CN110487497A - A kind of Bridge Crack recognition methods based on recursive search - Google Patents

A kind of Bridge Crack recognition methods based on recursive search Download PDF

Info

Publication number
CN110487497A
CN110487497A CN201910673031.6A CN201910673031A CN110487497A CN 110487497 A CN110487497 A CN 110487497A CN 201910673031 A CN201910673031 A CN 201910673031A CN 110487497 A CN110487497 A CN 110487497A
Authority
CN
China
Prior art keywords
crack
segment
segments
suspicious
image
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910673031.6A
Other languages
Chinese (zh)
Other versions
CN110487497B (en
Inventor
苏成悦
刘信宏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN201910673031.6A priority Critical patent/CN110487497B/en
Publication of CN110487497A publication Critical patent/CN110487497A/en
Application granted granted Critical
Publication of CN110487497B publication Critical patent/CN110487497B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M5/00—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings
    • G01M5/0008—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings of bridges
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M5/00—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings
    • G01M5/0033—Investigating the elasticity of structures, e.g. deflection of bridges or air-craft wings by determining damage, crack or wear
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Signal Processing (AREA)
  • Quality & Reliability (AREA)
  • Geometry (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Image Analysis (AREA)

Abstract

The present invention proposes a kind of Bridge Crack recognition methods based on recursive search, comprising the following steps: crack image to be identified is converted to crack grayscale image G, Threshold segmentation is carried out to the crack image using threshold split plot design, obtains segmented image T；Feature Selection is carried out to the segmented image T, obtains suspicious crack set of segments；Parameter initialization is seriatim carried out to the suspicious crack set of segments；According to the ordinate value of the aft terminal of the suspicious crack set of segments, suspicious crack set of segments is resequenced according to descending, obtains the crack set of segments P for completing sequence；It scans for identifying by the crack identification algorithm fracture set of segments P based on recursive search, output obtains completing recursive detected crack data；The crack data detected are cleaned, and are exported as recognition result.The present invention can effectively eliminate influence caused by the identification of interference characteristic fracture.

