CN105975972B - Bridge Crack detection and feature extracting method based on image - Google Patents
Bridge Crack detection and feature extracting method based on image Download PDFInfo
- Publication number
- CN105975972B CN105975972B CN201610269077.8A CN201610269077A CN105975972B CN 105975972 B CN105975972 B CN 105975972B CN 201610269077 A CN201610269077 A CN 201610269077A CN 105975972 B CN105975972 B CN 105975972B
- Authority
- CN
- China
- Prior art keywords
- crack
- image
- area
- gray
- obtains
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
- G01N2021/889—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques providing a bare video image, i.e. without visual measurement aids
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Immunology (AREA)
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biochemistry (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention provide it is a kind of based on image Bridge Crack detection and feature extracting method.A kind of Bridge Crack detection based on image includes the following steps: with feature extracting method Step 1: bridge floor Image Acquisition；Step 2: image preprocessing；Step 3: obtaining crack candidate regions；Step 4: crack area enhances；Step 5: feature extraction；Step 6: crack result is extracted.Bridge Crack based on image of the invention detects and feature extracting method, is based on bridge surface image, detects crack area and simultaneously extracts Expressive Features, algorithm is simple, efficient, and tests to test and show precision with higher.
Description
Technical field
The present invention relates to magnetic material measuring device field, in particular to a kind of Bridge Crack detection and spy based on image
Levy extracting method.
Background technique
With the national projects such as China's economic development, urbanization process quickening, high-speed rail develop rapidly, highway, railway or
It is that all kinds of bridge numbers of the leaping over obstacles in the rural water conservancy construction of city, built increasingly are increased sharply, bridge is in the national economic development
In play very important effect, while being also a kind of embodiment of China's comprehensive strength.Due to the universal existence of bridge, pontic
The safety of structure and persistence can not be ignored.Crack is as a kind of main crossstructure Disease Characters, to crossstructure
The harm that durability and safety generate is maximum, and therefore, crack is one of primary evaluation index of its health status.
Current detection method is still based on artificial detection, place that there are many deficiencies:
(1) detection efficiency is low: it is time-consuming, need to install or remove the equipment such as hand cradle；
(2) detection accuracy is low: mainly carrying out observation detection with human eye, is easy to be influenced by the subjective factor of people；
(3) large labor intensity: bridge is more, detects heavy workload, simple by being accomplished manually, intensity is very big；
(4) safety is low: testing staff need under to detecting under bridge, safety does not ensure；
(5) at high cost: to be detected, spent high using a large amount of human and material resources；
(6) level of informatization is low: can not accurately establish Bridge Crack historical data, be not easy to the management and dimension of Dangerous Bridge
Shield, can not also provide decision support information to government administration section.
These deficiencies cause current detection status not adapt to bridge construction and development instantly completely.
In recent years, the algorithm for extracting crack on road based on image vision method detection box is put forward one after another, this makes to attain the Way
There is biggish development in the automation in road crack, intellectualized detection.Bridge structure crack and pavement of road Crack Detection are similar
Seemingly, but the former is increasingly complex, is mainly shown as in terms of two: first, the complexity of bridge structure leads to view-based access control model image side
Method difficulty when obtaining data is significantly greatly increased, and bridge structure upper surface is almost the same with road, and data relatively easily obtain, at present
Also there is practical system to put into production, still, all do not occur effective Data acquisition and processing system so far for its following table
Face；Second, pavement of road textural characteristics are relatively easy, single, and FRACTURE CHARACTERISTICS is generally with uniformity, and bridge bottom surface line
Manage relative complex, it is a large amount of " noise " that there are a large amount of spots, stain, water stain, detection of sign etc., the difficulty of Crack Detection and extraction
It spends bigger.This two o'clock strongly limits the automatic detection of bridge structure crack and intelligent structural safety monitoring.
The different understanding of fracture feature so that it has been proposed that crack detection method it is also various, most of algorithm
The essential characteristic utilized is consistent, and the process of algorithm is also roughly the same: pretreatment, crack area detection and segmentation, after
Processing is described with feature.Crack seems simply as one kind, but have because of its background and structure feature itself variability and
The target of complexity, existing crack on road detection algorithm is there are still more defect, far from meeting its demand.
In short, the feature for detecting crack is varied, but simple and efficient detection or a difficult point,
How the crack of numerous and disorderly multiplicity preferably to be separated with background characteristics, how rapidly extracting FRACTURE CHARACTERISTICS quickly rebuilds crack knot
Structure feature is all challenging problem.This patent solves these problems to a certain extent, proposes based on form
Background removal algorithm and the method for being clustered and being divided based on grayscale information, realize the Crack Detection under complex scene, it
Two-dimensional topology optimized relation is realized using the counterincision crack structure containing structural information of characteristic curve afterwards.
