CN104504702A - Cement notch pavement crack recognition method based on square lattice searching method - Google Patents
Cement notch pavement crack recognition method based on square lattice searching method Download PDFInfo
- Publication number
- CN104504702A CN104504702A CN201410807044.5A CN201410807044A CN104504702A CN 104504702 A CN104504702 A CN 104504702A CN 201410807044 A CN201410807044 A CN 201410807044A CN 104504702 A CN104504702 A CN 104504702A
- Authority
- CN
- China
- Prior art keywords
- image
- crack
- pavement
- cutting
- grid
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a cement notch pavement crack recognition method based on a square lattice searching method. According to the method, firstly, a pavement image containing a notch image is firstly obtained, and in addition, the pavement image is preprocessed; the pavement image is subjected to edge detection, possible cracks are separated out from a pavement background image; the possible crack image is subjected to square lattice searching, and then, partial horizontal notches are removed according to the size of a square lattice center point communication region; the noise is eliminated; burrs are removed. The method has the advantages that the problem of judging notches into the cracks to be processed by mistake in the damaged pavement image recognition process is effectively solved, in addition, the calculation quantity is small, and the work efficiency is improved.
Description
Technical field
The invention belongs to technical field of image processing, be specifically related to the cement cutting pavement crack recognition methods based on grid search procedure.
Background technology
Late 1980s, China has built up Article 1 highway---and highway is praised in Shanghai.This indicates that the development of China's highway enters fast-developing period.Especially in recent years, the highway of China was with the speed increase of several thousand kilometers every year.Present expressway construction becomes promote domestic demands, promotes one of key factor of national economy fast development, is subject to the great attention of governments at all levels.
Be planned for the year two thousand twenty according to national Eleventh Five-Year Development Plan, China's highway mileage open to traffic will reach 100,000 kilometers.Along with the development that highway in China is built, especially high-grade highway and highway network is day by day perfect, and highway maintenance management has become the vital task of highway in China construction field.Comprehensively deep understanding Pavement Condition is the important foundation that science formulates maintenance plan.Use frequently due to the destruction of weather extremes, transition and highway is aging etc. that reason causes road surface occurs crack, crack is the old model of most disease.What the appearance of pavement crack was serious have impact on pavement quality, have impact on the serviceable life of traffic safety and highway.Therefore Timeliness coverage crack repairing, avoids the having a strong impact on of causing that further develop due to pavement crack to have important practical significance.
Along with the development of computer technology, Digital image technology has been applied to each more and more widely and has produced and sciemtifec and technical sphere.Adopt and carry out road surface breakage detection based on image recognition technology, damage is carried out to the pavement image gathered and identifies, calculate pavement damage ratio, for road condition analyzing provides data foundation.Cement pavement is due to the needs of pavement skid resistance effect, and when pavement construction, all very important person is the road surface cutting increasing a series of almost horizontal.The existence of this cutting, bring many difficulties to the cement pavement crack identification based on image, the gray-scale value that road surface cutting shows on image is darker, very approximate with crack, therefore during crack identification, often cutting and crack are identified simultaneously, cause many mistakes to identify interference.
In the prior art, have a kind of method that cement pavement image groove based on two-dimensional Fourier transform is removed, can reach the effect removing cutting equally, but this algorithm also also exists some problems at present, as calculated amount is large, spent time is long.
Summary of the invention
The defect existed for above-mentioned prior art or deficiency, the invention provides a kind of cement cutting pavement crack recognition methods based on grid search procedure, the method efficiently solves above-mentioned in the identifying of damaged road surface image, cutting is judged into by accident the problem of crack treatment, and calculated amount is little, improve work efficiency.
The technical scheme realizing the object of the invention is:
Based on a cement cutting pavement crack recognition methods for grid search procedure, comprise the steps:
(1) obtain the pavement image comprising cutting image, and road pavement image carries out pre-service;
(2) road pavement image carries out rim detection, is split in possible crack from the background image of road surface;
(3) grid search is carried out to possible crack pattern picture,
(4) according to the size with grid central point connected region, part of horizontal cutting is removed;
(5) noise is removed;
(6) burr is removed.
Mathematical Morphology Method is selected in described pavement image pre-service, and the method can filtering noise effectively, the detailed information in image can also be retained preferably.
Described edge detection operator adopts Canny operator, utilizes the non-maxima suppression feature of Canny operator, and setting threshold value retains the strong edge in image border.
Described deburring adopts sciagraphy to remove the burr be connected with crack.
Advantage of the present invention is: the method efficiently solves above-mentioned in the identifying of damaged road surface image, cutting is judged into by accident the problem of crack treatment, and calculated amount is little, improves work efficiency.
Accompanying drawing explanation
Fig. 1 is algorithm frame process flow diagram of the present invention.
Fig. 2 is crack image panes search schematic diagram.
Fig. 3 is several situation schematic diagram of grid search.
Fig. 4 is that grid searching threshold selects schematic diagram.
Fig. 5 is for removing noise schematic diagram.
Fig. 6 is for removing burr sciagraphy schematic diagram.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described in detail:
A kind of cement cutting pavement crack recognition methods based on grid search procedure is:
Obtain the pavement image comprising cutting image, and road pavement image carries out pre-service, comprise the pavement image of cutting image by video recording equipment shooting, the cutting image approximate level in pavement image and run through the straight line of image, these cuttings equidistantly arrange, and gray scale is substantially identical.Greasy dirt, foreign material, optical system distortion, relative motion, air-flow cause in shooting process image blurring, noise pollution etc. in image transmitting process is there is in pavement crack image; The task of Image semantic classification is by the noise remove of image, strengthens target energy in the picture, namely highlights crack principal character in the picture, for next step crack identification does data encasement; In pavement image, crack gray-scale value is lower, and road surface background gray levels is relatively high, and this gray feature can be utilized during image enhaucament to strengthen.
Mathematical Morphology Method can filtering noise effectively, the detailed information in image can also be retained preferably.The basic operation of morphological image process is dilation and erosion, after again original image being done and is expanded, increase the bright spot in image, bright spot in image is given prominence to, and do not comprise crack information in background image, carry out corrosion again and obtain closed operation result, the even part of uneven illumination in the corresponding former figure in bright wisp region of formation, and the spot containing interference crack identification pollutes.By subtracting each other with original image, obtain the pavement crack image removing background.
Road pavement image carries out rim detection, object is split from the background image of road surface in possible crack, crack and cutting are all the gray scale catastrophe points in image, when carrying out rim detection with edge detection operator, the edge of crack and cutting can be detected simultaneously, the edge detection operator that the present invention adopts is Canny operator, and according to the present embodiment, high threshold and the Low threshold of choosing operator are as follows:
Threshold value comprises high threshold and Low threshold, and when Canny operator high threshold is greater than 0.5, main crack is not detected, and there will be many noises when high threshold is too small.Low threshold is less.The image detail detected is more, and it is also more complete that crack retains, but remain again the details in too many non-crack when Low threshold is too small.By contrast, selected Canny operator Low threshold is 0.2, and high threshold is 0.5, carries out Image Edge-Detection.
The crack location value detected is set to 0, is black, other parts are set to 1, are white, obtain crack extract result images.
Road pavement image carries out grid search, in order to remove cutting and noise; Grid search procedure is adopted to carry out traversal search to image, to extract possible crack information.
As shown in Figure 2, the image after edge treated comprises cutting, crack and noise three kinds of information.When carrying out grid traversal search to image, the side's of making center of a lattice traversal extracts the every bit of image, according to the difference extracting image, set a suitable threshold value, and the size of in grid and grid center connected region is judged, when being greater than this threshold value, image being retained, when being less than this threshold value, image being removed.Following four kinds of situations can be divided into, as shown in Figure 3:
As shown in (a), there is two cuttings and a crack in grid, be all connected with grid central point, and connected region is greater than default threshold value, therefore retain;
As shown in (b), only there is a cutting in grid, although be connected with grid central point, connected region is less than default threshold value, therefore removes;
As shown in (c), there is a cutting and a noise in grid, be all connected with grid central point, and connected region is greater than default threshold value, therefore is mistaken for crack and is retained;
As shown in (d), there is two cuttings and a noise in grid, only have noise to be connected with grid central point, so logical region is less than default threshold value, therefore remove.
When carrying out grid search, cutting and noise is removed and the object that retains crack in order to reach, need a selected suitable threshold value in grid and the size of central point linking area judge, with this threshold value for foundation, the image in grid spaces is accepted or rejected.The selection of threshold size is specific as follows:
As shown in Figure 3, when carrying out grid search to image, there will be (a) and (b), (c), (d) four kinds of main conditions, two cuttings and a crack is there is in (a) medium square, the connected region be connected with grid central point is necessarily greater than the grid length of side w of 2 times, if retain this image, selected threshold value e is less than or equal to 2w; B only there is a cutting in () medium square, the connected region be connected with grid central point equals grid length of side w, if remove this image, selected threshold value e is greater than w; C there is a cutting and a noise in () medium square, the connected region be connected with grid central point is greater than grid length of side w, but due to the uncertainty of noise, can not choose suitable threshold value and this image is carried out right-on choice; D there is two cuttings and a noise in () medium square, equally due to the uncertainty of noise, the connected region be connected with grid central point cannot compare with w.
In sum, preliminary definite threshold e=2w, for (a) and (b) two kinds of situations, this threshold value can correct accepting or rejecting detected image.For (c) and (d) two kinds of situations, when the connected region be connected with grid central point is less than 2w, remove this noise image; When the connected region be connected with grid central point is more than or equal to 2w, retain this noise image.The noise image remained can only be removed by subsequent operation.
When carrying out grid search, the length of side w of selected grid directly affects the accuracy of image zooming-out, correctness and operation efficiency, and the concrete selection mode of its length of side w value is as follows:
In the process determining grid length of side w, distance d between two cuttings can be chosen as with reference to value, as shown in Figure 4,
As w<d, utilize square method to carry out traversal search to image, effectively can remove the road surface cutting in image shown in (a).When grid is searched for the crack (b) between two cuttings, in grid, only there is crack information, cause the value in cutting (a) and the crack (b) be connected with grid center very nearly the same, can not effectively identify cutting and crack.
As w>d, as shown in (c), in grid, comprise many cuttings and crack, although part cutting can be removed, but cross due to the length of side image that conference makes to extract and comprise redundant information, cause and extract image section distortion, inconvenient subsequent treatment.
In sum, the grid length of side w for grid search procedure can not be too small, can not be excessive.In the present invention, choosing the grid length of side is distance d between two cuttings of 1.5 times, i.e. w=1.5d, when now traversal search being carried out to image, to extract image accuracy the highest, and be conducive to subsequent treatment.
Remove noise, when carrying out grid search to image, have some noise images and be mistaken for cutting and crack and retained, the object of this step will reject these noises exactly, retains the trunk portion in crack.
As shown in Figure 5, pavement image after grid search comprises major fracture and misjudged noise two kinds of information, according to the difference of two kinds of image geometry parameters and feature, area, girth can be selected, external convex polygon carries out image denoising, select girth characteristic to carry out denoising to image in the present invention.
Carry out analysis to the image after grid search can obtain, road base slot image is interconnected, and remaining noise image is then mutually isolated, and the girth of known major fracture image is far longer than the girth of noise image.Grid length of side w when searching for according to grid and the length of noise thereof, selectable threshold is 10w.The girth of main slot image will certainly be greater than threshold value 10w and be retained; And noise image all can be less than threshold value 10w and be removed.
Remove burr, adopt sciagraphy to remove the burr be connected with crack, concrete grammar is as described below:
As shown in Figure 6, for the image after above-mentioned denoising sets up coordinate system x-y;
Endokinetic fissure image m point bottom starts, and carries out rim detection by arrow direction fracture image;
After carrying out rim detection to whole image, all have two y values at identical x coordinate place, its difference is the width in crack.But, due to the existence of cutting, identical x coordinate place, two y values that numerical value difference is larger can be gone out, be designated as left some L (t) under this x coordinate and right R (t), similarly, the left point of cutting is followed successively by L (t+1), L (t+2), L (t+3) etc. from below to up, and corresponding right point is followed successively by R (t+1), R (t+2), R (t+3) etc.;
In order to eliminate cutting, can choose a predetermined threshold value, the size of this threshold value is greater than the breadth extreme in crack and is less than the length that there is cutting.When under identical x coordinate, the difference of two y values is less than predetermined threshold, retain two y values; When under identical x coordinate, the difference of two y values is greater than predetermined threshold, namely the left point of cutting place and the difference of right point are greater than this predetermined threshold, then get its mean value:
K(t)=[L(t)+R(t)]/2, K(t+1)=[L(t+1)+R(t+1)]/2,…
Through above process of removing image burr step, the crack information on cutting road surface will be extracted accurately and quickly.Can there is the node of single pixel in the crack extracted, little on the impact of major cracks image information in cutting position.
Claims (4)
1., based on a cement cutting pavement crack recognition methods for grid search procedure, it is characterized in that: comprise the steps:
(1) obtain the pavement image comprising cutting image, and road pavement image carries out pre-service;
(2) road pavement image carries out rim detection, is split in possible crack from the background image of road surface;
(3) grid search is carried out to possible crack pattern picture,
(4) according to the size with grid central point connected region, part of horizontal cutting is removed;
(5) noise is removed;
(6) burr is removed.
2. method according to claim 1, is characterized in that: Mathematical Morphology Method is selected in described pavement image pre-service.
3. method according to claim 1, is characterized in that: described edge detection operator adopts Canny operator, utilizes the non-maxima suppression feature of Canny operator, and setting threshold value retains the strong edge in image border.
4. method according to claim 1, is characterized in that: described deburring adopts sciagraphy to remove the burr be connected with crack.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410807044.5A CN104504702A (en) | 2014-12-23 | 2014-12-23 | Cement notch pavement crack recognition method based on square lattice searching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410807044.5A CN104504702A (en) | 2014-12-23 | 2014-12-23 | Cement notch pavement crack recognition method based on square lattice searching method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104504702A true CN104504702A (en) | 2015-04-08 |
Family
ID=52946096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410807044.5A Pending CN104504702A (en) | 2014-12-23 | 2014-12-23 | Cement notch pavement crack recognition method based on square lattice searching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104504702A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
| CN106651872A (en) * | 2016-11-23 | 2017-05-10 | 北京理工大学 | Prewitt operator-based pavement crack recognition method and system |
| CN106934829A (en) * | 2017-02-14 | 2017-07-07 | 中铁大桥科学研究院有限公司 | The detection method and system of a kind of surface crack |
-
2014
- 2014-12-23 CN CN201410807044.5A patent/CN104504702A/en active Pending
Non-Patent Citations (2)
| Title |
|---|
| 英红 等: "基于消除刻槽的水泥路面新生裂缝识别方法", 《同济大学学报（自然科学版）》 * |
| 荣婧 等: "基于图像的水泥刻槽路面裂缝识别方法", 《公路交通科技》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
| CN105469094B (en) * | 2015-11-28 | 2019-02-26 | 重庆交通大学 | A kind of edge vectors line drawing method of road surface bianry image |
| CN106651872A (en) * | 2016-11-23 | 2017-05-10 | 北京理工大学 | Prewitt operator-based pavement crack recognition method and system |
| CN106651872B (en) * | 2016-11-23 | 2020-09-15 | 北京理工大学 | Pavement crack identification method and system based on Prewitt operator |
| CN106934829A (en) * | 2017-02-14 | 2017-07-07 | 中铁大桥科学研究院有限公司 | The detection method and system of a kind of surface crack |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018107939A1 (en) | Edge completeness-based optimal identification method for image segmentation | |
| CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
| CN110210475B (en) | License plate character image segmentation method based on non-binarization and edge detection | |
| CN104021574A (en) | Method for automatically identifying pavement diseases | |
| CN103258201B (en) | A kind of form lines extracting method of amalgamation of global and local message | |
| CN102999886B (en) | Image Edge Detector and scale grating grid precision detection system | |
| CN104036262B (en) | A kind of method and system of LPR car plates screening identification | |
| CN104504388A (en) | Pavement crack identification and feature extraction algorithm and system | |
| CN103048329A (en) | Pavement crack detecting method based on active contour model | |
| CN104732220A (en) | Specific color human body detection method oriented to surveillance videos | |
| CN103136528A (en) | Double-edge detection based vehicle license plate identification method | |
| CN103295013A (en) | Pared area based single-image shadow detection method | |
| CN110969620A (en) | Method and device for detecting magnetic shoe ripple defects | |
| CN103942816A (en) | Method for identifying number and size of wood veneer knots based on image processing | |
| CN113240623A (en) | Pavement disease detection method and device | |
| CN103440657A (en) | Method for online screening cracks of road | |
| CN112085699B (en) | Pavement crack extraction method based on two-dimensional image | |
| CN112184725B (en) | Method for extracting center of structured light bar of asphalt pavement image | |
| CN108053412A (en) | The remote sensing image optimum segmentation result for merging more quantitative assessing index determines method | |
| CN104504702A (en) | Cement notch pavement crack recognition method based on square lattice searching method | |
| CN109543498A (en) | A kind of method for detecting lane lines based on multitask network | |
| CN113066041A (en) | Pavement crack detection method based on stack sparse self-coding deep learning | |
| CN114972757B (en) | Tunnel water leakage area identification method and system | |
| CN113450461B (en) | Soil-discharging-warehouse geotechnical distribution cloud extraction method | |
| CN116452826A (en) | Coal gangue contour estimation method based on machine vision under shielding condition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150408 |
|
| WD01 | Invention patent application deemed withdrawn after publication |