Description

A kind of Bridge Crack recognition methods based on recursive search
Technical field
The present invention relates to technical field of computer vision, more particularly, to a kind of Bridge Crack based on recursive search Recognition methods.
Background technique
Bridge Crack identification technology is one and shoots object by camera, obtains image information, then carry out image analysis, from And obtain the engineering discipline of the information such as pose and the size of object.For measuring object, being identified and handled using computer can It is not influenced by artifact, precision is not limited by the objects of reference precision such as surveyors' staff, and compared to traditional artificial scale Measurement has the advantages such as quick, accurate, non-contact and low in cost, can greatly reduce cost of labor, effectively improve production Efficiency.
It is general using detection target when being identified to Bridge Crack, it then carries out carrying out non-cpntact measurement to target. When computer carries out detection target to Bridge Crack image, it is often used image segmentation processing method, in image dividing processing process In the automatic threshold segmentations methods such as fixed threshold split plot design, edge split plot design, maximum between-cluster variance can be used.However, fixed threshold Value split plot design, which has being illuminated by the light variation, to be influenced serious, and edge split plot design there are problems that by object texture serious interference, And maximum variance between clusters can effectively inhibit the influence of illumination variation, but when smooth surface generation mirror surface in target object part is anti- It penetrates or the gray scale of body surface and background is in the tonal gradation state of apparent 3 or 3 or more, tend to cause only The region segmentation of reflective or most highlighted tonal gradation is come out, and has ignored object other parts.In addition, in the mesh of crack image Mark shape be it is elongated in the case where carry out image segmentation, it is possible to lead to the fracture in target crack, so that whole crack performance At series crack segment, meanwhile, the interference characteristic present in the image and crack piece size quite, form quite or face Form and aspect were at that time, it is difficult to effectively realize effective detection of crack target.
Summary of the invention
The present invention is to overcome influence caused by can not eliminating the identification of interference characteristic fracture described in the above-mentioned prior art Defect provides a kind of Bridge Crack recognition methods based on recursive search.
In order to solve the above technical problems, technical scheme is as follows:
A kind of Bridge Crack recognition methods based on recursive search, comprising the following steps:
S1: being converted to crack grayscale image G for crack image to be identified, is split using threshold split plot design to described It stitches image and carries out Threshold segmentation, obtain segmented image T；
S2: Feature Selection is carried out to the segmented image T, obtains suspicious crack set of segments Cp={ Cpi, wherein i be can Doubt the serial number of crack segment；
S3: to the suspicious crack set of segments CpSeriatim carry out parameter initialization；
S4: according to the suspicious crack set of segments CpAft terminal ordinate value, by suspicious crack set of segments CpRoot It resequences according to descending, obtains the crack set of segments P for completing sequence；
S5: it scans for identifying by the crack identification algorithm fracture set of segments P based on recursive search, output obtains Complete recursive detected crack data；
S6: the crack data that cleaning detects, and exported as recognition result.
In the technical program, Threshold segmentation is carried out using threshold split plot design fracture image first, and screen Suspicious crack segment out, then using setting matching threshold, effectively improve robustness of the invention, searched finally by based on recurrence The crack identification algorithm of rope, is discharged that, form suitable with crack piece size be suitable or the comparable interference characteristic of color, effectively into Row search, detection and combination many cracks.
Preferably, in S1 step, the tool of Threshold segmentation is carried out to the crack image using threshold split plot design Steps are as follows for body:
S11: mean filter is carried out to the crack grayscale image G, obtains mean filter image M；
S12: image T=is enabled | M-G |, and image T is traversed, when its pixel gray value is greater than preset threshold offset, 255 are set by the gray value of the pixel, obtains the segmented image T for completing Threshold segmentation.
Preferably, in S2 step, carrying out Feature Selection to segmented image T, specific step is as follows:
S21: the connected domain profile C of detection segmented image Tk, wherein k indicates the sequence of the connected domain profile detected Number；
S22: the connected domain profile C is calculatedkArea s, perimeter l, minimum circumscribed rectangle long side w and minimum circumscribed rectangle Short side h calculates its length-width ratio f and circularity e and judges whether to meet suspicious crack Rule of judgment, if then by connected domain profile CkLabeled as suspicious crack segment Cpi, wherein the calculation formula of the length-width ratio f and circularity e is as follows:
S23: it after traversing all connected domains, marks described as crack segment CpiIt is integrated into suspicious crack set of segments Cp={ Cpi}。
Preferably, the suspicious crack Rule of judgment is connected domain profile CkLength-width ratio f > t0Or connected domain profile CkCircularity e meet t1< e and f > t2, wherein t0、t1、t2For the preset threshold value obtained by many experiments, and t0> t2。
Preferably, in S3 step, initiation parameter includes crack segment CpiExtreme coordinates, direction vector, length, mark Position, fragment and segment angle, wherein to the suspicious crack set of segments CpSeriatim carry out the specific of parameter initialization Steps are as follows:
Define crack segment CpiForward terminal be its profile in the smallest point of ordinate value；
Define crack segment CpiAft terminal be its profile in the maximum point of ordinate value；
Define crack segment CpiDirection vector be its minimum circumscribed rectangle longitudinal direction vector, direction be by rear end It is directed toward one end where forward terminal in one end where point；
Define crack segment CpiLength L (Cpi) be its minimum circumscribed rectangle long side length；
Define crack segment CpiThe crack flag bit flag=-1；
Define crack segment CpiWith crack segment CpjFragment D (Cpi,Cpj) it is crack segment CpiForward terminal with Crack segment CpjAft terminal between Euclidean distance, wherein j be crack segment serial number；
Define crack segment CpiWith crack segment CpjSegment included angle A (Cpi,Cpj) it is vectorWith crack segment CpjSide To the angle of vector, wherein vectorFor with crack segment CpiForward terminal be starting point, with crack segment CpjAft terminal be linked to be Vector.
Preferably, it in S5 step, is scanned for by the crack identification algorithm fracture set of segments P based on recursive search Specific step is as follows for identification:
S51: initialization crack quantity m=1；
S52: take the crack segment of first crack flag bit flag=-1 as currently processed from the set of segments P of crack Segment P0, by its crack, flag bit is set as flag=m, judges in the set of segments P of crack in addition to currently processed segment P0It is outer whether There are also the crack flag bit flag=-1 of other crack segments, if so, executing S53 step；S58 step is executed if it is not, then jumping Suddenly；
S53: take the crack segment of second crack flag bit flag=-1 as to be matched from the set of segments P of crack Section Pn；
S54: currently processed segment P is calculated0With segment P to be matchednFragment D (P0,Pn) and its length L (P0)、L (Pn), judge the fragment D (P0,Pn) whether it is less than preset distance threshold doff, if so, executing S55 step；If It is no, then further judge the fragment D (P0,Pn) whether meet D (P0,Pn) 5 × min of < (L (P0),L(Pn)), if so, It then jumps and executes S56 step；If it is not, then enabling m=m+1, then jumps and execute S52 step；
S55: currently processed segment P is calculated0With segment P to be matchednSegment included angle A (P0,Pn), judge segment included angle A (P0,Pn) whether it is less than preset angle threshold aoff, if so, by segment P to be matchednCrack flag bit be set as flag =m, and will current segment P to be matchednAs currently processed segment P0=Pn, take next crack segment as segment to be matched Pn=Pn+1, and jump and execute S57 step；If it is not, thening follow the steps S56；
S56: judge current segment P to be matchednIt whether is the last one element in the set of segments P of crack, if so, enabling Then m=m+1 is jumped and is executed S52 step；If it is not, then taking next crack segment as segment P to be matchedn=Pn+1, and hold Row S57 step；
S57: judge current segment P to be matchednWhether in the set of segments P of crack, S54 step is executed if so, jumping, If it is not, then executing S58 step；
S58: judge whether crack quantity m is equal with the element number in the set of segments P of crack, if so, by recurrence knot Crack set of segments P after beam is exported as the crack data for completing identification；If it is not, then that crack flag bit flag is identical Crack fragment combination be new crack segment, and corresponding crack segment is merged into a new composite unit in set P Element, and its parameter is initialized, then jump execution S51 and re-start recursive search.
Preferably, the calculation formula of distance threshold doff are as follows:
Doff=f (d) × min (L (P0),L(Pn))
The calculation formula of the angle threshold aoff are as follows:
Aoff=g (d)
Wherein, f (d) and g (d) is with currently processed segment P0With segment P to be matchednFragment D (P0,Pn) conduct The linear function formula of independent variable.
Preferably, in S58 step, to parameter initialization is carried out by new composite component, specific step is as follows:
The direction vector of the composite component: being initialized as the longitudinal direction vector of its minimum circumscribed rectangle by step A, side To for from bottom to top；
Step B: the forward terminal of the composite component is initialized as to the forward terminal of its last one daughter element；
Step C: the aft terminal of the composite component is initialized as to the aft terminal of its first daughter element；
Step D: -1 is set by the crack flag bit flag of the composite component.
Preferably, in S6 step, the specific steps for cleaning the crack data detected include: to judge the fracture number It is added whether be greater than preset threshold value l according to the length of whole daughter elements in middle combination crackc, if so, retaining the combination crack Data, if it is not, then abandoning the combination crack data.
Compared with prior art, the beneficial effect of technical solution of the present invention is: by using the crack based on recursive search Recognizer fracture segment identified, can effectively exclude that size is suitable, form is suitable or the comparable interference characteristic of color, Effectively eliminate influence caused by the identification of interference characteristic fracture；When carrying out crack identification using threshold value setting is carried out, make this hair Bright more robustness；The image partition method divided using threshold, can effectively weaken illumination variation, image background Influence of the disturbing factors such as texture to Threshold segmentation.
Detailed description of the invention
Fig. 1 is the flow chart of the Bridge Crack recognition methods based on recursive search of the present embodiment.
Fig. 2 is the flow chart that the fracture set of segments of the present embodiment scans for identification.
Fig. 3 is the image grayscale image in crack to be identified of the present embodiment.
Fig. 4 is the image for completing threshold segmentation of the present embodiment.
Fig. 5 is the suspicious crack fragmentary views of the present embodiment.
Fig. 6 is the crack image of the completion recognition detection of the present embodiment.
Fig. 7 is the crack segment rough schematic view of the present embodiment.
Fig. 8 is the crack fragment combination schematic diagram that the completion of the present embodiment is once searched for.
Fig. 9 is that the completion of the present embodiment initializes the crack fragmentary views of combined new element set.
Specific embodiment
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent；
In order to better illustrate this embodiment, the certain components of attached drawing have omission, zoom in or out, and do not represent actual product Size；
To those skilled in the art, it is to be understood that certain known features and its explanation, which may be omitted, in attached drawing 's.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
As shown in Figure 1, the flow chart of the Bridge Crack recognition methods based on recursive search for the present embodiment.
The present embodiment proposes a kind of Bridge Crack recognition methods based on recursive search, comprising the following steps:
S1: being converted to crack grayscale image G for crack image to be identified, is split using threshold split plot design to described It stitches image and carries out Threshold segmentation, obtain segmented image T.
As shown in Figure 3,4, respectively the image grayscale image in crack to be identified of the present embodiment and complete threshold The segmented image of segmentation.
In this step, the specific steps of Threshold segmentation are carried out such as to the crack image using threshold split plot design Under:
S11: mean filter is carried out to the crack grayscale image G, obtains mean filter image M；
S12: image T=is enabled | M-G |, and image T is traversed, when its pixel gray value is greater than preset threshold offset, 255 are set by the gray value of the pixel, obtains the segmented image T for completing Threshold segmentation.
Wherein, set threshold value offset is to be obtained by test of many times, in the present embodiment, set threshold value The value range of offset is 2~8.
S2: Feature Selection is carried out to the segmented image T, obtains suspicious crack set of segments Cp={ Cpi, wherein i be can Doubt the serial number of crack segment.
In this step, carrying out Feature Selection to segmented image T, specific step is as follows:
S21: the connected domain profile C of detection segmented image Tk, wherein k indicates the sequence of the connected domain profile detected Number；
S22: the connected domain profile C is calculatedkArea s, perimeter l, minimum circumscribed rectangle long side w and minimum circumscribed rectangle Short side h calculates its length-width ratio f and circularity e and judges whether to meet suspicious crack Rule of judgment, if then by connected domain profile CkLabeled as suspicious crack segment Cpi, wherein the calculation formula of the length-width ratio f and circularity e is as follows:
S23: it after traversing all connected domains, marks described as crack segment CpiIt is integrated into suspicious crack set of segments Cp={ Cpi}。
As shown in figure 5, being the suspicious crack fragmentary views of the present embodiment.
In this step, the suspicious crack Rule of judgment is connected domain profile CkLength-width ratio f > t0Or connected domain wheel Wide CkCircularity e meet t1< e and f > t2, wherein t0、t1、t2For the preset threshold value obtained by many experiments, and t0> t2.In the present embodiment, t0Value range be (2,4), t1Value range be (0.03,0.07), t2Value range be (1,3).
S3: to the suspicious crack set of segments CpSeriatim carry out parameter initialization.
In this step, the required parameter initialized includes crack segment CpiExtreme coordinates, direction vector, length Degree, flag bit, fragment and segment angle, wherein to the suspicious crack set of segments CpIt is initial seriatim to carry out parameter Specific step is as follows for change:
Define crack segment CpiForward terminal be its profile in the smallest point of ordinate value；
Define crack segment CpiAft terminal be its profile in the maximum point of ordinate value；
Define crack segment CpiDirection vector be its minimum circumscribed rectangle longitudinal direction vector, direction be by rear end It is directed toward one end where forward terminal in one end where point；
Define crack segment CpiLength L (Cpi) be its minimum circumscribed rectangle long side length；
Define crack segment CpiThe crack flag bit flag=-1；
Define crack segment CpiWith crack segment CpjFragment D (Cpi,Cpj) it is crack segment CpiForward terminal with Crack segment CpjAft terminal between Euclidean distance, wherein j be crack segment serial number；
Define crack segment CpiWith crack segment CpjSegment included angle A (Cpi,Cpj) it is vectorWith crack segment CpjSide To the angle of vector, wherein vectorFor with crack segment CpiForward terminal be starting point, with crack segment CpjAft terminal be linked to be Vector.
S4: according to the suspicious crack set of segments CpAft terminal ordinate value, by suspicious crack set of segments CpRoot It resequences according to descending, obtains the crack set of segments P for completing sequence.
S5: it scans for identifying by the crack identification algorithm fracture set of segments P based on recursive search, output obtains Complete recursive detected crack data.
As shown in Fig. 2, scanning for the flow chart of identification for the fracture set of segments of the present embodiment.
In this step, identification is scanned for by the crack identification algorithm fracture set of segments P based on recursive search Specific step is as follows:
S51: initialization crack quantity m=1；
S52: take the crack segment of first crack flag bit flag=-1 as currently processed from the set of segments P of crack Segment P0, by its crack, flag bit is set as flag=m, judges in the set of segments P of crack in addition to currently processed segment P0It is outer whether There are also the crack flag bit flag=-1 of other crack segments, if so, executing S53 step；S58 step is executed if it is not, then jumping Suddenly；
S53: take the crack segment of second crack flag bit flag=-1 as to be matched from the set of segments P of crack Section Pn；
S54: currently processed segment P is calculated0With segment P to be matchednFragment D (P0,Pn) and its length L (P0)、L (Pn), judge the fragment D (P0,Pn) whether it is less than preset distance threshold doff, if so, executing S55 step；If It is no, then further judge the fragment D (P0,Pn) whether meet D (P0,Pn) 5 × min of < (L (P0),L(Pn)), if so, It then jumps and executes S56 step；If it is not, then enabling m=m+1, then jumps and execute S52 step；
S55: currently processed segment P is calculated0With segment P to be matchednSegment included angle A (P0,Pn), judge segment included angle A (P0,Pn) whether it is less than preset angle threshold aoff, if so, by segment P to be matchednCrack flag bit be set as flag =m, and will current segment P to be matchednAs currently processed segment P0=Pn, take next crack segment as segment to be matched Pn=Pn+1, and jump and execute S57 step；If it is not, thening follow the steps S56；
S56: judge current segment P to be matchednIt whether is the last one element in the set of segments P of crack, if so, enabling Then m=m+1 is jumped and is executed S52 step；If it is not, then taking next crack segment as segment P to be matchedn=Pn+1, and hold Row S57 step；
S57: judge current segment P to be matchednWhether in the set of segments P of crack, S54 step is executed if so, jumping, If it is not, then executing S58 step；
S58: judge whether crack quantity m is equal with the element number in the set of segments P of crack, if so, by recurrence knot Crack set of segments P after beam is exported as the crack data for completing identification；If it is not, then that crack flag bit flag is identical Crack fragment combination be new crack segment, and corresponding crack segment is merged into a new composite unit in set P Element, and its parameter is initialized, then jump execution S51 and re-start recursive search.
Wherein, to parameter initialization is carried out by new composite component, specific step is as follows:
The direction vector of the composite component: being initialized as the longitudinal direction vector of its minimum circumscribed rectangle by step A, side To for from bottom to top；
Step B: the forward terminal of the composite component is initialized as to the forward terminal of its last one daughter element；
Step C: the aft terminal of the composite component is initialized as to the aft terminal of its first daughter element；
Step D: -1 is set by the crack flag bit flag of the composite component.
In addition, the calculation formula of the distance threshold doff in this step are as follows:
Doff=f (d) × min (L (P0),L(Pn))
The calculation formula of angle threshold aoff are as follows:
Aoff=g (d)
Wherein, f (d) and g (d) is with currently processed segment P0With segment P to be matchednFragment D (P0,Pn) conduct The linear function formula of independent variable, in the specific implementation process, distance threshold doff and angle threshold aoff dynamic adjust, and make this reality It applies the Bridge Crack recognition methods that example is proposed and has more robustness.In the present embodiment, the value range of f (d) is (0,5), g (d) value range is (0,45 °).
S6: the crack data that cleaning detects judge whole daughter elements that crack is combined in the crack data Length is added whether be greater than preset threshold value lc, if so, retaining the combination crack data, and defeated as recognition result progress Out, if it is not, then abandoning the combination crack data.
In the present embodiment, threshold value lcIt is determined by many experiments data, value range is (200,2000).
As shown in fig. 6, the crack image of the completion recognition detection for the present embodiment.
For scanning for identifying using the crack identification algorithm fracture set of segments based on recursive search, specific real During applying, crack segment rough schematic view is as shown in fig. 7, wherein P1~P10For the ordinate of the aft terminal according to crack segment The crack set of segments P that value is obtained according to descending sort.
In search identification process, crack quantity is subjected to initialization m=1 first, takes P1For currently processed segment, by it Crack flag bit is set as flag=m=1, carries out first time search；Take P2As segment to be matched, due to segment P1And segment P2Between segment included angle A (P1,P2) be greater than preset angle threshold aoff, therefore choose backward in the set of segments P of crack to Segment is matched, until segment P to be matched4Meet fragment D (P1,P4) it is less than its corresponding distance threshold doff=f (d) × min (L(P1),L(P4)), and segment included angle A (P1,P4) be less than preset angle threshold aoff=g (d), i.e. expression segment P4With segment P1Match, therefore by segment P4Crack flag bit be set as flag=m=1；
By segment P4As currently processed segment, segment P is chosen backward in the set of segments P of crack5As to be matched Section, due to due to segment P1With segment P2Between fragment D (P4,P5) it is greater than distance threshold doff, therefore in crack segment Segment to be matched is chosen in set P backward, until segment P to be matched6Meet fragment D (P4,P6) it is less than distance threshold doff, And segment included angle A (P4,P6) be less than preset angle threshold aoff, i.e. expression segment P6With segment P4Match, and by segment P4 Crack flag bit be set as flag=m=1；Similarly, by segment P6As currently processed segment, in the set of segments P of crack Segment P is chosen backward7As segment to be matched, repeat the above steps to identification segment P10When still without matching segment is found, then Export first group of combination crack segment { P1,P4,P6, and m=m+1=2 is set；
Since crack, set of segments P takes second of search the segment that first flag bit is -1, due to segment P1Upper It states flag bit in step and is set as flag=1, therefore take segment P in epicycle search2As currently processed segment, by its crack Flag bit is set as flag=m=2, takes P3As segment to be matched, above-mentioned search step is repeated, is split until completing all combinations Stitch set of segments, respectively { P1,P4,P6}、{P2}、{P3,P5,P8}、{P7And { P9,P10, then by said combination crack segment Set merges into an element respectively, i.e., by { P1,P4,P6Group is combined into new element { { P1_1,P1_2,P1_3, by { P2As new Element { P2, by { P3,P5,P8Group is combined into new element { { P3_1,P3_2,P3_3, by { P7It is used as new element { P4, by { P9,P10} Group is combined into new element { { P5_1,P5_2, as shown in figure 9, the new element combination diagram combined for the completion of the present embodiment, then Combine into the { { P of new element1_1,P1_2,P1_3}}、{P2}、{{P3_1,P3_2,P3_3}}、{P4}、{{P5_1,P5_2Be merged into newly Crack set of segments Pnew={ { P1,P4,P6},{P2},{P3,P5,P8},{P7},{P9,P10, and initialize a composite component Direction vector, forward terminal and aft terminal, as shown in Figure 8, Figure 9, the crack respectively once searched for for the completion of the present embodiment Fragment combination schematic diagram and initialization composite component parameter schematic diagram.If crack set of segments PnewElement number be less than set P Element number, then enable P=Pnew, then repeat the above steps to set P and complete the recursive search of all composite components, otherwise Terminate the search identification of fracture set of segments P, cleaning detects the crack that identification obtains and output.
In the present embodiment, identified by using the crack identification algorithm fracture segment based on recursive search, it can Effectively exclusion size is suitable, form is suitable or the comparable interference characteristic of color, effectively eliminates the identification of interference characteristic fracture and causes Influence；When carrying out crack identification using threshold value setting is carried out, the present invention is made to have more robustness；Using threshold point The image partition method cut can effectively weaken influence of the disturbing factors such as illumination variation, image background texture to Threshold segmentation.
The same or similar label correspond to the same or similar components；
The terms describing the positional relationship in the drawings are only for illustration, should not be understood as the limitation to this patent；
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention Protection scope within.