Summary of the invention
In order to solve since pavement of road textural characteristics are relatively easy, single in existing highway bridge detection process, FRACTURE CHARACTERISTICS
It is general with uniformity, and bridge bottom surface texture is relative complex, there are a large amount of spot, stain, water stain, detection of sign
Deng a large amount of " noises ", the big technical problem of the difficulty of Crack Detection and extraction, the present invention provide it is a kind of can splitting numerous and disorderly multiplicity
Seam is preferably separated with background characteristics, and energy rapidly extracting FRACTURE CHARACTERISTICS quickly rebuilds the bridge based on image of fissured structure feature
Beam Crack Detection and feature extracting method.
Bridge Crack detection and feature extracting method provided by the invention based on image, includes the following steps:
Step 1: bridge floor Image Acquisition；
Step 2: image preprocessing: using Gaussian smoothing filter, weakens the noise jamming in image, specifically: utilize height
This smothing filtering removes isolated noise point existing for concrete structural surface, while retaining the structure of crack area；
Step 3: obtaining crack candidate regions: the method for utilizing the self adapting morphology background removal based on image definition,
It is preliminary to remove the non-crack area of image, crack candidate region is obtained, specifically: using gray level image closed operation principle, and according to
The vision crack generating principle for the readability and people being imaged when every photograph taking, i.e., it is usual along the luminous intensity in crack
Secretly much than other background areas, and the length in crack is significantly larger than the principle of width, automatically generates background removal
Coefficient obtains image background, i.e., non-crack area, to obtain preliminary crack area by the method；
The method that the morphological background removes includes the following steps:
Step 3 one passes through morphological operation and obtains background image: using the vision crack generating principle of people, i.e., along splitting
Usually secretly much than other background areas, and the length in crack is significantly larger than the principle of width to the luminous intensity of seam, obtains back
Scape image；
Step 3 two automatically generates background removal coefficient: suitable back is automatically generated within the scope of 0 to 1 penalty coefficient
Scape removes coefficient；
In the step 3 two, the penalty coefficient is inversely proportional selected with clarity, and the clarity evaluation method uses
Tenen Grad clarity evaluation method, includes the following steps:
Step 321 determines gradient magnitude: according to the arithmetic square root of horizontal gradient and vertical gradient quadratic sum, determining
Gradient magnitude；
Step 3 two or two determines final clarity evaluation of estimate: according to preset threshold value, obtaining final clarity evaluation
Value；
Step 3 three obtains image background: utilizing gray level image closed operation principle, the gray value of image is stretched to 0 and is arrived
255, then image background is obtained by linear transformation function；
Step 4: crack area enhance: using gray scale cluster method, further enhance back as a result, then making
Final crack area is obtained with the binarization segmentation method of local auto-adaptive；
The method clustered using gray scale in the step 4 are as follows: using crack area darker in color in background area,
And otherness is little between crack area and clustering principle, the gray value and gray scale cluster centre to each pixel are logical
It crosses clustering function to be compared, obtains the high gray level image of energy value, further remove the non-crack area of image, enhancing, which obtains, to be split
Stitch candidate region；
Specifically comprise the following steps:
Step 4 one by one, the set of the gray value of reliable area: global binary segmentation is carried out to image, obtains reliable area
Then range seeks common ground with former gray level image, the gray value within the scope of reliable area is added in set；
Step 4 one or two obtains gray scale cluster centre: passing through the collection of the gray value of the reliable area in the candidate region of crack
The arithmetic mean of instantaneous value of conjunction obtains gray scale cluster centre；
Step 4 one or three, the judgement of image grayscale cluster value: led to by gray value to each pixel and gray scale cluster centre
It crosses clustering function to be compared, obtains the high gray level image of energy value；
The binarization segmentation method of the local auto-adaptive in the step 4 obtains final crack area and specifically wraps
Include following steps:
Step 421 obtains the inter-class variance that gray level is threshold value: passing through background, the gray average of target part and general
Rate obtains the inter-class variance that gray level is threshold value；
Step 4 two or two, search maximum between-cluster variance determine segmentation threshold: being 0 search for arriving maximum gray scale in gray level
Maximum between-cluster variance is found in space, determines segmentation threshold；
Step 4 two or three determines the corresponding segmentation threshold of each pixel value: according to local gray feature is counted, according to image
The gray average of middle window obtains local auto-adaptive threshold value；
Step 4 two or four, local auto-adaptive binary segmentation obtain final crack area: passing through segmentation threshold and adaptive thresholding
Overall situation and partial situation's threshold value is combined using the method for overall situation and partial situation's weighting segmentation, is retained the feature of overall situation and partial situation, reached by value
To comparatively ideal segmentation, final crack area is obtained；
Step 5: feature extraction: utilizing bianry image skeletonization method, extract the framework characteristic of crack area, according to splitting
The continuity of seam and the spatial structure characteristic of flatness optimize skeletonizing result；
Described utilizes bianry image skeletonization method, extracts the framework characteristic of crack area, specially utilizes Guo-
Hall Thinning8- neighborhood thinning algorithm generates the skeleton structure line in bianry image region；
It is described according to the continuity in crack and the spatial structure characteristic of flatness, optimize skeletonizing as a result, specifically: it is right
The hole as present on crack area or fracture after the micronization processes of region, and destroy has good adjacency and company originally
The fissured structure of the general character, fracture fragment carry out two-dimensional topology relationship analysis, carry out key point connection to key point, by neighbouring or
Crack fragments mosaicing with good connection is together；
Step 6: crack result is extracted.
In the Bridge Crack detection and a kind of preferred embodiment of feature extracting method provided by the invention based on image,
After the step 3 three further include: using the image background of acquisition described in dimensional Gaussian the disposal of gentle filter.
Compared with the existing technology, the Bridge Crack detection of the invention based on image has as follows with feature extracting method
The utility model has the advantages that
One, it is based on bridge surface image, detect crack area and extracts Expressive Features, enhances crack area, inhibits noise
With non-crack area, the crack of numerous and disorderly multiplicity can preferably be separated with background characteristics, energy rapidly extracting FRACTURE CHARACTERISTICS is quick
Fissured structure feature is rebuild, algorithm is simple, efficient, and tests test and show precision with higher.
Two, using Gaussian smoothing filter, gray level image closed operation principle, morphologic background removal algorithm, utilize gray scale
The method of cluster, the binarization segmentation method of local auto-adaptive obtain final crack area, utilize bianry image skeletonizing side
Method, extracts the framework characteristic of crack area and according to various methods such as the spatial structure characteristics of the continuity in crack and flatness,
Concrete structural surface is removed there are noise jamming existing for a large amount of isolated noise point and concrete, cement surface, is protected simultaneously
The structure of crack area has been stayed, and denoising result has been optimized, has made measurement result precision with higher.
Three, using Gaussian smoothing filter, removing concrete structural surface, there are the isolated noises such as a large amount of spot, stain
Point, while retaining the structure of crack area." TenenGrad " clarity evaluation method has been used, the tax of penalty coefficient α is obtained
Value, then by background removal algorithm, there is more precipitous variation characteristic, noiseproof feature is good, focus sensitivity height and reliability
High feature identical compared with the visual experience of human eye, further noise reduction；In method using gray scale cluster, in reliable area model
Gray value in enclosing is added to a simple screening for gathering that this process is fracture candidate region in fact, does not make directly
The noise jamming as existing for concrete, cement surface is mainly considered as crack area in region after using background removal,
This interference is weakened further by global binary segmentation, chooses candidate regions the most significant as far as possible as reliable crack area
Domain；The method combined using overall situation and partial situation's threshold value can weaken image irradiation unevenness and local blank area to a certain extent
The influence in domain improves anti-noise ability, improves segmentation effect.Final split is obtained using the binarization segmentation method of local auto-adaptive
Region is stitched, it is uneven and local empty can to weaken image irradiation to a certain extent for the method combined using overall situation and partial situation's threshold value
The influence of white region improves anti-noise ability, improves segmentation effect.Using bianry image skeletonization method, crack area is extracted
The method of framework characteristic can be stable generate the skeleton line of single pixel width, but during refinement, it is easy to by original
The influence of extracted region result, especially when, there are when hole and subbranch, this is generally deposited in crack area detection in region
Herein before carrying out thinning algorithm, the hole in region is being filled up using morphological dilations；Using according to the continuity in crack and
The method of the spatial structure characteristic of flatness, when meeting preferable propinquity or good connectivity, so that it may by the two
The crack fragments mosaicing at place locally can be approximately straight line connection joining method simplification together.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing, in which:
Fig. 1 is the knot of the Bridge Crack detection and one preferred embodiment of feature extracting method provided by the invention based on image
Structure schematic diagram；
Fig. 2 is that the morphological background of the Bridge Crack detection and feature extracting method shown in FIG. 1 based on image removes
The flow chart of one preferred embodiment of method；
Fig. 3 is the Bridge Crack detection shown in FIG. 1 based on image and the side of feature extracting method clustered using gray scale
The flow chart of one preferred embodiment of method；
Fig. 4 is the two-value of the Bridge Crack detection and the local auto-adaptive of feature extracting method shown in FIG. 1 based on image
Change the flow chart that dividing method obtains final one preferred embodiment of crack area.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that the described embodiments are merely a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts all other
Embodiment shall fall within the protection scope of the present invention.
Also referring to Fig. 1 to Fig. 4, wherein Fig. 1 is the Bridge Crack detection and feature provided by the invention based on image
The structural schematic diagram of one preferred embodiment of extracting method, Fig. 2 are that the Bridge Crack detection shown in FIG. 1 based on image is mentioned with feature
The flow chart for one preferred embodiment of method for taking the morphological background of method to remove, Fig. 3 are the bridges shown in FIG. 1 based on image
The flow chart of one preferred embodiment of method using gray scale cluster of Crack Detection and feature extracting method, Fig. 4 is shown in FIG. 1
The binarization segmentation method of Bridge Crack detection and the local auto-adaptive of feature extracting method based on image obtains final split
Stitch the flow chart of one preferred embodiment of region.
The Bridge Crack detection based on image includes the following steps: with feature extracting method
Step S1, bridge floor Image Acquisition；
Step S2: image preprocessing: using Gaussian smoothing filter, weakens the noise jamming in image；
Specifically: Gaussian smoothing filter is utilized, the isolated noises such as spot, stain existing for concrete structural surface are removed
Point, while retaining the structure of crack area.
Step S3, crack candidate regions are obtained: utilizing the method for the self adapting morphology background removal based on image definition
S30, it is preliminary to remove the non-crack area of image, crack candidate region is obtained, specifically: gray level image closed operation principle is utilized, and
The vision crack generating principle for the readability and people being imaged when according to every photograph taking, i.e., along the luminous intensity in crack
Usually secretly much than other background areas, and the length in crack is significantly larger than the principle of width, automatically generates background shifting
The coefficient removed obtains image background, i.e., non-crack area, to obtain preliminary crack area by the method；
The method S30 that the morphological background removes includes the following steps:
Step S31, pass through morphological operation and obtain background image: using the vision crack generating principle of people, i.e., along splitting
Usually secretly much than other background areas, and the length in crack is significantly larger than the principle of width to the luminous intensity of seam, obtains back
Scape image；
Step S32, it automatically generates background removal coefficient: automatically generating suitable background within the scope of 0 to 1 penalty coefficient
Remove coefficient；
The penalty coefficient be inversely proportional with clarity it is selected, the clarity evaluation method use Tenen Grad clarity
Evaluation method S320, includes the following steps:
Step S321, it determines gradient magnitude: according to the arithmetic square root of horizontal gradient and vertical gradient quadratic sum, determining ladder
Spend size；
Step S322, it determines final clarity evaluation of estimate: according to preset threshold value, obtaining final clarity evaluation
Value.
Step S33, it obtains image background: utilizing gray level image closed operation principle, the gray value of image is stretched to 0 and is arrived
255, then image background is obtained by linear transformation function；
Step S34, using the image background of acquisition described in dimensional Gaussian the disposal of gentle filter.
Step S4, crack area enhance: using gray scale cluster method S410, further enhance back as a result, right
Final crack area S420 is obtained using the binarization segmentation method of local auto-adaptive afterwards；
The method S410 clustered using gray scale are as follows: using crack area darker in color in background area, and crack area
Between otherness is little and clustering principle, clustering function is passed through to the gray value and gray scale cluster centre of each pixel
It is compared, obtains the high gray level image of energy value, further remove the non-crack area of image, enhancing obtains crack candidate regions
Domain；
Specifically comprise the following steps:
Step S411, the set of the gray value of reliable area: global binary segmentation is carried out to image, obtains reliable area model
It encloses, then seeks common ground with former gray level image I, the gray value within the scope of reliable area is added in set Is；
Step S412, it obtains gray scale cluster centre: passing through the set of the gray value of the reliable area in the candidate region of crack
Arithmetic mean of instantaneous value, obtain gray scale cluster centre；
Step S413, image grayscale cluster value judges: being passed through by the gray value to each pixel with gray scale cluster centre
Clustering function is compared, and obtains the high gray level image of energy value；
The binarization segmentation method of the local auto-adaptive obtains final crack area S420 and specifically comprises the following steps:
Step S421, the inter-class variance that gray level is threshold value is obtained: by background, the gray average of target part and general
Rate obtains the inter-class variance that gray level is threshold value；
Step S422, search maximum between-cluster variance determines segmentation threshold: empty in the search that gray level is 0 to maximum gray scale
Interior searching maximum between-cluster variance, determines segmentation threshold；
Step S423, the corresponding segmentation threshold of each pixel value is determined: according to local gray feature is counted, according in image
The gray average of window obtains local auto-adaptive threshold value；
Step S424, local auto-adaptive binary segmentation obtains final crack area: by segmentation threshold and adaptive threshold,
Using overall situation and partial situation weighting segmentation method, overall situation and partial situation's threshold value is combined, the feature of overall situation and partial situation is retained, reach compared with
Ideal segmentation, obtains final crack area.
Step S5, feature extraction: bianry image skeletonization method is utilized, the framework characteristic of crack area is extracted, according to splitting
The continuity of seam and the spatial structure characteristic of flatness optimize skeletonizing result；
Described utilizes bianry image skeletonization method, extracts the framework characteristic of crack area, specially utilizes Guo-
Hall Thinning8- neighborhood thinning algorithm generates the skeleton structure line in bianry image region.
It is described according to the continuity in crack and the spatial structure characteristic of flatness, optimize skeletonizing as a result, specifically: it is right
The hole as present on crack area or fracture after the micronization processes of region, and destroy has good adjacency and company originally
The fissured structure of the general character, fracture fragment carry out two-dimensional topology relationship analysis, carry out key point connection to key point, by neighbouring or
Crack fragments mosaicing with good connection is together.
Step S6: crack result is extracted.
Bridge Crack detection based on image of the invention has as follows with feature extracting method 100 and its measurement method
The utility model has the advantages that
One, it is based on bridge surface image, detect crack area and extracts Expressive Features, enhances crack area, inhibits noise
With non-crack area, the crack of numerous and disorderly multiplicity can preferably be separated with background characteristics, energy rapidly extracting FRACTURE CHARACTERISTICS is quick
Fissured structure feature is rebuild, algorithm is simple, efficient, and tests test and show precision with higher.
Two, using Gaussian smoothing filter, gray level image closed operation principle, morphologic background removal algorithm S30, utilize ash
The binarization segmentation method of the method S410, local auto-adaptive that spend cluster obtain final crack area S420, utilize binary map
As skeletonization method, the framework characteristic of crack area is extracted and according to the continuity in crack and spatial structure characteristic of flatness etc.
Various methods, there are noises existing for a large amount of isolated noise point and concrete, cement surface to do for removal concrete structural surface
It disturbs, while retaining the structure of crack area, and optimize to denoising result, make measurement result precision with higher.
Three, using Gaussian smoothing filter, removing concrete structural surface, there are the isolated noises such as a large amount of spot, stain
Point, while retaining the structure of crack area." TenenGrad " clarity evaluation method S320 has been used, penalty coefficient α is obtained
Assignment, then by background removal algorithm S30, there is more precipitous variation characteristic, noiseproof feature is good, focus sensitivity height and
High reliablity feature identical compared with the visual experience of human eye, further noise reduction；In method S410 using gray scale cluster,
Gray value within the scope of reliable area is added to a simple screening for gathering that this process is fracture candidate region in fact,
Region after not using background removal directly is as crack area, mainly in view of existing due to concrete, cement surface
Noise jamming, weaken this interference further by global binary segmentation, choose candidate regions the most significant as far as possible and make
For reliable crack area；The method combined using overall situation and partial situation's threshold value can weaken image irradiation unevenness to a certain extent
With the influence of local white space, anti-noise ability is improved, improves segmentation effect.Utilize the binarization segmentation method of local auto-adaptive
Final crack area S420 is obtained, the method combined using overall situation and partial situation's threshold value can weaken image to a certain extent
The influence of uneven illumination and local white space, improves anti-noise ability, improves segmentation effect.Utilize bianry image skeletonizing side
Method, the method for extracting the framework characteristic of crack area can be stable generate the skeleton line of single pixel width, but in refinement
In the process, it is easy to be influenced by former extracted region result, especially when, there are when hole and subbranch, this is in crack in region
It is generally existing in region detection, herein before carrying out thinning algorithm, the hole in region is filled up using morphological dilations；Using root
According to the method for the spatial structure characteristic of the continuity and flatness in crack, meeting preferable propinquity or good connectivity
When, so that it may it locally can be approximately that straight line connects joining method simplification together by the crack fragments mosaicing where the two
It connects.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (2)
1. a kind of Bridge Crack detection and feature extracting method based on image, which comprises the steps of:
Step 1: bridge floor Image Acquisition；
Step 2: image preprocessing: using Gaussian smoothing filter, weakens the noise jamming in image, specifically: it is flat using Gauss
Sliding filtering, removes isolated noise point existing for concrete structural surface, while retaining the structure of crack area；
Step 3: obtaining crack candidate regions: using the method for the self adapting morphology background removal based on image definition, tentatively
The non-crack area of image is removed, crack candidate region is obtained, specifically: gray level image closed operation principle is utilized, and according to every
The vision crack generating principle for the readability and people being imaged when photograph taking, i.e., along crack luminous intensity usually than
Secretly much, and the length in crack is significantly larger than the principle of width, automatically generates the coefficient of background removal for other background areas,
Image background, i.e., non-crack area, to obtain preliminary crack area are obtained by the method；
The method that the morphological background removes includes the following steps:
Step 3 one obtains background image by morphological operation: using the vision crack generating principle of people, i.e., along crack
Usually secretly much than other background areas, and the length in crack is significantly larger than the principle of width to luminous intensity, obtains Background
Picture；
Step 3 two automatically generates background removal coefficient: automatically generating suitable background within the scope of 0 to 1 penalty coefficient and moves
Except coefficient；
In the step 3 two, the penalty coefficient be inversely proportional with clarity it is selected, the clarity evaluation method use Tenen
Grad clarity evaluation method, includes the following steps:
Step 321 determines gradient magnitude: according to the arithmetic square root of horizontal gradient and vertical gradient quadratic sum, determining gradient
Size；
Step 3 two or two determines final clarity evaluation of estimate: according to preset threshold value, obtaining final clarity evaluation of estimate；
Step 3 three obtains image background: gray level image closed operation principle utilized, the gray value of image is stretched to 0 to 255,
Image background is obtained by linear transformation function again；
Step 4: crack area enhances: the method clustered using gray scale, then further enhance back as a result, use office
The adaptive binarization segmentation method in portion obtains final crack area；
The method clustered using gray scale in the step 4 are as follows: using crack area darker in color in background area, and split
Otherness is little between seam region and clustering principle, gray value and gray scale cluster centre to each pixel pass through poly-
Class function is compared, and obtains the high gray level image of energy value, further removes the non-crack area of image, and enhancing obtains crack and waits
Favored area；
Specifically comprise the following steps:
Step 4 one by one, the set of the gray value of reliable area: global binary segmentation is carried out to image, obtains reliable area model
It encloses, then seeks common ground with former gray level image, the gray value within the scope of reliable area is added in set；
Step 4 one or two obtains gray scale cluster centre: by the set of the gray value of the reliable area in the candidate region of crack
Arithmetic mean of instantaneous value obtains gray scale cluster centre；
Step 4 one or three, the judgement of image grayscale cluster value: by gray value to each pixel and gray scale cluster centre by gathering
Class function is compared, and obtains the high gray level image of energy value；
The binarization segmentation method of the local auto-adaptive in the step 4 obtain final crack area specifically include as
Lower step:
Step 421 obtains the inter-class variance that gray level is threshold value: by background, the gray average and probability of target part,
Obtain the inter-class variance that gray level is threshold value；
Step 4 two or two, search maximum between-cluster variance determine segmentation threshold: being 0 search space for arriving maximum gray scale in gray level
Interior searching maximum between-cluster variance, determines segmentation threshold；
Step 4 two or three determines the corresponding segmentation threshold of each pixel value: according to local gray feature is counted, according to window in image
The gray average of mouth obtains local auto-adaptive threshold value；
Step 4 two or four, local auto-adaptive binary segmentation obtain final crack area: by segmentation threshold and adaptive threshold, making
With the method for overall situation and partial situation's weighting segmentation, overall situation and partial situation's threshold value is combined, the feature of overall situation and partial situation is retained, reached and relatively manage
The segmentation thought obtains final crack area；
Step 5: feature extraction: utilizing bianry image skeletonization method, the framework characteristic of crack area is extracted, according to crack
The spatial structure characteristic of continuity and flatness optimizes skeletonizing result；
Described utilizes bianry image skeletonization method, extracts the framework characteristic of crack area, specially utilizes Guo-Hall
Thinning8- neighborhood thinning algorithm generates the skeleton structure line in bianry image region；
It is described according to the continuity in crack and the spatial structure characteristic of flatness, optimize skeletonizing as a result, specifically: to region
The hole as present on crack area or fracture after micronization processes, and destroy has good adjacency and connectivity originally
Fissured structure, fracture fragment carries out two-dimensional topology relationship analysis, carries out key point connection to key point, will be neighbouring or have
The crack fragments mosaicing of good connection is together；
Step 6: crack result is extracted.
2. the Bridge Crack detection and feature extracting method according to claim 1 based on image, which is characterized in that described
After step 3 three further include: using the image background of acquisition described in dimensional Gaussian the disposal of gentle filter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610269077.8A CN105975972B (en) | 2016-04-27 | 2016-04-27 | Bridge Crack detection and feature extracting method based on image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610269077.8A CN105975972B (en) | 2016-04-27 | 2016-04-27 | Bridge Crack detection and feature extracting method based on image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105975972A CN105975972A (en) | 2016-09-28 |
| CN105975972B true CN105975972B (en) | 2019-05-07 |
Family
ID=56993306
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610269077.8A Active CN105975972B (en) | 2016-04-27 | 2016-04-27 | Bridge Crack detection and feature extracting method based on image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105975972B (en) |
Families Citing this family (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106645171A (en) * | 2016-12-22 | 2017-05-10 | 江苏工程职业技术学院 | Method for detecting concrete cracks by using wall-climbing robot |
| CN107123109A (en) * | 2017-01-13 | 2017-09-01 | 陕西师范大学 | A kind of window sliding algorithm detected for Bridge Crack |
| CN107169953B (en) * | 2017-04-07 | 2020-02-21 | 西安电子科技大学 | Bridge concrete surface crack detection method based on HOG characteristics |
| CN107463934A (en) * | 2017-08-08 | 2017-12-12 | 浙江国自机器人技术有限公司 | A kind of tunnel slot detection method and device |
| CN108106801B (en) * | 2017-11-15 | 2021-11-05 | 温州市交通工程试验检测有限公司 | Bridge and tunnel disease non-contact detection system and detection method |
| CN107967685A (en) * | 2017-12-11 | 2018-04-27 | 中交第二公路勘察设计研究院有限公司 | A kind of bridge pier and tower crack harmless quantitative detection method based on unmanned aerial vehicle remote sensing |
| CN108805855B (en) * | 2018-01-09 | 2021-09-24 | 湖南科技大学 | Method for quickly identifying concrete cracks |
| CN108801151A (en) * | 2018-05-29 | 2018-11-13 | 云南富龙高速公路建设指挥部 | Pavement crack detection device based on smart mobile phone and detection method |
| CN108764345B (en) * | 2018-05-30 | 2022-04-26 | 河海大学常州校区 | Underwater dam crack detection method based on local and global clustering |
| CN108921847B (en) * | 2018-08-08 | 2022-06-24 | 长沙理工大学 | Bridge bottom detection method based on machine vision |
| CN109410205B (en) * | 2018-11-01 | 2021-05-04 | 西安建筑科技大学 | Crack extraction method under complex pavement background |
| CN109374651B (en) * | 2018-11-02 | 2021-04-27 | 深圳大学 | Large bridge detection method and detection device based on tethered unmanned aerial vehicle |
| CN109459443A (en) * | 2018-11-08 | 2019-03-12 | 青海民族大学 | A kind of detection of extremely frigid zones Bridge Crack and analytic method based on machine vision |
| CN109493339B (en) * | 2018-11-20 | 2022-02-18 | 北京嘉恒中自图像技术有限公司 | Endoscope imaging-based method for detecting defects of pores on inner surface of casting |
| CN109580657B (en) * | 2019-01-23 | 2021-08-20 | 郑州工程技术学院 | Crack detection method in bridge quality detection |
| CN110021017B (en) * | 2019-04-02 | 2022-06-17 | 南通大学 | Method for extracting central line of welding line |
| CN109919942B (en) * | 2019-04-04 | 2020-01-14 | 哈尔滨工业大学 | Bridge crack intelligent detection method based on high-precision noise reduction theory |
| CN110189290A (en) * | 2019-04-08 | 2019-08-30 | 广东工业大学 | Metal surface fine defects detection method and device based on deep learning |
| CN110175987B (en) * | 2019-04-23 | 2022-10-04 | 江西科技学院 | Bridge crack detection method and system |
| CN110223283B (en) * | 2019-06-10 | 2022-04-19 | 西安汇智信息科技有限公司 | Bridge crack detection method |
| CN110853034A (en) * | 2020-01-15 | 2020-02-28 | 清华四川能源互联网研究院 | Crack detection method, crack detection device, electronic equipment and computer-readable storage medium |
| CN112581433B (en) * | 2020-12-07 | 2022-10-11 | 上海大学 | Geometric information extraction method for thermal barrier coating cracks |
| CN112801972A (en) * | 2021-01-25 | 2021-05-14 | 武汉理工大学 | Bridge defect detection method, device, system and storage medium |
| CN113240649B (en) * | 2021-05-19 | 2023-12-19 | 姚坚超 | Building component detection method and system based on image analysis |
| CN113592861B (en) * | 2021-09-27 | 2021-12-24 | 江苏中云筑智慧运维研究院有限公司 | Bridge crack detection method based on dynamic threshold |
| CN114418957B (en) * | 2021-12-24 | 2022-11-18 | 广州大学 | Global and local binary pattern image crack segmentation method based on robot vision |
| CN115294080B (en) * | 2022-08-15 | 2023-09-08 | 山东大学 | Automatic slotting robot for highway cracks and working method and application thereof |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101915764A (en) * | 2010-08-10 | 2010-12-15 | 武汉武大卓越科技有限责任公司 | Road surface crack detection method based on dynamic programming |
| CN103048329A (en) * | 2012-12-11 | 2013-04-17 | 北京恒达锦程图像技术有限公司 | Pavement crack detecting method based on active contour model |
| CN103440657A (en) * | 2013-08-27 | 2013-12-11 | 武汉大学 | Method for online screening cracks of road |
-
2016
- 2016-04-27 CN CN201610269077.8A patent/CN105975972B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101915764A (en) * | 2010-08-10 | 2010-12-15 | 武汉武大卓越科技有限责任公司 | Road surface crack detection method based on dynamic programming |
| CN103048329A (en) * | 2012-12-11 | 2013-04-17 | 北京恒达锦程图像技术有限公司 | Pavement crack detecting method based on active contour model |
| CN103440657A (en) * | 2013-08-27 | 2013-12-11 | 武汉大学 | Method for online screening cracks of road |
Non-Patent Citations (2)
| Title |
|---|
| "Fast Fully Parallel Thinning Algorithms";ZICHENG GUO etc，;《Cvgip Image Understanding》;19921231;第55卷(第3期);摘要、第2节、图2 |
| "桥梁裂缝检测中图像识别方法";齐超;《万方数据知识服务平台》;20150629;摘要、第2.2节、3.4-3.6节、5.1.4-5.1.5节 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105975972A (en) | 2016-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105975972B (en) | Bridge Crack detection and feature extracting method based on image | |
| Wang et al. | Pavement crack image acquisition methods and crack extraction algorithms: A review | |
| CN104008553B (en) | Crack detection method with image gradient information and watershed method conflated | |
| CN104504388B (en) | A kind of pavement crack identification and feature extraction algorithm and system | |
| CN101706445B (en) | Image processing mthod for beef marbling grain grade scoring | |
| CN104392224B (en) | A kind of highway pavement crack detecting method | |
| Rau et al. | Bridge crack detection using multi-rotary UAV and object-base image analysis | |
| CN108009591A (en) | A kind of contact network key component identification method based on deep learning | |
| CN105678792A (en) | Method and system for extracting body profile | |
| CN110009618A (en) | A kind of Axle Surface quality determining method and device | |
| CN105512666A (en) | River garbage identification method based on videos | |
| CN108830877A (en) | The stereoscopic micro-image quantitative description of rock core | |
| CN107240079A (en) | A kind of road surface crack detection method based on image procossing | |
| CN109191432A (en) | The remote sensing images cloud detection method of optic of filtering multi-resolution decomposition is converted based on domain | |
| CN103295199A (en) | Intelligent repair assistance system for cracks of ancient wall murals | |
| CN110415241A (en) | A kind of surface of concrete structure quality determining method based on computer vision | |
| CN103048329A (en) | Pavement crack detecting method based on active contour model | |
| CN106096603A (en) | A kind of dynamic flame detection method merging multiple features and device | |
| CN109685788A (en) | A kind of flooring defect image automatic testing method based on morphological feature | |
| CN106023226A (en) | Crack automatic detection method based on three-dimensional virtual pavement | |
| CN104537651A (en) | Proportion detecting algorithm and system for cracks in road surface image | |
| CN110473187A (en) | A kind of line scanning three-dimensional pavement crack extract method of object-oriented | |
| Stricker et al. | Road surface segmentation-pixel-perfect distress and object detection for road assessment | |
| CN105678735A (en) | Target salience detection method for fog images | |
| CN110009622A (en) | A kind of display panel open defect detection network and its defect inspection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |