the region located below may be determined as a pedestrian.

  Further, in the driving support device according to the above invention, the imaging unit is configured to image the surroundings of the vehicle including a road, and detects a road region corresponding to the road from the captured image obtained by the imaging unit. The pedestrian detection means further includes the high brightness area and the lower area based on a positional relationship between the road area detected by the road area detection means and the area located below. It may be configured to determine whether or not the region including the region located at is a pedestrian.

  Furthermore, the driving support apparatus according to the above invention may further include image correction means for correcting the captured image in accordance with the amount of vibration of the vehicle.

  Further, a driving support device according to the present invention is mounted on a vehicle, and in a driving support device that supports driving of the vehicle, from an imaging unit that images the periphery of the vehicle, and a captured image obtained by the imaging unit, Pedestrian detection means for detecting a pedestrian, the imaging means is a far-infrared camera, and the pedestrian detection means has a high aspect ratio and a high luminance ratio of a region having higher brightness than other regions. It is configured to determine whether or not the high luminance region is a pedestrian based on a result of comparing the region and the shape pattern of the whole body of the person.

  Further, in the driving support device according to the above invention, the imaging unit is configured to image the surroundings of the vehicle including a road, and detects a road region corresponding to the road from the captured image obtained by the imaging unit. And a pedestrian detecting means, wherein the pedestrian detecting means is a pedestrian based on the positional relationship between the road area detected by the road area detecting means and the high brightness. It may be configured to determine whether or not.

  Furthermore, the driving support apparatus according to the above invention may further include image correction means for correcting the captured image in accordance with the amount of vibration of the vehicle.

  According to the driving assistance apparatus of the present invention, it is possible to more stably detect a pedestrian on the road while the vehicle is traveling.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing the configuration of the driving support apparatus of the present invention. This driving support device is mounted on a motorcycle and includes an ECU (Electronic Control Unit) 10 that executes image processing as shown in FIG. The ECU 10 includes a microcomputer including a CPU 11, a ROM 12, a RAM 13, an interface (I / F) 14, and the like as main components, and performs an operation described later when a computer program stored in the ROM 12 is executed by the CPU 10. Further, data generated or acquired by the operation and to be temporarily stored is stored in the RAM 13.

  The ROM 12 stores predetermined data such as head feature data, body feature data, whole body feature data, and pedestrian feature data, which will be described later.

  The ECU 10 is connected to the far-infrared camera 21, the display unit 22, and the alarm sound output device 23 via the I / F 14.

  The far-infrared camera 21 is an imaging device that measures the temperature of the object surface and displays the temperature distribution as a two-dimensional black and white image. The far-infrared camera 21 repeatedly executes imaging processing at predetermined time intervals in accordance with a command output from the ECU 10 via the I / F 14 and outputs a captured image obtained as a result to the ECU 10 via the I / F 14. To do.

  The display device 22 is a display used for notifying the motorcycle driver of information related to detection of a pedestrian. The display device 22 may perform color display or monochrome display.

  The warning sound output device 23 is a device for outputting a sound warning that is issued to alert the driver when a pedestrian is detected. The alarm sound output device 23 executes an alarm sound output process in accordance with a command output from the ECU 10 via the I / F 14.

  Note that, as described above, in the present embodiment, the ECU 10 performs an operation described later, but the present invention is not limited to this aspect, and for example, a dedicated processor for performing image processing, etc. The driving support device of the present invention may be configured.

  Moreover, in this Embodiment, although the driving assistance apparatus 1 is mounted in the two-wheeled vehicle, it is not necessarily limited to this, Of course, it may be mounted in the four-wheeled vehicle etc.

  Next, operation | movement of the driving assistance apparatus 1 comprised as mentioned above is demonstrated.

  FIG. 2 is a flowchart showing a flow of operation of the driving support apparatus 1 according to Embodiment 1 of the present invention. The operation of the driving support device 1 shown here is controlled by the ECU 10. Further, the following operation is executed at an appropriate timing during the operation of the motorcycle.

  First, the driving support device 1 executes an initialization process such as securing a predetermined storage area or setting an initial value to a predetermined variable (S101), and then acquired by the far-infrared camera 21. An input of a captured image is accepted (S102).

  Next, the driving assistance apparatus 1 performs a vibration cancellation process (S103). Here, the vibration canceling process is to avoid a situation in which the detection of the pedestrian is not stably performed due to the captured image being disturbed due to the vibration of the vehicle during traveling. A process for correcting a captured image.

[Vibration cancellation processing]
FIG. 3 is a flowchart showing a flow of operation of the driving support apparatus 1 according to the present invention accompanying the vibration canceling process. FIG. 4 is an explanatory diagram showing a concept of an effective pixel area and a processing process of vibration cancellation processing to be described later. Hereinafter, the vibration canceling process will be described with reference to FIGS. 3 and 4.

  First, the driving support apparatus 1 reads a captured image (hereinafter referred to as a previous captured image) already obtained by the far-infrared camera 21 (S201), and sets an effective pixel region for the read previous captured image (S201). S202).

  Here, the effective pixel area is an area provided in a pixel area (hereinafter referred to as an imageable pixel area) that can be imaged by the far-infrared camera 21, and is used for pedestrian detection.

  FIG. 4A shows the previous captured image. In FIG. 4A, R2 indicates the effective pixel region set in step S202. R1 represents an imageable pixel region.

  As shown in FIG. 4A, the effective pixel region R2 is a pixel region having a predetermined size provided in the imageable pixel region R1. Processing related to pedestrian detection, which will be described later, is executed using the image in the effective pixel region R2.

  After executing step S202, the driving support device 1 reads a captured image (hereinafter referred to as a current captured image) obtained by the far-infrared camera 21 next to the previous captured image (S203).

  Then, the driving assistance device 1 executes a gray search using the previous captured image that precedes in time and the current captured image that follows the current captured image (S204). Specifically, the degree of coincidence between the previous captured image and the current captured image is calculated, and a point in the current captured image that matches the previous captured image is detected.

  The driving support device 1 determines how much deviation has occurred between the previous captured image and the current captured image by the gray search in step S204, and calculates the shift as a shift amount (S205).

  Next, the driving assistance apparatus 1 moves the effective pixel area in the current captured image based on the shift amount calculated in step S205 (S206). Hereinafter, this process will be described with reference to FIGS. 4B and 4C.

  FIG. 4B shows a captured image this time. As can be seen from a comparison between FIGS. 4A and 4B, it is assumed that there is a difference between the previous captured image and the current captured image due to vibration during traveling of the two-wheeled vehicle on which the driving support device 1 is mounted. Therefore, the display range of objects (pedestrians, buildings, roads, etc.) appearing in the image differs between the image in the effective pixel region R2 in the previous captured image and the image in the effective pixel region R2 in the current captured image. (For example, the effective pixel region R2 in the previous captured image includes the whole body of the pedestrian, but the effective pixel region R2 in the current captured image does not include a part of the pedestrian's head.) Has occurred.)

  Similar to FIG. 4B, FIG. 4C also shows the current captured image. As shown in FIG. 4C, the driving support device 1 moves the effective pixel region R2 by the shift amount calculated in step S205 in step S206. In FIG. 4C, the effective pixel region after movement is indicated by R3. As a result, the image in the effective pixel region R3 in the current captured image is the same as the image in the effective pixel region R2 in the previous captured image.

  The current captured image within the effective pixel region corrected in this way is stored in the RAM 13 as captured image data by the driving support device 1 (S207).

  By moving the effective pixel region as described above, it is possible to correct a shift caused by vibration.

    In addition, the vibration of the vehicle is larger in the case of a two-wheeled vehicle than in the case of a four-wheeled vehicle. Therefore, it can be said that the vibration canceling process as described above is more effective in the case of a two-wheeled vehicle.

  The driving assistance device 1 that has finished the vibration canceling process executes a road detection process as shown in FIG. 2 (S104). Here, the road detection process refers to a process for detecting a road area indicating a road in the captured image data.

[Road detection processing]
FIG. 5 is a flowchart showing a flow of operations of the driving support apparatus 1 according to the present invention accompanying the road detection process.

  The driving support device 1 performs an image differentiation process on the captured image data stored in the RAM 13 in order to calculate the change value of the brightness of the image (S301). As a result, an edge portion of the image is detected.

  Next, the driving support device 1 performs binarization processing on the captured image data in which the edge portion is detected, and detects an area where the luminance change is small (S302). Then, the driving support device 1 executes a labeling process for sequentially attaching labels to the connected areas in the binarized captured image data (S303).

  Next, the driving support device 1 sets a specific one of the labeled regions as a processing target of the road detection process (S304), and the area of the processing target is a value set from the entire area. It is determined whether the position is large and the position is below (S305). Here, the value set from the entire area is, for example, when the far-infrared camera 21 is mounted substantially horizontally and the road is set to appear in the lower half of the screen (in this case, the road is the maximum on the entire screen). In 1/2), surrounding buildings and the like are also reflected, so the area is about 1/5 to 1/10 of the entire screen. As described above, this value is appropriately set depending on the mounting position of the far-infrared camera 21.

  The determination process in step S305 is performed depending on whether the area to be processed is larger than the value set as described above, whether the area is in the lower half of the effective pixel area, or the like.

  If it is determined in step S305 that the area to be processed is not larger than the value set from the entire area or that the position is not below (NO in S305), the driving support device 1 proceeds to step S309 described later. .

  On the other hand, when it is determined in step S305 that the area to be processed is larger than the value set from the entire area and the position is below (YES in S305), the driving support device 1 determines that the process target is a road. The road area data relating to the road area is generated using the data relating to the position, area, etc. of the processing target (S306). Then, the driving assistance device 1 stores the road area data in the RAM 13 (S307).

  Next, the driving assistance device 1 determines whether or not all labeled regions have been set as processing targets (S308), and determines that there is a region that has not yet been processed (NO in S308). ), The process returns to step S304, and the above steps S304 to S307 are executed. On the other hand, when it determines with all the area | region becoming processing object (it is YES at S308), the driving assistance apparatus 1 complete | finishes a road detection process.

  As described above, in the present invention, the road area is detected based on the degree of change in luminance of the image. Here, a small luminance change in a captured image obtained by a far-infrared camera means that a temperature change is small. In the nighttime, the temperature on the road is usually substantially constant, and thus it is possible to detect the road more stably by paying attention to the luminance change.

  The driving support device 1 that has finished the road detection process performs the building detection process as shown in FIG. 2 (S105). Here, the building detection process refers to a process for detecting a building region indicating a building in the captured image data.

[Building detection processing]
FIG. 6 is a flowchart showing a flow of operations of the driving support apparatus 1 according to the present invention accompanying the building detection process.

  First, the driving assistance device 1 sets one specific area among the already labeled areas as a processing target of the building detection process (S401), and whether or not the area of the processing target is a predetermined value or more. Is determined (S402). If it is determined that the value is not equal to or greater than the predetermined value (NO in S402), the driving assistance apparatus 1 proceeds to step S408 described later.

  On the other hand, if it is determined in step S402 that the area of the processing target is equal to or larger than the predetermined value (YES in S402), the driving support device 1 recognizes the horizontal and vertical sizes of the processing target, It is determined whether the vertical length is 2 or more (S403). Here, when it is determined that the horizontal / vertical direction is not 2 or more (NO in S403), the driving support device 1 determines whether the vertical / horizontal direction is 5 or more (S404). If it is determined that the length / width is not 5 or more (NO in S404), the driving support device 1 determines whether the area is equal to or larger than a predetermined value (a value larger than the predetermined value in Step S402). Is determined (S405). If it is determined that the value is not equal to or greater than the predetermined value (NO in S405), the driving assistance apparatus 1 proceeds to step S408 described later.

  If it is determined in step S403 that horizontal / vertical is 2 or more (YES in S403), if it is determined in step S404 that vertical / horizontal is 5 or more (YES in S404), or the area is determined in step S405. When it is determined that the value is equal to or greater than the predetermined value (YES in S405), the driving support device 1 determines that the processing target is a region (building region) corresponding to a building, and relates to the position, area, and the like of the processing target. Using the data, building area data relating to the building area is generated (S406). Then, the driving support device 1 stores the building area data in the RAM 13 (S407).

  Next, the driving assistance apparatus 1 determines whether or not all labeled regions have been set as processing targets (S408), and determines that there is a region that has not yet been processed (NO in S408). ), Returning to step S401, the above steps S401 to S407 are executed. On the other hand, when it determines with all the area | region becoming processing object (it is YES at S408), the driving assistance apparatus 1 complete | finishes a building detection process.

  As described above, when it is determined that the region to be processed has a predetermined area and has a predetermined shape, the driving support device 1 determines that the processing target indicates a building. In the present embodiment, when the horizontal / longitudinal shape is 2 or more, when the vertical / horizontal shape is 5 or more, or when the area is a predetermined value or more, the region is a building. It is judged that it shows. In addition, these are illustrations, and it cannot be overemphasized that the area | region of other shapes is good also as a building.

  The driving support device 1 that has completed the building detection process performs a disturbance elimination process for erasing an area determined to be a building as shown in FIG. 2 (S106). Next, the driving assistance apparatus 1 performs a 1st pedestrian detection process (S107). Here, a 1st pedestrian detection process means the process which detects a pedestrian by detecting separately a pedestrian's head (a face is included) and the other.

[First pedestrian detection process]
FIG. 7 is a flowchart showing a flow of the operation of the driving support device 1 of the present invention accompanying the first pedestrian detection process.

  First, the driving support device 1 acquires head feature data indicating a head feature value that represents a pedestrian's head feature as a numerical value from the ROM 12 (S501), and also calculates a pedestrian's body feature as a numerical value. The body feature amount data indicating the body feature amount represented by (2) is acquired in the same manner (S502).

  Here, the head feature amount means data obtained by digitizing a human head shape pattern, and specific examples include data obtained by digitizing a vertically long elliptical region. The body feature amount means data obtained by digitizing a shape pattern other than the human head, and specific examples include data obtained by digitizing a vertically long rectangular area.

  Next, the driving assistance apparatus 1 extracts the area | region (head candidate) estimated to correspond to a pedestrian's head from the already labeled area | region (S503). Specifically, a region having a higher brightness (that is, a higher temperature) than the other regions and having an area of a predetermined value or less is extracted as a head candidate.

  Next, the driving support device 1 sets a specific head candidate as a processing target (S504), and the processing target has the head feature amount indicated by the head feature amount data acquired in step S501. It is determined whether or not there is (S505).

  If it is determined in step S505 that the processing target does not have the head feature amount (NO in S505), the driving support device 1 proceeds to step S508 described later. On the other hand, when it is determined that the processing target has a head feature amount (YES in S505), the driving support device 1 indicates that the processing target indicates a pedestrian's head (head region). And using the data relating to the position, area, etc. of the processing target, head region data relating to the pedestrian's head is generated (S506). Then, the driving assistance device 1 stores the head area data in the RAM 13 (S507).

  Next, the driving assistance device 1 determines whether or not all head candidates have been set as processing targets (S508), and determines that there is a head candidate that has not yet been processed (NO in S508). ), The process returns to step S504, and the above steps S504 to S507 are executed. On the other hand, if it is determined that all head candidates have been processed (YES in S508), the driving support device 1 determines a specific head region from among the head regions indicated in the head region data stored in the RAM 13. One is selected, and a labeled area that is in contact with the lower part of the head area is set as a processing target (S509).

  Next, the driving assistance apparatus 1 determines whether or not the processing target has the body feature amount indicated by the body feature amount data acquired in step S502 (S510). Here, if it is determined that the processing target does not have a body feature amount (NO in S510), the driving support device 1 proceeds to step S514 described later.

  On the other hand, when it is determined in step S510 that the processing target has a body feature amount (YES in S510), the driving support device 1 determines that the processing target and the head in contact with the processing target are above. It is determined that the area is a pedestrian candidate, and it is determined whether or not the lower part of the processing target is in contact with the road area indicated by the road area data stored in the RAM 13 (S511).

  If it is determined in step S511 that the lower part of the processing target is not in contact with the road area (NO in S511), the driving assistance apparatus 1 proceeds to step S514 described later. On the other hand, when it is determined that the lower part of the processing target is in contact with the road area (YES in S511), the driving support device 1 indicates that the processing target and the head area in contact with the processing target are above the pedestrian. It is determined that the region is a corresponding region (pedestrian region), and pedestrian region data regarding the pedestrian region is generated using data regarding the position, area, etc. of the processing target (S512). Then, the driving assistance device 1 stores the pedestrian area data in the RAM 13 (S513).

  Next, the driving assistance apparatus 1 determines whether or not a labeled area located below all head areas has been set as a processing target (S514), and there is an area that has not yet been processed. (NO in S514), the process returns to step S509, and the above steps S509 to S513 are executed. On the other hand, when it determines with all the areas having been processed (it is YES at S514), the driving assistance device 1 complete | finishes a 1st pedestrian detection process.

  When a far-infrared camera is used as an imaging unit as in the present embodiment, a captured image in which a region having a high temperature is displayed with high luminance is obtained. Usually, the temperature of the surface of the part other than the pedestrian's head is lowered by clothing, while the temperature of the surface of the head is increased, so the head is more stable than the part other than the head. It can often be detected. Therefore, as in the first pedestrian detection process described above, a region corresponding to the pedestrian's head is detected, and then a region corresponding to the body of the pedestrian in contact with the region below is detected. It becomes possible to detect pedestrians more stably.

  FIG. 8 is a conceptual diagram showing an example of a captured image used in the driving support device 1 of the present invention. In FIG. 8, W indicates a pedestrian area detected by the driving support device 1.

  As shown in FIG. 8, the temperature of the surface of the pedestrian's torso is low due to clothing, and is therefore often expressed with lower brightness than the surrounding area. On the other hand, since the temperature of the surface of the head is high, it is expressed with higher luminance than the surrounding area. In such a case, the driving assistance device 1 detects the head of the pedestrian in contact with the lower part of the pedestrian area W after detecting the head that can be detected more stably first. Is detected. Thereby, a pedestrian can be detected more stably.

  Further, as described above in step S511, it is possible to more stably detect a pedestrian that should be noted by the driver of the motorcycle by determining whether or not the lower part of the processing target is in contact with the road. Furthermore, when a shape that resembles a person appears in the captured image, although it is not actually a person, it is erroneously detected that the shape is a pedestrian by checking the positional relationship between the shape and the road. It is possible to prevent this.

  In this embodiment, it is determined in step S511 whether or not the processing target is in contact with the road. However, this processing is omitted, and the processing target has a head feature value and a body feature value. The processing target may be determined as a pedestrian area. In that case, the driving assistance apparatus 1 does not need to perform the road detection process mentioned above.

  The driving assistance device 1 that has completed the first pedestrian detection process executes the second pedestrian detection process as shown in FIG. 2 (S108). Here, the second pedestrian detection process refers to a process of detecting a pedestrian by detecting the whole body of the pedestrian as a whole.

[Second pedestrian detection process]
FIG. 9 is a flowchart showing a flow of the operation of the driving support apparatus 1 of the present invention accompanying the second pedestrian detection process.

  First, the driving support device 1 acquires whole body feature quantity data indicating the whole body feature quantity representing the whole body feature of the pedestrian numerically from the ROM 12 (S601), and also represents the feature of the shape pattern of the pedestrian numerically. The pedestrian feature value data indicating the pedestrian feature value thus obtained is acquired in the same manner (S602).

  Here, the whole body feature value means data obtained by digitizing the shape pattern of a person's whole body, and a specific example is a digitized vertical region having an aspect ratio of about 2: 1 to 3: 1. As mentioned. The pedestrian feature value means data obtained by digitizing a person's shape pattern, and the presence / absence determination is made as described later.

  The driving support device 1 extracts a labeled region having a higher brightness than the other regions and having an area of a predetermined value as a whole body candidate corresponding to the whole body of the pedestrian (S603). Then, the driving support device 1 sets a specific one of the whole body candidates as a processing target (S604).

  Next, the driving assistance apparatus 1 determines whether or not the processing target has the whole body feature amount indicated by the whole body feature amount data acquired in step S601 (S605).

  If it is determined in step S605 that the processing target does not have the whole body feature amount (NO in S605), the driving support device 1 proceeds to step S608 described later. On the other hand, when it is determined that the processing target has a whole body feature amount (YES in S605), the driving support device 1 is a region (whole body region) corresponding to the whole body of the pedestrian. And using the data relating to the position, area, etc. of the processing target, the whole body region data relating to the whole body of the pedestrian is generated (S606). Then, the driving support device 1 stores the whole body region data in the RAM 13 (S607).

  Next, the driving assistance apparatus 1 determines whether all the whole body candidates are set as processing targets (S608), and when it is determined that there are whole body candidates that are not yet processing targets (NO in S608), Returning to step S604, the above steps S604 to S607 are executed. On the other hand, when it is determined that all the whole body candidates have been processed (YES in S608), the driving support device 1 selects a specific one of the whole body areas indicated in the whole body area data stored in the RAM 13. The process target is set (S609).

Next, the driving assistance device 1 determines whether or not the processing target has the pedestrian feature amount indicated by the pedestrian feature amount data acquired in step S602 (S610). Specifically, the processing target is divided into a plurality of areas, and the processing target is determined to be a pedestrian by determining whether the processing target corresponds to a human shape pattern based on the density of the divided areas. It is determined whether or not it has a feature amount. Here, when it is determined that the processing target does not have the pedestrian feature amount (NO in S610), the driving support device 1 proceeds to Step S614 described later.
On the other hand, when it determines with the process target having a pedestrian feature-value in step S608 (it is YES at S608), the driving assistance apparatus 1 judges that the process target is a pedestrian candidate, It is determined whether the lower part to be processed is in contact with the road area indicated by the road area data stored in the RAM 13 (S611).
If it is determined in step S611 that the lower part of the processing target is not in contact with the road area (NO in S611), the driving assistance apparatus 1 proceeds to step S614 described later. On the other hand, when it is determined that the lower part of the processing target is in contact with the road area (YES in S611), the driving support device 1 determines that the processing target is an area corresponding to a pedestrian (pedestrian area). The pedestrian area data relating to the pedestrian is generated using the data relating to the position, area, etc. of the processing target (S612). Then, the driving support device 1 stores the pedestrian area data in the RAM 13 (S613).

  Next, the driving assistance device 1 determines whether or not all the whole body regions have been set as processing targets (S614), and if it is determined that there is a whole body region that has not yet been processed (NO in S614), Returning to step S609, the above steps S609 to S613 are executed. On the other hand, when it determines with all the whole body area | region becoming a process target (it is YES at S614), the driving assistance device 1 complete | finishes a 2nd pedestrian detection process.

  FIG. 10 is an explanatory diagram showing the concept of the pedestrian area detected by the second pedestrian detection process. The specific region W shown in FIG. 10 is a vertically long region having an aspect ratio of about 2: 1 to 3: 1. Therefore, the driving assistance device 1 determines in step S605 that this region W has a whole body feature amount.

  Also, as shown in FIG. 10, when the region W is divided into 5 × 8 horizontal regions, the density of the divided regions corresponds to a human shape pattern. In this case, the driving assistance apparatus 1 determines in step S610 that the region W has a pedestrian feature amount.

    As described above, the driving support device 1 determines that the region W has the whole body feature amount and the pedestrian feature amount. If the lower portion of the region W is in contact with the road region, the driving support device 1 walks in this region W. Pedestrian area data is generated and stored as a person area.

  FIG. 11 is a conceptual diagram showing an example of a captured image used in the driving support device 1 of the present invention. In FIG. 11, W indicates a pedestrian area detected by the driving support device 1.

  As shown in FIG. 11, when a region indicating the pedestrian's whole body appears in the captured image, the driving support device 1 has the whole body feature amount and the pedestrian feature amount as described above. It is determined that And when it determines with the lower part of the area | region contacting the road area | region, the said area | region will be detected as the pedestrian area | region W. FIG.

  In this embodiment, it is determined whether or not the processing target is in contact with the road in step S611. However, this processing is omitted, and the processing target has a whole body feature amount and a pedestrian feature amount. The processing target may be determined as a pedestrian area. In that case, the driving assistance apparatus 1 does not need to perform the road detection process mentioned above.

  The driving assistance apparatus 1 that has finished the second pedestrian detection process executes a process result output process as shown in FIG. 2 (S109). Here, the process result output process refers to a process for notifying the driver of the motorcycle of the process results of the first pedestrian detection process and the second pedestrian detection process.

[Detection result output processing]
FIG. 12 is a flowchart showing a flow of operations of the driving support apparatus 1 according to the present invention accompanying the detection result output process.

  The driving support device 1 determines whether or not pedestrian area data is stored in the RAM 13 (S701). Here, when it is determined that the pedestrian area data is not stored (NO in S701), the driving support device 1 proceeds to Step S707 described later. On the other hand, when it is determined that the pedestrian area data is stored (YES in S701), the driving support device 1 sets a specific one of the stored pedestrian area data as a processing target. (S702).

  Next, the driving assistance device 1 performs a pedestrian frame drawing process indicating a pedestrian area based on the pedestrian area data to be processed (S703). Then, the driving support device 1 determines whether or not all pedestrian area data has been processed (S704), and determines that there is pedestrian area data that has not yet become a predetermined target (NO in S704). ), The process returns to step S702, and steps S702 and S703 are executed.

  If it is determined in step S704 that all pedestrian area data has been processed (YES in S704), the driving support apparatus 1 determines whether 5 seconds have elapsed since the previous display of the pedestrian frame. Is determined (S705). Here, if it is determined that 5 seconds have not elapsed (NO in S705), the driving assistance apparatus 1 proceeds to step S707 described later. On the other hand, when it is determined that 5 seconds have elapsed (YES in S705), the driving support device 1 causes the alarm sound output device 23 to output a sound alarm (S706).

  In step S707, the driving support device 1 updates the image display by causing the display device 22 to display the drawn pedestrian frame.

  In this way, the driving assistance device 1 notifies the motorcycle driver that a pedestrian has been detected. Thereby, the driver can recognize the presence of the pedestrian.

  As described above, in the driving support device 1 of the present invention, unlike the conventional driving support device, since it is not necessary to use stereo vision, only one camera as an imaging unit is sufficient. Therefore, even when the driving support device is mounted on a motorcycle, no particular trouble occurs.

  The driving support device according to the present invention is useful for a driving support device that notifies the driver of a vehicle whether or not a pedestrian has been detected.

It is a block diagram which shows the structure of the driving assistance device of this invention. It is a flowchart which shows the flow of operation | movement of the driving assistance device in Embodiment 1 of this invention. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a vibration cancellation process. It is explanatory drawing which shows the concept of an effective pixel area | region, and the process of a vibration cancellation process. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a road detection process. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a building detection process. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a 1st pedestrian detection process. It is a conceptual diagram which shows an example of the captured image utilized in the driving assistance device of this invention. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a 2nd pedestrian detection process. It is explanatory drawing which shows the concept of the pedestrian area | region detected by the 2nd pedestrian detection process. It is a conceptual diagram which shows an example of the captured image utilized in the driving assistance device of this invention. It is a flowchart which shows the flow of operation | movement of the driving assistance device of this invention accompanying a detection result output process.

Explanation of symbols

1 Driving support device 10 ECU
11 CPU
12 ROM
13 RAM
14 Interface 21 Far-infrared camera 22 Display device 23 Alarm sound output device W Pedestrian area

Claims (20)

In a driving support device mounted on a vehicle and supporting driving of the vehicle,
Imaging means for imaging the periphery of the vehicle including a road;
Road area detection means for detecting a road area corresponding to a road from a captured image obtained by the imaging means;
Pedestrian candidate detection means for detecting a pedestrian candidate from the captured image;
Whether or not the detected pedestrian candidate is a pedestrian based on the positional relationship between the road area detected by the road area detection means and the pedestrian candidate detected by the pedestrian candidate detection means. A driving support device comprising: a pedestrian determination means for determining.   The pedestrian determination means is configured to determine that the detected pedestrian candidate is a pedestrian when a lower part of the detected pedestrian candidate is in contact with the detected road region. The driving support device according to claim 1.   The road area detection means is configured to determine whether or not the specific area is a road area based on the size of the area of the specific area in the captured image and its position. The driving support device according to claim 1 or 2.   The road area detection means determines that the specific area is a road area when the area of the specific area is larger than a value set based on the entire area and the specific area is positioned below. The driving support device according to claim 3, configured as described above. The imaging means is a far-infrared camera;
The said road area detection means is comprised so that it may determine whether the said specific area is a road area based on the brightness | luminance change of the specific area in the said captured image. The driving support device according to 1.   The driving support according to claim 5, wherein the road area detection unit is configured to determine the specific area as a road area when a change in luminance of the specific area is smaller than other areas. apparatus. The imaging means is a far-infrared camera;
The driving support device according to claim 1, wherein the pedestrian candidate detection unit is configured to determine a region having a higher brightness than other regions as a pedestrian candidate. The imaging means is a far-infrared camera;
The pedestrian candidate detection means compares a region having a higher brightness than the other regions with a shape pattern of a human head, and based on the comparison result, the region having a higher brightness is the head of the pedestrian candidate. It is determined whether or not it is a head region corresponding to a part, and when it is determined that the high luminance region is a head region, the region including the high luminance region is determined as a pedestrian candidate The driving support device according to claim 1, which is configured as follows.   The pedestrian candidate detection means compares a region located below the high brightness region with a shape pattern other than a human head, and based on the comparison result, the region located below is a pedestrian. It is determined whether or not the region corresponds to a region other than the candidate's head, and when it is determined that the region located below is a region corresponding to a region other than the pedestrian candidate's head, the region having the high brightness The driving support device according to claim 8, wherein the driving support device is configured to determine a region including the lower region as a pedestrian candidate. The imaging means is a far-infrared camera;
The pedestrian candidate detection means is configured to determine whether or not the high luminance area is a pedestrian candidate based on the aspect ratio of the high luminance area compared to other areas. Item 2. The driving support device according to Item 1.   The pedestrian candidate detection means compares the high luminance area with a shape pattern of the whole body of the person, and determines whether the high luminance area is a pedestrian candidate based on the comparison result. The driving support device according to claim 10, configured as described above.   The driving support device according to claim 1, further comprising image correction means for correcting the captured image in accordance with a vibration amount of the vehicle. The imaging means is configured to acquire a plurality of captured images at predetermined time intervals,
The image correcting unit is configured to compare a preceding captured image obtained by the imaging unit with a subsequent captured image and correct the subsequent captured image based on the comparison result. The driving support device according to claim 12. In a driving support device mounted on a vehicle and supporting driving of the vehicle,
Imaging means for imaging the periphery of the vehicle;
Pedestrian detection means for detecting a pedestrian from a captured image obtained by the imaging means,
The imaging means is a far-infrared camera;
The pedestrian detection means compares a region having a higher brightness than the other regions with the shape pattern of the human head, and based on the comparison result, the region having the higher brightness is applied to the pedestrian's head. It is configured to determine whether or not it is a corresponding head region, and when it is determined that the high luminance region is a head region, the region including the high luminance region is determined to be a pedestrian. A driving assistance device.   The pedestrian detection means compares a region located below the high luminance region with a shape pattern other than a human head, and based on the comparison result, the region located below is the pedestrian's It is determined whether the region corresponds to a region other than the head, and when it is determined that the region located below is a region corresponding to a region other than the pedestrian's head, the high luminance region and the lower portion The driving support device according to claim 14, wherein the driving support device is configured to determine a region including a region located at a pedestrian. The imaging means is configured to image the periphery of the vehicle including a road,
Road area detection means for detecting a road area corresponding to a road from the captured image obtained by the imaging means;
The pedestrian detection unit is configured to walk on a region including the high luminance region and the lower region based on a positional relationship between the road region detected by the road region detection unit and the lower region. The driving support device according to claim 15, wherein the driving support device is configured to determine whether or not the vehicle is a person.   The driving support device according to claim 14, further comprising image correction means for correcting the captured image in accordance with a vibration amount of the vehicle. In a driving support device mounted on a vehicle and supporting driving of the vehicle,
Imaging means for imaging the periphery of the vehicle;
Pedestrian detection means for detecting a pedestrian from a captured image obtained by the imaging means,
The imaging means is a far-infrared camera;
The pedestrian detection unit is configured to walk the high-brightness region on the basis of the aspect ratio of the high-brightness region compared to other regions and the result of comparing the high-brightness region and the shape pattern of the whole body of the person. A driving support device configured to determine whether or not the vehicle is a person. The imaging means is configured to image the periphery of the vehicle including a road,
Road area detection means for detecting a road area corresponding to a road from the captured image obtained by the imaging means;
The pedestrian detection means is configured to determine whether or not the high brightness area is a pedestrian based on a positional relationship between the road area detected by the road area detection means and the high brightness. The driving support device according to claim 18, wherein   The driving support device according to claim 18, further comprising image correction means for correcting the captured image in accordance with a vibration amount of the vehicle.