Claims (9)

1. a kind of Bridge Crack recognition methods based on recursive search, which comprises the following steps:
S1: crack image to be identified is converted into crack grayscale image G, using threshold split plot design to the crack pattern As carrying out Threshold segmentation, segmented image T is obtained；
S2: Feature Selection is carried out to the segmented image T, obtains suspicious crack set of segments Cp={ Cpi, wherein i is suspicious splits The serial number of patch section；
S3: to the suspicious crack set of segments CpParameter initialization is seriatim carried out, the parameter includes suspicious crack segment Extreme coordinates；
S4: according to the suspicious crack set of segments CpAft terminal ordinate value, by suspicious crack set of segments CpAccording to drop Sequence rearrangement obtains the crack set of segments P for completing sequence；
S5: by the crack identification algorithm fracture set of segments P based on recursive search, output obtains completing recursive identified Crack data；
S6: the crack data that cleaning detects, and exported as recognition result.
2. Bridge Crack recognition methods according to claim 1, it is characterised in that: dynamic using part in the S1 step State thresholding method carries out Threshold segmentation to the crack image, and specific step is as follows:
S11: mean filter is carried out to the crack grayscale image G, obtains mean filter image M；
S12: enabling image T=M-G, and traverses image T, when its pixel gray value is greater than preset threshold offset, by the picture The gray value of vegetarian refreshments is set as 255, obtains the segmented image T for completing Threshold segmentation.
3. Bridge Crack recognition methods according to claim 1, it is characterised in that: in the S2 step, to the segmentation Image T carries out Feature Selection, and specific step is as follows:
S21: the connected domain profile C of detection segmented image Tk, wherein k indicates the serial number of the connected domain profile detected；
S22: the connected domain profile C is calculatedkArea s, perimeter l, minimum circumscribed rectangle long side w and minimum circumscribed rectangle short side H calculates its length-width ratio f and circularity e and judges whether to meet suspicious crack Rule of judgment, if then by connected domain profile CkMark It is denoted as suspicious crack segment Cpi, wherein the calculation formula of the length-width ratio f and circularity e is as follows:
S23: it after traversing all connected domains, marks described as crack segment CpiIt is integrated into suspicious crack set of segments Cp= {Cpi}。
4. Bridge Crack recognition methods according to claim 3, it is characterised in that: the suspicious crack Rule of judgment is, Connected domain profile CkLength-width ratio f > t0Or connected domain profile CkCircularity e meet t1< e and f > t2, wherein t0、t1、 t2For the preset threshold value obtained by many experiments, and t0> t2。
5. Bridge Crack recognition methods according to claim 1, it is characterised in that: in the S3 step, initiation parameter It further include crack segment CpiDirection vector, length, flag bit, fragment and segment angle, wherein to the suspicious crack Set of segments CpSeriatim carrying out parameter initialization, specific step is as follows:
Define crack segment CpiForward terminal be its profile in the smallest point of ordinate value；
Define crack segment CpiAft terminal be its profile in the maximum point of ordinate value；
Define crack segment CpiDirection vector be its minimum circumscribed rectangle longitudinal direction vector, direction is by aft terminal institute One end be directed toward forward terminal where one end；
Define crack segment CpiLength L (Cpi) be its minimum circumscribed rectangle long side length；
Define crack segment CpiThe crack flag bit flag=-1；
Define crack segment CpiWith crack segment CpjFragment D (Cpi,Cpj) it is crack segment CpiForward terminal and crack piece Section CpjAft terminal between Euclidean distance, wherein j be crack segment serial number；
Define crack segment CpiWith crack segment CpjSegment included angle A (Cpi,Cpj) it is vectorWith crack segment CpjDirection to The angle of amount, wherein vectorFor with crack segment CpiForward terminal be starting point, with crack segment CpjAft terminal be linked to be to Amount.
6. Bridge Crack recognition methods according to claim 5, it is characterised in that: in the S5 step, passed by being based on Returning the crack identification algorithm fracture set of segments P of search to scan for identification, specific step is as follows:
S51: initialization crack quantity m=1；
S52: take the crack segment of first crack flag bit flag=-1 as currently processed segment from the set of segments P of crack P0, by its crack, flag bit is set as flag=m, judges in the set of segments P of crack in addition to currently processed segment P0Outside whether also The crack flag bit flag=-1 of other crack segments, if so, executing S53 step；S58 step is executed if it is not, then jumping；
S53: take the crack segment of second crack flag bit flag=-1 as segment P to be matched from the set of segments P of crackn；
S54: currently processed segment P is calculated0With segment P to be matchednFragment D (P0,Pn) and its length L (P0)、L(Pn), Judge the fragment D (P0,Pn) whether it is less than preset distance threshold doff, if so, executing S55 step；If it is not, then Further judge the fragment D (P0,Pn) whether meet D (P0,Pn) 5 × min of < (L (P0),L(Pn)), if so, jumping Execute S56 step；If it is not, then enabling m=m+1, then jumps and execute S52 step；
S55: currently processed segment P is calculated0With segment P to be matchednSegment included angle A (P0,Pn), judge segment included angle A (P0,Pn) Whether preset angle threshold aoff is less than, if so, by segment P to be matchednCrack flag bit be set as flag=m, and It will current segment P to be matchednAs currently processed segment P0=Pn, take next crack segment as segment P to be matchedn= Pn+1, and jump and execute S57 step；If it is not, thening follow the steps S56；
S56: judge current segment P to be matchednIt whether is the last one element in the set of segments P of crack, if so, enabling m=m+ 1, it then jumps and executes S52 step；If it is not, then taking next crack segment as segment P to be matchedn=Pn+1, and execute S57 Step；
S57: judge current segment P to be matchednWhether in the set of segments P of crack, S54 step is executed if so, jumping, if it is not, Then execute S58 step；
S58: judge whether crack quantity m is equal with the element number in the set of segments P of crack, if so, by after recurrence Crack set of segments P as complete identification crack data exported；If it is not, then splitting crack flag bit flag is identical Seam fragment combination is new crack segment, and corresponding crack segment is merged into a new composite component in set P, and Its parameter is initialized, then jumps execution S51 and re-starts recursive search.
7. Bridge Crack recognition methods according to claim 6, it is characterised in that: the calculating of the distance threshold doff is public Formula are as follows:
Doff=f (d) × min (L (P0),L(Pn))
The calculation formula of the angle threshold aoff are as follows:
Aoff=g (d)
Wherein, f (d) and g (d) is with currently processed segment P0With segment P to be matchednFragment D (P0,Pn) it is used as independent variable Linear function formula.
8. Bridge Crack recognition methods according to claim 6, it is characterised in that: in the S58 step, to by new group Closing element progress parameter initialization, specific step is as follows:
Step A: the direction vector of the composite component is initialized as to the longitudinal direction vector of its minimum circumscribed rectangle, direction is From bottom to top；
Step B: the forward terminal of the composite component is initialized as to the forward terminal of its last one daughter element；
Step C: the aft terminal of the composite component is initialized as to the aft terminal of its first daughter element；
Step D: -1 is set by the crack flag bit flag of the composite component.
9. Bridge Crack recognition methods according to claim 1, it is characterised in that: in the S6 step, cleaning is detected The specific steps of obtained crack data include: to judge that the length that whole daughter elements in crack are combined in the crack data is added Whether preset threshold value l is greater thanc, if so, retaining the combination crack data, if it is not, then abandoning the combination crack data.
CN201910673031.6A 2019-07-24 2019-07-24 Bridge crack identification method based on recursive search Active CN110487497B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910673031.6A CN110487497B (en) 2019-07-24 2019-07-24 Bridge crack identification method based on recursive search

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910673031.6A CN110487497B (en) 2019-07-24 2019-07-24 Bridge crack identification method based on recursive search

Publications (2)

Publication Number Publication Date
CN110487497A true CN110487497A (en) 2019-11-22
CN110487497B CN110487497B (en) 2021-09-24

Family

ID=68548207

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910673031.6A Active CN110487497B (en) 2019-07-24 2019-07-24 Bridge crack identification method based on recursive search

Country Status (1)

Country Link
CN (1) CN110487497B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111105408A (en) * 2019-12-27 2020-05-05 南京大学 Building surface crack detection method and system based on image processing
CN111784667A (en) * 2020-06-30 2020-10-16 北京海益同展信息科技有限公司 Crack identification method and device
CN116630899A (en) * 2023-07-21 2023-08-22 四川公路工程咨询监理有限公司 Highway side slope disease monitoring and early warning system

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104008553A (en) * 2014-06-17 2014-08-27 武汉武大卓越科技有限责任公司 Crack detection method with image gradient direction histogram and watershed method conflated

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104008553A (en) * 2014-06-17 2014-08-27 武汉武大卓越科技有限责任公司 Crack detection method with image gradient direction histogram and watershed method conflated

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
苏成悦 等: "用光学方法实现多灰度级图像的区域分割", 《应用光学》 *
贾迪 等: "面向线阵列CCD道路影像的裂缝识别", 《中国图象图形学报》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111105408A (en) * 2019-12-27 2020-05-05 南京大学 Building surface crack detection method and system based on image processing
CN111105408B (en) * 2019-12-27 2022-05-03 南京大学 Building surface crack detection method and system based on image processing
CN111784667A (en) * 2020-06-30 2020-10-16 北京海益同展信息科技有限公司 Crack identification method and device
CN111784667B (en) * 2020-06-30 2023-09-26 京东科技信息技术有限公司 Crack identification method and device
CN116630899A (en) * 2023-07-21 2023-08-22 四川公路工程咨询监理有限公司 Highway side slope disease monitoring and early warning system

Also Published As

Publication number Publication date
CN110487497B (en) 2021-09-24

Similar Documents

Publication Publication Date Title
US11922615B2 (en) Information processing device, information processing method, and storage medium
CN110487497A (en) A kind of Bridge Crack recognition methods based on recursive search
CN106157323B (en) A kind of insulator division and extracting method of dynamic division threshold value and block search combination
JP4137969B2 (en) Eye detection device, eye detection method, and program
CN106023257B (en) A kind of method for tracking target based on rotor wing unmanned aerial vehicle platform
CN108765402B (en) Non-woven fabric defect detection and classification method
CN104700112A (en) Method for detecting parasite eggs in excrement based on morphological characteristics
CN115294159B (en) Method for dividing corroded area of metal fastener
CN113096121B (en) Pavement crack detection method and system based on cross fracture mechanics and image processing
Feng et al. A separating method of adjacent apples based on machine vision and chain code information
CN110749598A (en) Silkworm cocoon surface defect detection method integrating color, shape and texture characteristics
CN110175574A (en) A kind of Road network extraction method and device
CN111624680A (en) Precipitation forecast displacement correction algorithm based on object diagnosis
JP2008146278A (en) Cell outline extraction device, cell outline extraction method and program
CN110458019B (en) Water surface target detection method for eliminating reflection interference under scarce cognitive sample condition
CN106897723B (en) Target real-time identification method based on characteristic matching
KR101101142B1 (en) A system and method for identifying the iris of persons in unrestricted condition
CN115731257A (en) Leaf form information extraction method based on image
WO2001071663A1 (en) Cell lineage extracting method
KR101423153B1 (en) Invariant radial iris segmentation
Wang et al. Optical fiber defect detection method based on DSSD network
CN113192100B (en) Time-sharing overlapped plant image key feature area edge path acquisition method
CN112434712B (en) Local binary pattern texture image feature extraction method and system based on scale and angle self-adaptive selection
CN114700941A (en) Strawberry picking method based on binocular vision and robot system
CN111046861B (en) Method for identifying infrared image, method for constructing identification model and application

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant