CN112907549A - Shoe printing pattern fracture characteristic detection and description method and system - Google Patents

Shoe printing pattern fracture characteristic detection and description method and system Download PDF

Info

Publication number
CN112907549A
CN112907549A CN202110225954.2A CN202110225954A CN112907549A CN 112907549 A CN112907549 A CN 112907549A CN 202110225954 A CN202110225954 A CN 202110225954A CN 112907549 A CN112907549 A CN 112907549A
Authority
CN
China
Prior art keywords
fracture
image
pattern
connected domain
broken
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
CN202110225954.2A
Other languages
Chinese (zh)
Other versions
CN112907549B (en
Inventor
王新年
李宝瑞
齐国清
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.)
Dalian Maritime University
Original Assignee
Dalian Maritime University
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 Dalian Maritime University filed Critical Dalian Maritime University
Priority to CN202110225954.2A priority Critical patent/CN112907549B/en
Publication of CN112907549A publication Critical patent/CN112907549A/en
Application granted granted Critical
Publication of CN112907549B publication Critical patent/CN112907549B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06T7/0004—Industrial image inspection
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge 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
    • G06T7/168—Segmentation; Edge detection involving transform domain methods
    • 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
    • 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
    • G06T2207/20048—Transform domain processing
    • G06T2207/20061—Hough transform
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Geometry (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention provides a method for detecting and describing shoe print pattern fracture characteristics in a shoe print image, which is characterized by comprising the following steps of: preprocessing the input shoe print image; detecting the pattern fracture blocks on the input shoe print image; the pattern fracture block is an image block containing a fracture pattern; detecting the broken patterns; predicting the pattern fracture area; the pattern fracture area is described. The method does not depend on the directions of the patterns at the two ends of the fracture area when the fracture area is detected, and can also effectively detect the pattern fracture area when the directions of the patterns at the two ends of the fracture area have large difference. Meanwhile, the fracture area is predicted by combining the global appearance information and the local geometric information of the patterns, so that the description result of the fracture area is more accurate. The geometric information of the crack characteristics in the shoe print can be effectively obtained, and the comparison work of the wearing characteristics of the shoe print is facilitated.

Description

Shoe printing pattern fracture characteristic detection and description method and system
Technical Field
The invention relates to the technical field of a quantitative analysis method and a system for characteristics of shoe printing, in particular to a method and a system for detecting and describing shoe printing pattern fracture characteristics.
Background
At present, a fracture detection and prediction algorithm based on fingerprint lines and high-resolution remote sensing image road extraction and character refinement is available, and a fracture detection and prediction algorithm based on the fingerprint lines is a fracture line connection method adopting a reference method; an improved circular template matching tracking method is adopted in a fracture detection and prediction algorithm based on high-resolution remote sensing image road extraction; the fracture detection and prediction algorithm based on the refined characters has a prediction method by utilizing the structural characteristics of the characters and pre-calculated statistical information, and the specific content of each method is as follows:
(1) a breaking detection and prediction algorithm based on fingerprint lines is a breaking line connection method adopting a reference method, the method comprises the steps of firstly scanning and refining a fingerprint image point by point to find a breaking point p, calculating a normal line at the breaking point, considering the breaking point as a breaking point to be matched if an intersection point q exists between the normal line and the adjacent fingerprint lines and the distance between the intersection point q and the breaking point p is smaller than a threshold value, then searching for matched breaking points, calculating the normal line and the intersection point between the normal line and the adjacent fingerprint lines of all the found unrepaired breaking points, finally screening the unrepaired breaking points according to screening conditions to obtain the best matched breaking point, and referring to the adjacent fingerprint lines to connect the breaking point with the best matched breaking point. (2) The method comprises the steps of firstly judging a fracture area according to two characteristics of short length of a fracture and small angle difference of line segments on two sides of the fracture, and then tracking the road area by using an improved circular template matching method to predict the fracture by taking a fracture endpoint as a seed point. (3) The method comprises the steps of firstly predicting the fracture length according to the size of an image matrix after character normalization and the summarized statistical information, then detecting end points in a refined character image according to certain continuous model rules, then fitting a possible continuous model by using end points and a certain number of connected elements, then performing pre-extension according to the predicted length, judging the end points as break points if the continuous model extends and then fitting the break points again to determine the continuous model for interpolation prediction.
The existing fracture area detection and prediction algorithm has the following problems: (1) the fingerprint image broken line prediction method based on the fingerprint line has the problems that the broken line detection and prediction can only be carried out on the line, the broken line prediction method can not be used for predicting broken areas with different widths, the reference method needs to depend on the fingerprint line near the broken line, and the effect of the broken prediction can be influenced when the near line is defective. (2) The fracture detection and prediction algorithm based on the high-resolution remote sensing image road extraction has the problems that when the fracture is detected and connected by using the improved circular template matching and tracking method, the fracture can only be connected aiming at fracture areas with short fracture length and small angle difference of line segments on two sides of the fracture, so that the applicability of the method is limited. (3) The fracture detection and prediction method based on the thinned characters needs to determine the size of an image matrix in advance according to specific situations to be predicted when the method is applied specifically, and the fracture length is predicted through a large amount of statistical information, so that the method is not very convenient to use.
Disclosure of Invention
In light of the above-mentioned technical problems, a method and system for detecting and describing shoe pattern fracture characteristics are provided. The invention mainly utilizes a method for detecting and describing shoe print pattern fracture characteristics in shoe print images, which comprises the following steps:
step S1: preprocessing the input shoe print image;
step S2: detecting the pattern fracture blocks on the input shoe print image; the pattern fracture block is an image block containing a fracture pattern;
step S3: detecting the broken patterns;
step S4: predicting the pattern fracture area;
step S5: the pattern fracture area is described.
Further, the preprocessing of the shoe print image further comprises the steps of:
s11: converting the input shoe print image into a gray level image, and performing binarization processing on the gray level image to obtain a shoe print binary image;
s12: carrying out corrosion and expansion morphological processing on the shoe printing binary image;
s13: dividing the shoe print binary image into a plurality of image blocks with fixed sizes and without overlapping, respectively calculating the entropy of each image block, and enabling the entropy to be larger than a preset threshold value E1The image blocks are put into a pattern fracture candidate block set C, and the mark is C ═ C1,c2,…,cv,…,cV}; where V represents the number of pattern break candidates.
Further, the step of detecting the pattern fracture block on the input shoe print image further comprises the following steps:
s21: traversing each pattern break candidate block cvV1, 2, …, V, the following steps S22-S26 are performed to form a pattern breaking block set R ═ R1,R2,…,Rj,…,RJAnd its corresponding broken component set Gj,j＝1,2,…,J；
S22: detecting the pattern fracture candidate block cvForm a set of connected domains
Figure RE-GDA0003029841690000031
Wherein v represents a fracture candidate block identifier and T represents the number of connected domains;
s23: judging the shape complexity of the connected domain;
calculating the ratio of the area of each connected region to the smallest convex polygon if the ratio is larger than a predetermined threshold value E2Then marking the connected domain as a common connected domain; if the area ratio is less than or equal to a predetermined threshold E2Then marking the connected domain as a complex connected domain;
s24: performing fracture component judgment on the common communication domain;
s25: performing fracture component judgment on the complex connected domain;
s26: and judging the pattern fracture block.
Further, a broken component determination is performed on the general connected domain:
s241: traversing each of the common connected domains;
s242: respectively calculating the minimum distances from four vertexes of the minimum circumscribed rectangle of the common connected domain to the image boundary of the pattern fracture candidate block, and enabling the minimum distances to be smaller than a preset threshold value E3The vertices of (a) are marked as vertices near the image boundary; calculating the number of the top points close to the image boundary and recording as eta;
s243: calculating the length difference between the long axis and the short axis of the common connected domain, and recording as rho;
s244: calculating the maximum overlapping length of the common connected domain and the pattern fracture candidate block image boundary, and recording as q;
s245: calculating the ratio of the length of the long axis of the common connected domain to the maximum overlapping length q, and recording as mu;
s246: the number of vertices η that will be met is less than a predetermined threshold E4And the length difference rho between the major axis and the minor axis is larger than a predetermined threshold value E5And the ratio mu of the length of the long axis to the maximum overlapping length q is larger than a predetermined threshold value E6The common connected domain of (a) is labeled as a suspected fracture component.
Further, a broken component determination is performed on the complex connected domain:
s251: traversing each of the complex connected domains;
s252: thinning the complex connected domain, and extracting a skeleton image of the complex connected domain; points, through which 3 skeleton lines pass, in the skeleton image are skeleton branch points; detecting a skeleton branch point in the skeleton image and judging whether the connected domain is detachable;
if a skeleton branch point is detected in the skeleton line image, the number of skeleton points in a 5 × 5 neighborhood with the skeleton branch point as the center is calculated, and if the number is greater than a threshold value E7Marking the complex connected domain as a detachable connected domain; if no skeleton branch point is detected in the skeleton line image, extracting the outer contour of the complex connected domain, carrying out Hough transform straight line detection on the outer contour of the complex connected domain, calculating the angle between the detected straight line segment and the horizontal direction and the angle difference between the two straight line segments, and if the angle difference between the two straight line segments is 80-90 degrees, marking the connected domain as a detachable connected domain; if the angle difference of no two straight line segments is 80-90 degrees, marking as an undetachable connected domain;
s253: splitting the splittable connected domain into a plurality of sub-connected domains;
s254: judging a component suspected of being broken;
for the detachable connected domain, executing steps S242-246 on the detached sub-connected domain, and marking the sub-connected domain meeting the judgment condition as a suspected fracture part;
for the undetachable connected domain, comparing the size of the circumscribed rectangle of the undetachable connected domain with the size of the pattern fracture candidate block, and if the height difference or the width difference between the circumscribed rectangle and the image block is smaller than a preset threshold value E8Then the connected domain is marked as a suspected fractured component.
Further, in the process of judging the pattern fracture block: if the pattern fracture candidate block contains two or more suspected fracture parts, the pattern fracture candidate block is a pattern fracture block and is added into a pattern fracture block set R, and simultaneously the suspected fracture parts corresponding to the fracture block are added into a fracture part set GjWherein j represents the identity of the broken block,
Figure RE-GDA0003029841690000041
k represents the block correspondenceNumber of parts suspected of being broken.
Further, the breaking pattern is detected:
s31: traversing the corresponding broken part set G of each pattern broken blockjThen, steps S32 to S33 are executed, and J is 1,2, …, J, so as to form a fracture pattern image set W corresponding to each fracture block1,W2,…,WJ；
S32: assembling the broken parts GjThe fracture parts in the method are combined to form a fracture pattern image set to be judged, and k is preset to be 1;
s321: to-be-broken member
Figure RE-GDA0003029841690000051
And
Figure RE-GDA0003029841690000052
respectively forming pattern images of the cracks to be judged
Figure RE-GDA0003029841690000053
Wherein
Figure RE-GDA0003029841690000054
Is marked as a first broken part,
Figure RE-GDA0003029841690000055
recording as a second fracture part, adding to a to-be-determined fracture pattern image set FkIn (1),
Figure RE-GDA0003029841690000056
wherein k is a broken part identifier;
s322: making K equal to K +1, if K is less than or equal to K-1, repeating the step S321, otherwise, stopping the execution;
s323: acquiring the fracture pattern image set F to be judged1,F2,…,Fk,…,FK-1；
S33: traversing the to-be-determined fracture pattern image set F1,F2,…,FK-1For the pattern image to be judged in the set
Figure RE-GDA0003029841690000057
Performing the following steps S34-S36, u ═ 1,2, …, K-K;
s34: judging the fracture based on the vertical projection of the fracture pattern image to be judged;
s341: vertically projecting the to-be-determined fracture pattern image, replacing a non-0 value in a projection histogram with 1, and then representing continuous 0 or 1 by using single 0 or 1; calculating the number of peaks in the projection represented by the number of 1;
s342: calculating the overlapping range and the main shaft angle of the projections of the two connected domains in the pattern image to be judged; step S341 is executed again after the to-be-determined fracture pattern image meeting one of the following conditions is rotated, and then step S343 is executed, otherwise, step S343 is executed directly, and the angle of image rotation is the angle theta between the connection line of the central points of the two connected domains and the horizontal direction;
a. the projection histogram of the pattern image to be determined only contains 1 after processing, and the overlapping range of the projections of the two connected domains is smaller than a preset threshold value E9；
b. The projection histogram of the fracture pattern image to be judged comprises 0 and 1 after processing, wherein the number of 1 is 1, and one of the following conditions is satisfied:
condition 1: the overlapping range of the two connected domain projections is less than a predetermined threshold value E9；
Condition 2: the angles of the two communicating areas are both larger than 80 degrees or the angle difference is larger than 60 degrees;
s343: if the projection of the to-be-determined breaking pattern image has two or more peaks, the image meets the breaking determination condition based on the vertical projection, and the step S35 is executed on the image, otherwise, the step S34 is executed on other to-be-determined breaking pattern images in the to-be-determined breaking pattern image set;
s35: determining fracture patterns based on geometric characteristics;
s351: detecting two connected domains in the pattern image to be judged, and acquiring a broken part image only containing a single broken part;
s352: determining a geometric characteristic judgment area of the fractured part; thinning the connected domain, and extracting a skeleton line of the connected domain; respectively extracting 4 continuous skeleton points taking end points at two ends of a skeleton line as starting points from the skeleton line image, and performing expansion operation on the continuous skeleton points by using circular structural elements; respectively putting the two expansion results and the other fracture part in the image of the fracture pattern to be judged into the same image, calculating and comparing the minimum distance between the two expansion results and the fracture part, and taking the expansion result with smaller minimum distance as a geometric characteristic judgment area of the fracture part;
placing the geometric characteristic judgment area into an empty image, and acquiring a geometric characteristic judgment area image corresponding to the broken part;
the calculation method of the minimum distance between the two areas comprises the following steps: respectively calculating the minimum value of the distance from one region to the nearest point of the other region, wherein the formula is as follows:
Figure RE-GDA0003029841690000061
Figure RE-GDA0003029841690000062
wherein A and B correspond to two regions respectively, a and B are points in the regions A and B, and d (a and B) is the distance between the two points; the minimum distance between the two regions is:
H(A,B)＝min{h(A,B),h(B,A)}；
s353: acquiring a broken part extension line segment; calculating the angle of a main shaft of a connected domain in the image of the broken part, rotating the image of the broken part and the image of the corresponding geometric characteristic judgment region until the angle of the main shaft of the connected domain is 0, then detecting the central positions of the broken part and the geometric characteristic judgment region, and judging the position of the geometric characteristic judgment region relative to the broken part in the horizontal direction;
calculating the main shaft angle of the geometric characteristic judgment area, rotating the image of the geometric characteristic judgment area until the main shaft angle of the geometric characteristic judgment area is 0, and detecting the central position of the geometric characteristic judgment area; creating an empty image with the same size as the image of the geometric characteristic determination area, drawing a horizontal line segment with the width of 1 to the image boundary along the direction of the geometric characteristic determination area relative to the broken part by taking the central position of the geometric characteristic determination area as a starting point in the empty image, and obtaining an extended line segment image;
reversely rotating the extension line segment image according to the main shaft angle of the geometric characteristic judgment area to obtain the extension line segment image of the broken part;
s354: judging the position relation of the extension line; putting the extension line segments of the two broken parts obtained in the step S353 into the same image, detecting and counting the number of connected domains in the image:
if the number of connected domains is 1, then: judging that the two extension line segments are intersected, extracting the intersection point of the two extension line segments, putting the intersection point and a first fracture part in the pattern image to be judged into the same image, and calculating the minimum distance from the intersection point to the first fracture part;
if the number of connected domains is 2, then: calculating the angle difference and the minimum distance of two connected domains in the graph, and if the angle difference of the two connected domains is smaller than a preset threshold value E10And the minimum distance is less than a predetermined threshold E11Judging that the directions of the two extension line segments are consistent;
s355: judging the geometric characteristics for the second time; performing median filtering on the fracture pattern image to be determined, smoothing the edge of a connected domain in the image, repeating the steps S351-S354, and determining the geometric characteristics again;
integrating the judgment results of the two geometric characteristics; if at least one intersection of the two extension line segments is determined in the two determinations, the two extension line segments are intersected; if the directions of the two extension line sections are consistent at least once in the two determinations, the directions of the two extension line sections are consistent;
s36: judging broken patterns; if the geometric characteristic judgment result of the to-be-judged fracture pattern image meets one of the following conditions, the image is the fracture pattern image: condition 1: the directions of the extension line sections of the broken parts are consistent; condition 2: the extension line segments of the breaking parts are intersected, and the distance between the intersection point and the first breaking part is the minimum;
if the same to-be-determined fracture pattern image set contains to-be-determined fracture pattern images respectively meeting the conditions 1 and 2, the to-be-determined fracture pattern image meeting the condition 1 is a fracture pattern image;
adding a fracture pattern image to a fracture pattern image set WjWherein j represents the identity of the broken block,
Figure RE-GDA0003029841690000071
z represents the number of breaking patterns in each breaking block.
Further, the pattern fracture area prediction method further comprises the following steps:
s41: traversing the fracture pattern image set W corresponding to each pattern fracture block1,W2,…,WJFor the broken pattern image therein
Figure RE-GDA0003029841690000081
The following steps S42-S44 are executed, where Z is 1,2, …, Z, and a breaking region prediction image set M of each pattern breaking block is formed1,M2,…,MJWhere j represents the identity of the broken block,
Figure RE-GDA0003029841690000082
s42: fracture region prediction based on global appearance features;
presetting a similarity parameter beta, and finding a fracture block R in the shoe print binary imagejThe N similar areas with the similarity larger than beta are marked as
Figure RE-GDA0003029841690000083
Fusing the N similar areas to obtain a fracture pattern image
Figure RE-GDA0003029841690000084
Fracture region predicted image based on global apparent features
Figure RE-GDA0003029841690000085
The fused image is:
Figure RE-GDA0003029841690000086
wherein wiComprises the following steps:
Figure RE-GDA0003029841690000087
wherein
Figure RE-GDA0003029841690000088
Is a distance weight, wdiIs a broken block RjSimilar region
Figure RE-GDA0003029841690000089
The center distance of (d);
Figure RE-GDA00030298416900000810
is a broken block RjSimilar region
Figure RE-GDA00030298416900000811
The similarity of (2);
s43: fracture region prediction based on local geometric features;
s44: fusing the prediction results of the fracture area; firstly, predicting the image of the fracture area
Figure RE-GDA00030298416900000812
And
Figure RE-GDA00030298416900000813
respectively with broken pattern images
Figure RE-GDA00030298416900000814
Subtracting to obtain a predicted image of the fracture region
Figure RE-GDA00030298416900000815
And
Figure RE-GDA00030298416900000816
corresponding two difference images
Figure RE-GDA00030298416900000817
And
Figure RE-GDA00030298416900000818
then detecting the image
Figure RE-GDA00030298416900000819
And
Figure RE-GDA00030298416900000820
of the connected component, removing
Figure RE-GDA00030298416900000821
And
Figure RE-GDA00030298416900000822
there is no connected domain of intersection, finally the image is processed
Figure RE-GDA00030298416900000823
As images of broken patterns
Figure RE-GDA00030298416900000824
Predicted image of the fracture region of (1), is recorded as
Figure RE-GDA00030298416900000825
And added to the set of predicted images M of the fracture regionjPerforming the following steps; rotated fracture region prediction image
Figure RE-GDA00030298416900000826
The above operation is performed after the reverse rotation is required.
Furthermore, the description of the pattern breaking areas also has the following steps:
s51: traversing each fracture area prediction image set M1,M2,…,MJTo break thereinSplit region predicted image
Figure RE-GDA0003029841690000091
Performing the following steps S52-S53, Z being 1,2, …, Z, to realize the boundary description of the pattern fracture area;
s52: detecting the predicted image of the fracture region
Figure RE-GDA0003029841690000092
The connected domain in the system is used for executing the following operation on each connected domain and extracting the connected domain to be described; detecting the external rectangular area of the connected domain, and counting the images of the fracture area
Figure RE-GDA0003029841690000093
If the number of the connected domains in the edge expanding region of the same region is 1, removing the breakage region predicted image
Figure RE-GDA0003029841690000094
The connected domain corresponding to the region; the final residual connected domain is a prediction image of a fracture region
Figure RE-GDA0003029841690000095
The connected domain to be described in (1);
s53: predicting images of fracture regions
Figure RE-GDA0003029841690000096
The connected domain to be described in the method carries out edge detection, extracts the outline of the connected domain to be described and realizes the fracture area
Figure RE-GDA0003029841690000097
The boundary description of (1).
Furthermore, the invention also comprises a system for detecting and describing shoe print pattern fracture characteristics in a shoe print image, which is characterized by comprising the following steps: the device comprises an image preprocessing module, a pattern fracture block detection module, a fracture pattern detection module, a pattern fracture region prediction module and a pattern fracture region description module;
the image preprocessing module is used for removing noise in the image by performing graying and binarization on the shoe print image and using morphological processing; dividing the shoe print image into image blocks with fixed sizes and non-overlapping, screening pattern fracture candidate blocks and sending the pattern fracture candidate blocks to the pattern fracture block detection module;
the pattern fracture block detection module is used for judging suspected fracture parts of the connected domains according to the forms of the connected domains in the pattern fracture candidate blocks, detecting whether the pattern fracture candidate blocks contain the suspected fracture parts or not, and screening the pattern fracture blocks from the pattern fracture candidate blocks;
the broken pattern detection module is used for detecting broken patterns in the pattern broken blocks;
the pattern fracture area prediction module is used for predicting fracture areas in the pattern fracture blocks;
and the pattern fracture area description module is used for describing the outline of the pattern fracture area.
Compared with the prior art, the invention has the following advantages:
the method does not depend on the directions of the patterns at the two ends of the fracture area when the fracture area is detected, and can also effectively detect the pattern fracture area when the directions of the patterns at the two ends of the fracture area have large difference.
The method of the invention predicts the fracture area by combining the global appearance information and the local geometric information of the patterns, so that the description result of the fracture area is more accurate.
The method is used for describing the fracture characteristics in the shoe print image, realizes geometric description of the fracture area, can effectively acquire the geometric information of the fracture characteristics in the shoe print, and facilitates the comparison work of the wear characteristics of the shoe print.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of the method for detecting and describing the breaking characteristics of the shoe printing patterns.
FIG. 2 is a diagram of a system for detecting and describing the breaking characteristics of the shoe printing patterns according to the present invention.
FIG. 3 is a schematic view of a pattern block of the present invention.
Fig. 4 is a schematic view of a general pattern of the present invention.
Fig. 5 is a schematic view of a complex pattern according to the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
As shown in fig. 1-2, the present invention provides a method for detecting and describing shoe print pattern fracture characteristics in a shoe print image, comprising the following steps:
step S1: and preprocessing the input shoe print image. The pre-processing of the shoe print image further comprises the steps of:
step S11: converting the input shoe print image into a gray level image, and performing binarization processing on the gray level image to obtain a shoe print binary image;
step S12: carrying out corrosion and expansion morphological processing on the shoe printing binary image;
step S13: dividing the shoe print binary image into a plurality of image blocks with fixed sizes and without overlapping, respectively calculating the entropy of each image block, and enabling the entropy to be larger than a preset threshold value E1The image blocks are put into a pattern fracture candidate block set C, and the mark is C ═ C1,c2,…,cv,…,cV}; where V represents the number of pattern break candidates.
Further, after the image preprocessing step ends, step S2 is executed: detecting the pattern fracture blocks on the input shoe print image; the pattern fracture block is an image block containing a fracture pattern.
The step of detecting the pattern fracture blocks on the input shoe print image further comprises the following steps:
step S21: traversing each pattern break candidate block cvV1, 2, …, V, the following steps S22-S26 are performed to form a pattern breaking block set R ═ R1,R2,…,Rj,…,RJAnd its corresponding broken component set Gj,j＝1,2,…,J；
Step S22: detecting the pattern fracture candidate block cvForm a set of connected domains
Figure RE-GDA0003029841690000111
Wherein v represents a fracture candidate block identifier and T represents the number of connected domains;
step S23: and judging the shape complexity of the connected domain. Calculating the ratio of the area of each connected region to the smallest convex polygon if the ratio is larger than a predetermined threshold value E2Then mark the connected domain as a normal connected domain(ii) a If the area ratio is less than or equal to a predetermined threshold E2Then marking the connected domain as a complex connected domain; as a preferred embodiment, the threshold E2Preferably 0.65.
Step S24: and judging the broken parts of the common communication area. The specific determination process is as follows:
step S241: traversing each of the common connected domains;
step S242: respectively calculating the minimum distances from four vertexes of the minimum circumscribed rectangle of the common connected domain to the image boundary of the pattern fracture candidate block, and enabling the minimum distances to be smaller than a preset threshold value E3The vertices of (a) are marked as vertices near the image boundary; calculating the number of the top points close to the image boundary and recording as eta;
step S243: calculating the length difference between the long axis and the short axis of the common connected domain, and recording as rho;
step S244: calculating the maximum overlapping length of the common connected domain and the pattern fracture candidate block image boundary, and recording as q;
step S245: calculating the ratio of the length of the long axis of the common connected domain to the maximum overlapping length q, and recording as mu;
step S246: the number of vertices η that will be met is less than a predetermined threshold E4And the length difference rho between the major axis and the minor axis is larger than a predetermined threshold value E5And the ratio mu of the length of the long axis to the maximum overlapping length q is larger than a predetermined threshold value E6The common connected domain of (a) is labeled as a suspected fracture component. Preferably, in practice, the threshold value E3Preferably 3, threshold value E4Preferably 20, threshold value E5Preferably 1.5.
Step S25: and judging the broken parts of the complex connected domain. The specific steps for determining the broken parts of the complex connected domain comprise the following steps:
step S251: traversing each of the complex connected domains;
step S252: thinning the complex connected domain, and extracting a skeleton image of the complex connected domain; the skeleton image of the complex connected component is an image including only connected component skeleton lines. The intersection point where 3 skeleton lines in the skeleton image pass is a skeleton branch point; detecting a skeleton branch point in the skeleton image and judging whether the connected domain is detachable;
if a skeleton branch point is detected in the skeleton line image, the number of skeleton points in a 5 × 5 neighborhood with the skeleton branch point as the center is calculated, and if the number is greater than a threshold value E7Marking the complex connected domain as a detachable connected domain; if no skeleton branch point is detected in the skeleton line image, extracting the outer contour of the complex connected domain, carrying out Hough transform straight line detection on the outer contour of the complex connected domain, calculating the angle between the detected straight line segment and the horizontal direction and the angle difference between the two straight line segments, and if the angle difference between the two straight line segments is 80-90 degrees, marking the connected domain as a detachable connected domain; if the angle difference of no two straight line segments is 80-90 degrees, marking as an undetachable connected domain;
step S253: and splitting the splittable connected domain into a plurality of sub-connected domains. As a preferred embodiment, the resolution method employed in the present application is Cao R, Tan C L.A model of stroke extraction from chip chromatography images [ C ]. Pattern Recognition,2000.proceedings.15th International Conference on.IEEE,2000.
Step S254: and (5) judging a suspected fracture part. For the detachable connected domain, executing steps S242-246 on the detached sub-connected domain, and marking the sub-connected domain meeting the judgment condition as a suspected fracture part;
for the undetachable connected domain, comparing the size of the circumscribed rectangle of the undetachable connected domain with the size of the pattern fracture candidate block, and if the height difference or the width difference between the circumscribed rectangle and the image block is smaller than a preset threshold value E8Then the connected domain is marked as a suspected fractured component.
Step S26: and judging the pattern fracture block. In the process of judging the pattern fracture block: if the pattern fracture candidate block contains two or more suspected fracture parts, the pattern fracture candidate block is a pattern fracture block and is added into the pattern fracture block set R, and simultaneously the suspected fracture parts corresponding to the fracture block are added into the fracture part setGjWherein j represents the identity of the broken block,
Figure RE-GDA0003029841690000131
k represents the number of suspected broken parts corresponding to the block. And step S13, the input shoe print image is blocked, and a pattern fracture candidate block is screened out according to the entropy of the image block. As described in step S21, the candidate blocks in the set C are traversed to perform the determination of the fragmentation block, where the fragmentation candidate block refers to the currently traversed candidate block.
Further, preferably, step S3: and detecting the breaking patterns. The detection of the breaking pattern further comprises the following steps:
step S31: traversing the corresponding broken part set G of each pattern broken blockjThen, steps S32 to S33 are executed, and J is 1,2, …, J, so as to form a fracture pattern image set W corresponding to each fracture block1,W2,…,WJ；
Step S32: assembling the broken parts GjThe fracture parts in the method are combined to form a fracture pattern image set to be judged, and k is preset to be 1;
step S321: to-be-broken member
Figure RE-GDA0003029841690000132
And
Figure RE-GDA0003029841690000133
respectively forming pattern images of the cracks to be judged
Figure RE-GDA0003029841690000134
Wherein
Figure RE-GDA0003029841690000135
Is marked as a first broken part,
Figure RE-GDA0003029841690000136
recording as a second fracture part, adding to a to-be-determined fracture pattern image set FkIn (1),
Figure RE-GDA0003029841690000137
wherein k is a broken part identifier;
step S322: making K equal to K +1, if K is less than or equal to K-1, repeating the step S321, otherwise, stopping the execution;
step S323: acquiring the fracture pattern image set F to be judged1,F2,…,Fk,…,FK-1；
Step S33: traversing the to-be-determined fracture pattern image set F1,F2,…,FK-1For the pattern image to be judged in the set
Figure RE-GDA0003029841690000141
Performing the following steps S34-S36, u ═ 1,2, …, K-K;
step S34: judging the fracture based on the vertical projection of the fracture pattern image to be judged; preferably, the projection means that each column of the circular image is sequentially judged whether the pixel value of each row is nonzero, and the number of all nonzero pixels in the column is counted. The method is to vertically project the pattern image to be judged
Step S341: vertically projecting the to-be-determined fracture pattern image, replacing a non-0 value in a projection histogram with 1, and then representing continuous 0 or 1 by using single 0 or 1; calculating the number of peaks in the projection represented by the number of 1;
step S342: calculating the overlapping range and the main shaft angle of the projections of the two connected domains in the pattern image to be judged; step S341 is executed again after the to-be-determined fracture pattern image meeting one of the following conditions is rotated, and then step S343 is executed, otherwise, step S343 is executed directly, and the angle of image rotation is the angle theta between the connection line of the central points of the two connected domains and the horizontal direction;
a. the projection histogram of the pattern image to be determined only contains 1 after processing, and the overlapping range of the projections of the two connected domains is smaller than a preset threshold value E9；
b. The projection histogram of the fracture pattern image to be judged comprises 0 and 1 after processing, wherein the number of 1 is 1, and one of the following conditions is satisfied:
condition 1: the overlapping range of the two connected domain projections is less than a predetermined threshold value E9；
Condition 2: the angles of the two communicating areas are both larger than 80 degrees or the angle difference is larger than 60 degrees;
step S343: if the projection of the to-be-determined breaking pattern image has two or more peaks, the image meets the breaking determination condition based on the vertical projection, and the step S35 is executed on the image, otherwise, the step S34 is executed on other to-be-determined breaking pattern images in the to-be-determined breaking pattern image set;
step S35: determining fracture patterns based on geometric characteristics;
step S351: detecting two connected domains in the pattern image to be judged, and acquiring a broken part image only containing a single broken part;
step S352: determining a geometric characteristic judgment area of the fractured part; thinning the connected domain, and extracting a skeleton line of the connected domain; respectively extracting 4 continuous skeleton points taking end points at two ends of a skeleton line as starting points from the skeleton line image, and performing expansion operation on the continuous skeleton points by using circular structural elements; respectively putting the two expansion results and the other fracture part in the image of the fracture pattern to be judged into the same image, calculating and comparing the minimum distance between the two expansion results and the fracture part, and taking the expansion result with smaller minimum distance as a geometric characteristic judgment area of the fracture part;
placing the geometric characteristic judgment area into an empty image, and acquiring a geometric characteristic judgment area image corresponding to the broken part;
the calculation method of the minimum distance between the two areas comprises the following steps: respectively calculating the minimum value of the distance from one region to the nearest point of the other region, wherein the formula is as follows:
Figure RE-GDA0003029841690000151
Figure RE-GDA0003029841690000152
wherein A and B correspond to two regions respectively, a and B are points in the regions A and B, and d (a and B) is the distance between the two points; the minimum distance between the two regions is:
H(A,B)＝min{h(A,B),h(B,A)}；
step S353: acquiring a broken part extension line segment; calculating the angle of a main shaft of a connected domain in the image of the broken part, rotating the image of the broken part and the image of the corresponding geometric characteristic judgment region until the angle of the main shaft of the connected domain is 0, then detecting the central positions of the broken part and the geometric characteristic judgment region, and judging the position of the geometric characteristic judgment region relative to the broken part in the horizontal direction;
calculating the main shaft angle of the geometric characteristic judgment area, rotating the image of the geometric characteristic judgment area until the main shaft angle of the geometric characteristic judgment area is 0, and detecting the central position of the geometric characteristic judgment area; creating an empty image with the same size as the image of the geometric characteristic determination area, drawing a horizontal line segment with the width of 1 to the image boundary along the direction of the geometric characteristic determination area relative to the broken part by taking the central position of the geometric characteristic determination area as a starting point in the empty image, and obtaining an extended line segment image;
reversely rotating the extension line segment image according to the main shaft angle of the geometric characteristic judgment area to obtain the extension line segment image of the broken part;
step S354: judging the position relation of the extension line; putting the extension line segments of the two broken parts obtained in the step S353 into the same image, detecting and counting the number of connected domains in the image:
if the number of connected domains is 1, then: judging that the two extension line segments are intersected, extracting the intersection point of the two extension line segments, putting the intersection point and a first fracture part in the pattern image to be judged into the same image, and calculating the minimum distance from the intersection point to the first fracture part;
if the number of connected domains is 2, then: calculating the angle difference and the minimum distance of two connected domains in the graph, and if the angle difference of the two connected domains is smaller than a preset threshold value E10And the minimum distance is less than a predetermined threshold E11Judging that the directions of the two extension line segments are consistent;
step S355: judging the geometric characteristics for the second time; performing median filtering on the fracture pattern image to be determined, smoothing the edge of a connected domain in the image, repeating the steps S351-S354, and determining the geometric characteristics again;
integrating the judgment results of the two geometric characteristics; if at least one intersection of the two extension line segments is determined in the two determinations, the two extension line segments are intersected; if the directions of the two extension line sections are consistent at least once in the two determinations, the directions of the two extension line sections are consistent;
step S36: judging broken patterns; if the geometric characteristic judgment result of the to-be-judged fracture pattern image meets one of the following conditions, the image is the fracture pattern image: condition 1: the directions of the extension line sections of the broken parts are consistent; condition 2: the extension line segments of the breaking parts are intersected, and the distance between the intersection point and the first breaking part is the minimum;
if the same to-be-determined fracture pattern image set contains to-be-determined fracture pattern images respectively meeting the conditions 1 and 2, the to-be-determined fracture pattern image meeting the condition 1 is a fracture pattern image;
adding a fracture pattern image to a fracture pattern image set WjWherein j represents the identity of the broken block,
Figure RE-GDA0003029841690000161
z represents the number of breaking patterns in each breaking block.
Step S4: predicting the pattern fracture area; the pattern fracture region refers to a portion where a pattern is lost by fracture. The method also comprises the following steps for predicting the pattern fracture area:
step S41: traversing the fracture pattern image set W corresponding to each pattern fracture block1,W2,…,WJFor the broken pattern image therein
Figure RE-GDA0003029841690000162
The following steps S42-S44 are executed, where Z is 1,2, …, Z, and a breaking region prediction image set M of each pattern breaking block is formed1,M2,…,MJWhereinj represents the identity of the broken block,
Figure RE-GDA0003029841690000163
step S42: fracture region prediction based on global appearance features;
presetting a similarity parameter beta, and finding a fracture block R in the shoe print binary imagejThe N similar areas with the similarity larger than beta are marked as
Figure RE-GDA0003029841690000171
Fusing the N similar areas to obtain a fracture pattern image
Figure RE-GDA0003029841690000172
Fracture region predicted image based on global apparent features
Figure RE-GDA0003029841690000173
The fused image is:
Figure RE-GDA0003029841690000174
wherein wiComprises the following steps:
Figure RE-GDA0003029841690000175
wherein
Figure RE-GDA0003029841690000176
Is a distance weight, wdiIs a broken block RjSimilar region
Figure RE-GDA0003029841690000177
The center distance of (d);
Figure RE-GDA0003029841690000178
is a broken block RjSimilar region
Figure RE-GDA0003029841690000179
The similarity of (2);
step S43: fracture region prediction based on local geometric features;
step S431: fitting a curve based on the skeleton line; detecting the broken pattern image
Figure RE-GDA00030298416900001710
And thinning the connected domain, and extracting the skeleton line of the connected domain. If the geometrical characteristics of the two connected domains of the first fracture part determine that the extension line segments of the fracture part are not intersected and the directions of the extension line segments of the fracture part are not consistent, the fracture pattern image is subjected to
Figure RE-GDA00030298416900001711
And thinning the median after filtering.
And performing curve fitting on the skeleton line, and drawing a curve image according to a fitting result. Wherein the order of the fit is adaptively selected. And traversing the fitting orders from 1 to 10, calculating the number of coincident points of each order fitting curve and the skeleton line and the mean value thereof, and taking the order corresponding to the coincident point closest to the mean value as the final fitting order.
Pattern image if broken
Figure RE-GDA00030298416900001712
In the fracture discrimination based on projection, the projection histogram of the fracture discrimination has 0 and only one 1 after processing, and the angles of two connected domains are both larger than 80 degrees or the angle difference is larger than 60 degrees, then the fracture pattern image is subjected to angle theta
Figure RE-GDA00030298416900001713
The rotation is performed to enable curve fitting based on the skeleton line.
Step S432: preliminary prediction of a fracture area; the curve image and the fracture pattern image are combined
Figure RE-GDA00030298416900001714
Put into the sameAnd (3) performing expansion operation on the curve segment which is not overlapped with the connected domain by using the circular structural element to obtain a primary prediction result of the fracture region. For the size of the structural element, firstly, the stroke width transformation algorithm is utilized to obtain the width values of each point in two connected domains, and then the median value of the width values of the connected domains with smaller maximum width is used as the diameter of the circular structural element.
Step S433: correcting the preliminary prediction result of the fracture area; performing edge detection on the preliminary prediction result of the fracture area, counting the number of connected domains in an edge image, and performing the following operations according to the number of the connected domains:
if the number of connected domains is 2, then: and respectively carrying out curve fitting on the two edge segments in the edge image.
If the number of connected domains is 1, then: and respectively extracting edge sections on two sides of the primary prediction result of the fracture area by using a Sobel operator, and then respectively carrying out curve fitting on the two extracted edge sections.
The fitting order selection method of curve fitting is as described in step S431.
Respectively drawing curve images according to the curve fitting result, putting the curve images and the fracture area images into the same image, filling holes formed after the curve segments and the communication areas are combined, and obtaining fracture pattern images
Figure RE-GDA0003029841690000181
Local geometric structural feature-based prediction image of fracture region
Figure RE-GDA0003029841690000182
Step S44: fusing the prediction results of the fracture area; firstly, predicting the image of the fracture area
Figure RE-GDA0003029841690000183
And
Figure RE-GDA0003029841690000184
respectively with broken pattern images
Figure RE-GDA0003029841690000185
Subtracting to obtain a predicted image of the fracture region
Figure RE-GDA0003029841690000186
And
Figure RE-GDA0003029841690000187
corresponding two difference images
Figure RE-GDA0003029841690000188
And
Figure RE-GDA0003029841690000189
then detecting the image
Figure RE-GDA00030298416900001810
And
Figure RE-GDA00030298416900001811
of the connected component, removing
Figure RE-GDA00030298416900001812
And
Figure RE-GDA00030298416900001813
there is no connected domain of intersection, finally the image is processed
Figure RE-GDA00030298416900001814
As images of broken patterns
Figure RE-GDA00030298416900001815
Predicted image of the fracture region of (1), is recorded as
Figure RE-GDA00030298416900001816
And added to the set of predicted images M of the fracture regionjPerforming the following steps; rotated fracture region prediction image
Figure RE-GDA00030298416900001817
Require reverse rotationAnd then the operation is carried out.
Step S5: the pattern fracture area is described. The following steps are also provided for the pattern breaking area description:
step S51: traversing each fracture area prediction image set M1,M2,…,MJPredicting the image of the broken region
Figure RE-GDA00030298416900001818
Performing the following steps S52-S53, Z being 1,2, …, Z, to realize the boundary description of the pattern fracture area;
step S52: detecting the predicted image of the fracture region
Figure RE-GDA00030298416900001819
The connected domain in the system is used for executing the following operation on each connected domain and extracting the connected domain to be described; detecting the external rectangular area of the connected domain, and counting the images of the fracture area
Figure RE-GDA00030298416900001820
If the number of the connected domains in the edge expanding region of the same region is 1, removing the breakage region predicted image
Figure RE-GDA00030298416900001821
The connected domain corresponding to the region; the final residual connected domain is a prediction image of a fracture region
Figure RE-GDA0003029841690000191
The connected domain to be described in (1);
step S53: predicting images of fracture regions
Figure RE-GDA0003029841690000192
The connected domain to be described in the method carries out edge detection, extracts the outline of the connected domain to be described and realizes the fracture area
Figure RE-GDA0003029841690000193
The boundary description of (1).
As a preferred embodiment, the present invention further comprises a system for detecting and describing shoe print pattern breakage characteristics in a shoe print image, comprising: the device comprises an image preprocessing module, a pattern fracture block detection module, a fracture pattern detection module, a pattern fracture region prediction module and a pattern fracture region description module;
the image preprocessing module is used for removing noise in the image by performing graying and binarization on the shoe print image and using morphological processing; dividing the shoe print image into image blocks with fixed sizes and non-overlapping, screening pattern fracture candidate blocks and sending the pattern fracture candidate blocks to the pattern fracture block detection module;
the pattern fracture block detection module is used for judging suspected fracture parts of the connected domains according to the forms of the connected domains in the pattern fracture candidate blocks, detecting whether the pattern fracture candidate blocks contain the suspected fracture parts or not, and screening the pattern fracture blocks from the pattern fracture candidate blocks;
the broken pattern detection module is used for detecting broken patterns in the pattern broken blocks;
the pattern fracture area prediction module is used for predicting fracture areas in the pattern fracture blocks;
a pattern fracture area description module for performing outline description on the pattern fracture area
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments. In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments. In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A method for detecting and describing shoe print pattern fracture characteristics in shoe print images is characterized by comprising the following steps:
s1: preprocessing the input shoe print image;
s2: detecting the pattern fracture blocks on the input shoe print image; the pattern fracture block is an image block containing a fracture pattern;
s3: detecting the broken patterns;
s4: predicting the pattern fracture area;
s5: the pattern fracture area is described.
2. The method for detecting and describing shoe pattern breakage characteristics in shoe print images as claimed in claim 1, wherein the preprocessing of the shoe print images further comprises the steps of:
s11: converting the input shoe print image into a gray level image, and performing binarization processing on the gray level image to obtain a shoe print binary image;
s12: carrying out corrosion and expansion morphological processing on the shoe printing binary image;
s13: dividing the shoe print binary image into a plurality of image blocks with fixed sizes and without overlapping, respectively calculating the entropy of each image block, and enabling the entropy to be larger than a preset threshold value E1The image blocks are put into a pattern fracture candidate block set C, and the mark is C ═ C1,c2,…,cv,…,cV}; where V represents the number of pattern break candidates.
3. The method for detecting and describing shoe print pattern breaking characteristics in shoe print images according to claim 1, wherein the step of detecting the pattern breaking blocks on the input shoe print images further comprises the following steps:
s21: traversing each pattern break candidate block cvV1, 2, …, V, the following steps S22-S26 are performed to form a pattern breaking block set R ═ R1,R2,…,Rj,…,RJAnd its corresponding broken component set Gj,j＝1,2,…,J；
S22: detecting the pattern fracture candidate block cvForm a set of connected domains
Figure RE-RE-FDA0003029841680000011
Wherein v represents a fracture candidate block identifier and T represents the number of connected domains;
s23: judging the shape complexity of the connected domain;
calculating the ratio of the area of each connected region to the smallest convex polygon if the ratio is larger than a predetermined threshold value E2Then marking the connected domain as a common connected domain; if the area ratio is less than or equal to a predetermined threshold E2Then marking the connected domain as a complex connected domain;
s24: performing fracture component judgment on the common communication domain;
s25: performing fracture component judgment on the complex connected domain;
s26: and judging the pattern fracture block.
4. The method for detecting and describing shoe print pattern fracture characteristics in a shoe print image according to claim 1, wherein the fracture component judgment is carried out on the common connected domain:
s241: traversing each of the common connected domains;
s242: respectively calculating the minimum distances from four vertexes of the minimum circumscribed rectangle of the common connected domain to the image boundary of the pattern fracture candidate block, and enabling the minimum distances to be smaller than a preset threshold value E3The vertices of (a) are marked as vertices near the image boundary; calculating the number of the top points close to the image boundary and recording as eta;
s243: calculating the length difference between the long axis and the short axis of the common connected domain, and recording as rho;
s244: calculating the maximum overlapping length of the common connected domain and the pattern fracture candidate block image boundary, and recording as q;
s245: calculating the ratio of the length of the long axis of the common connected domain to the maximum overlapping length q, and recording as mu;
s246: the number of vertices η that will be met is less than a predetermined threshold E4And the length difference rho between the major axis and the minor axis is larger than a predetermined threshold value E5And the ratio mu of the length of the long axis to the maximum overlapping length q is larger than a predetermined threshold value E6The common connected domain of (a) is labeled as a suspected fracture component.
5. The method for detecting and describing shoe print pattern fracture characteristics in a shoe print image according to claim 1, wherein the complex connected domain is subjected to fracture part judgment:
s251: traversing each of the complex connected domains;
s252: thinning the complex connected domain, and extracting a skeleton image of the complex connected domain; points, through which 3 skeleton lines pass, in the skeleton image are skeleton branch points; detecting a skeleton branch point in the skeleton image and judging whether the connected domain is detachable;
if a skeleton branch point is detected in the skeleton line image, the number of skeleton points in a 5 × 5 neighborhood with the skeleton branch point as the center is calculated, and if the number is greater than a threshold value E7Marking the complex connected domain as a detachable connected domain; if no skeleton branch point is detected in the skeleton line image, extracting the outer contour of the complex connected domain, carrying out Hough transform straight line detection on the outer contour of the complex connected domain, calculating the angle between the detected straight line segment and the horizontal direction and the angle difference between the two straight line segments, and if the angle difference between the two straight line segments is 80-90 degrees, marking the connected domain as a detachable connected domain; if the angle difference of no two straight line segments is 80-90 degrees, marking as an undetachable connected domain;
s253: splitting the splittable connected domain into a plurality of sub-connected domains;
s254: judging a component suspected of being broken;
for the detachable connected domain, executing steps S242-246 on the detached sub-connected domain, and marking the sub-connected domain meeting the judgment condition as a suspected fracture part;
for the undetachable connected domain, comparing the size of the circumscribed rectangle of the undetachable connected domain with the size of the pattern fracture candidate block, and if the height difference or the width difference between the circumscribed rectangle and the image block is smaller than a preset threshold value E8Then the connected domain is marked as a suspected fractured component.
6. The method for detecting and describing shoe print pattern fracture characteristics in a shoe print image according to claim 1, wherein in the process of judging the pattern fracture block:
if the pattern fracture candidate block contains two or more suspected fracture parts, the pattern fracture candidate block is a pattern fracture block and is added into a pattern fracture block set R, and simultaneously the suspected fracture parts corresponding to the fracture block are added into a fracture part set GjWherein j represents the identity of the broken block,
Figure RE-RE-FDA0003029841680000031
k represents the number of suspected broken parts corresponding to the block.
7. The method for detecting and describing breaking characteristics of shoe printing patterns in shoe printing images according to claim 1, wherein the breaking patterns are detected by the following steps:
s31: traversing the corresponding broken part set G of each pattern broken blockjThen, steps S32 to S33 are executed, and J is 1,2, …, J, so as to form a fracture pattern image set W corresponding to each fracture block1,W2,…,WJ；
S32: assembling the broken parts GjThe fracture parts in the method are combined to form a fracture pattern image set to be judged, and k is preset to be 1;
s321: to-be-broken member
Figure RE-RE-FDA0003029841680000032
And
Figure RE-RE-FDA0003029841680000033
k +1, K +2, …, K each forming a pattern image to be determined
Figure RE-RE-FDA0003029841680000034
u-1, 2, …, K-K, wherein
Figure RE-RE-FDA0003029841680000035
Is marked as a first broken part,
Figure RE-RE-FDA0003029841680000036
recording as a second fracture part, adding to a to-be-determined fracture pattern image set FkIn (1),
Figure RE-RE-FDA0003029841680000041
wherein k is a broken part identifier;
s322: making K equal to K +1, if K is less than or equal to K-1, repeating the step S321, otherwise, stopping the execution;
s323: acquiring the fracture pattern image set F to be judged1,F2,…,Fk,…,FK-1；
S33: traversing the to-be-determined fracture pattern image set F1,F2,…,FK-1For the pattern image to be judged in the set
Figure RE-RE-FDA0003029841680000042
Performing the following steps S34-S36, u ═ 1,2, …, K-K;
s34: judging the fracture based on the vertical projection of the fracture pattern image to be judged;
s341: vertically projecting the to-be-determined fracture pattern image, replacing a non-0 value in a projection histogram with 1, and then representing continuous 0 or 1 by using single 0 or 1; calculating the number of peaks in the projection represented by the number of 1;
s342: calculating the overlapping range and the main shaft angle of the projections of the two connected domains in the pattern image to be judged; step S341 is executed again after the to-be-determined fracture pattern image meeting one of the following conditions is rotated, and then step S343 is executed, otherwise, step S343 is executed directly, and the angle of image rotation is the angle theta between the connection line of the central points of the two connected domains and the horizontal direction;
a. the projection histogram of the pattern image to be determined only contains 1 after processing, and the overlapping range of the projections of the two connected domains is smaller than a preset threshold value E9；
b. The projection histogram of the fracture pattern image to be judged comprises 0 and 1 after processing, wherein the number of 1 is 1, and one of the following conditions is satisfied:
condition 1: the overlapping range of the two connected domain projections is less than a predetermined threshold value E9；
Condition 2: the angles of the two communicating areas are both larger than 80 degrees or the angle difference is larger than 60 degrees;
s343: if the projection of the to-be-determined breaking pattern image has two or more peaks, the image meets the breaking determination condition based on the vertical projection, and the step S35 is executed on the image, otherwise, the step S34 is executed on other to-be-determined breaking pattern images in the to-be-determined breaking pattern image set;
s35: determining fracture patterns based on geometric characteristics;
s351: detecting two connected domains in the pattern image to be judged, and acquiring a broken part image only containing a single broken part;
s352: determining a geometric characteristic judgment area of the fractured part; thinning the connected domain, and extracting a skeleton line of the connected domain; respectively extracting 4 continuous skeleton points taking end points at two ends of a skeleton line as starting points from the skeleton line image, and performing expansion operation on the continuous skeleton points by using circular structural elements; respectively putting the two expansion results and the other fracture part in the image of the fracture pattern to be judged into the same image, calculating and comparing the minimum distance between the two expansion results and the fracture part, and taking the expansion result with smaller minimum distance as a geometric characteristic judgment area of the fracture part;
placing the geometric characteristic judgment area into an empty image, and acquiring a geometric characteristic judgment area image corresponding to the broken part;
the calculation method of the minimum distance between the two areas comprises the following steps: respectively calculating the minimum value of the distance from one region to the nearest point of the other region, wherein the formula is as follows:
Figure RE-RE-FDA0003029841680000051
Figure RE-RE-FDA0003029841680000052
wherein A and B correspond to two regions respectively, a and B are points in the regions A and B, and d (a and B) is the distance between the two points; the minimum distance between the two regions is:
H(A,B)＝min{h(A,B),h(B,A)}；
s353: acquiring a broken part extension line segment; calculating the angle of a main shaft of a connected domain in the image of the broken part, rotating the image of the broken part and the image of the corresponding geometric characteristic judgment region until the angle of the main shaft of the connected domain is 0, then detecting the central positions of the broken part and the geometric characteristic judgment region, and judging the position of the geometric characteristic judgment region relative to the broken part in the horizontal direction;
calculating the main shaft angle of the geometric characteristic judgment area, rotating the image of the geometric characteristic judgment area until the main shaft angle of the geometric characteristic judgment area is 0, and detecting the central position of the geometric characteristic judgment area; creating an empty image with the same size as the image of the geometric characteristic determination area, drawing a horizontal line segment with the width of 1 to the image boundary along the direction of the geometric characteristic determination area relative to the broken part by taking the central position of the geometric characteristic determination area as a starting point in the empty image, and obtaining an extended line segment image;
reversely rotating the extension line segment image according to the main shaft angle of the geometric characteristic judgment area to obtain the extension line segment image of the broken part;
s354: judging the position relation of the extension line; putting the extension line segments of the two broken parts obtained in the step S353 into the same image, detecting and counting the number of connected domains in the image:
if the number of connected domains is 1, then: judging that the two extension line segments are intersected, extracting the intersection point of the two extension line segments, putting the intersection point and a first fracture part in the pattern image to be judged into the same image, and calculating the minimum distance from the intersection point to the first fracture part;
if the number of connected domains is 2, then: calculating the angle difference and the minimum distance of two connected domains in the graph, and if the angle difference of the two connected domains is smaller than a preset threshold value E10And the minimum distance is less than a predetermined threshold E11Judging that the directions of the two extension line segments are consistent;
s355: judging the geometric characteristics for the second time; performing median filtering on the fracture pattern image to be determined, smoothing the edge of a connected domain in the image, repeating the steps S351-S354, and determining the geometric characteristics again;
integrating the judgment results of the two geometric characteristics; if at least one intersection of the two extension line segments is determined in the two determinations, the two extension line segments are intersected; if the directions of the two extension line sections are consistent at least once in the two determinations, the directions of the two extension line sections are consistent;
s36: judging broken patterns; if the geometric characteristic judgment result of the to-be-judged fracture pattern image meets one of the following conditions, the image is the fracture pattern image: condition 1: the directions of the extension line sections of the broken parts are consistent; condition 2: the extension line segments of the breaking parts are intersected, and the distance between the intersection point and the first breaking part is the minimum;
if the same to-be-determined fracture pattern image set contains to-be-determined fracture pattern images respectively meeting the conditions 1 and 2, the to-be-determined fracture pattern image meeting the condition 1 is a fracture pattern image;
adding a fracture pattern image to a fracture pattern image set WjWherein j represents the identity of the broken block,
Figure RE-RE-FDA0003029841680000061
z represents the number of breaking patterns in each breaking block。
8. A method for detecting and describing shoe print pattern break features in shoe print images according to claim 1, wherein the following steps are further provided for pattern break area prediction:
s41: traversing the fracture pattern image set W corresponding to each pattern fracture block1,W2,…,WJFor the broken pattern image therein
Figure RE-RE-FDA0003029841680000062
The following steps S42-S44 are executed, where Z is 1,2, …, Z, and a breaking region prediction image set M of each pattern breaking block is formed1,M2,…,MJWhere j represents the identity of the broken block,
Figure RE-RE-FDA0003029841680000063
s42: fracture region prediction based on global appearance features;
presetting a similarity parameter beta, and finding a fracture block R in the shoe print binary imagejThe N similar areas with the similarity larger than beta are marked as
Figure RE-RE-FDA0003029841680000064
Fusing the N similar areas to obtain a fracture pattern image
Figure RE-RE-FDA0003029841680000071
Fracture region predicted image based on global apparent features
Figure RE-RE-FDA0003029841680000072
The fused image is:
Figure RE-RE-FDA0003029841680000073
wherein wiComprises the following steps:
Figure RE-RE-FDA0003029841680000074
wherein
Figure RE-RE-FDA0003029841680000075
Is a distance weight, wdiIs a broken block RjSimilar region
Figure RE-RE-FDA0003029841680000076
The center distance of (d);
Figure RE-RE-FDA0003029841680000077
is a broken block RjSimilar region
Figure RE-RE-FDA0003029841680000078
The similarity of (2);
s43: fracture region prediction based on local geometric features;
s44: fusing the prediction results of the fracture area; firstly, predicting the image of the fracture area
Figure RE-RE-FDA0003029841680000079
And
Figure RE-RE-FDA00030298416800000710
respectively with broken pattern images
Figure RE-RE-FDA00030298416800000711
Subtracting to obtain a predicted image of the fracture region
Figure RE-RE-FDA00030298416800000712
And
Figure RE-RE-FDA00030298416800000713
corresponding two difference images
Figure RE-RE-FDA00030298416800000714
And
Figure RE-RE-FDA00030298416800000715
then detecting the image
Figure RE-RE-FDA00030298416800000716
And
Figure RE-RE-FDA00030298416800000717
of the connected component, removing
Figure RE-RE-FDA00030298416800000718
And
Figure RE-RE-FDA00030298416800000719
there is no connected domain of intersection, finally the image is processed
Figure RE-RE-FDA00030298416800000720
As images of broken patterns
Figure RE-RE-FDA00030298416800000721
Predicted image of the fracture region of (1), is recorded as
Figure RE-RE-FDA00030298416800000722
And added to the set of predicted images M of the fracture regionjPerforming the following steps; rotated fracture region prediction image
Figure RE-RE-FDA00030298416800000723
The above operation is performed after the reverse rotation is required.
9. A method for detecting and describing shoe print pattern breaking characteristics in shoe print images according to claim 1, characterized in that the pattern breaking area description further comprises the following steps:
s51: traversing each fracture area prediction image set M1,M2,…,MJPredicting the image of the broken region
Figure RE-RE-FDA00030298416800000724
Performing the following steps S52-S53, Z being 1,2, …, Z, to realize the boundary description of the pattern fracture area;
s52: detecting the predicted image of the fracture region
Figure RE-RE-FDA00030298416800000725
The connected domain in the system is used for executing the following operation on each connected domain and extracting the connected domain to be described; detecting the external rectangular area of the connected domain, and counting the images of the fracture area
Figure RE-RE-FDA00030298416800000726
If the number of the connected domains in the edge expanding region of the same region is 1, removing the breakage region predicted image
Figure RE-RE-FDA00030298416800000727
The connected domain corresponding to the region; the final residual connected domain is a prediction image of a fracture region
Figure RE-RE-FDA00030298416800000728
The connected domain to be described in (1);
s53: predicting images of fracture regions
Figure RE-RE-FDA0003029841680000081
The connected domain to be described in the method carries out edge detection, extracts the outline of the connected domain to be described and realizes the fracture area
Figure RE-RE-FDA0003029841680000082
The boundary description of (1).
10. A system for detecting and characterizing shoe pattern breaks in an image of a shoe print using the method of claims 1-9, comprising: the device comprises an image preprocessing module, a pattern fracture block detection module, a fracture pattern detection module, a pattern fracture region prediction module and a pattern fracture region description module;
the image preprocessing module is used for removing noise in the image by performing graying and binarization on the shoe print image and using morphological processing; dividing the shoe print image into image blocks with fixed sizes and non-overlapping, screening pattern fracture candidate blocks and sending the pattern fracture candidate blocks to the pattern fracture block detection module;
the pattern fracture block detection module is used for judging suspected fracture parts of the connected domains according to the forms of the connected domains in the pattern fracture candidate blocks, detecting whether the pattern fracture candidate blocks contain the suspected fracture parts or not, and screening the pattern fracture blocks from the pattern fracture candidate blocks;
the broken pattern detection module is used for detecting broken patterns in the pattern broken blocks;
the pattern fracture area prediction module is used for predicting fracture areas in the pattern fracture blocks;
and the pattern fracture area description module is used for describing the outline of the pattern fracture area.
CN202110225954.2A 2021-03-01 2021-03-01 Method and system for detecting and describing breaking characteristics of shoe print patterns Active CN112907549B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110225954.2A CN112907549B (en) 2021-03-01 2021-03-01 Method and system for detecting and describing breaking characteristics of shoe print patterns

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110225954.2A CN112907549B (en) 2021-03-01 2021-03-01 Method and system for detecting and describing breaking characteristics of shoe print patterns

Publications (2)

Publication Number Publication Date
CN112907549A true CN112907549A (en) 2021-06-04
CN112907549B CN112907549B (en) 2023-09-05

Family

ID=76106860

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110225954.2A Active CN112907549B (en) 2021-03-01 2021-03-01 Method and system for detecting and describing breaking characteristics of shoe print patterns

Country Status (1)

Country Link
CN (1) CN112907549B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105005565A (en) * 2014-04-18 2015-10-28 大连恒锐科技股份有限公司 Onsite sole trace pattern image retrieval method
CN107247929A (en) * 2017-05-26 2017-10-13 大连海事大学 A kind of footwear stamp line progressive refinement formula extracting method of combination priori
US20190312985A1 (en) * 2018-04-06 2019-10-10 Motorola Solutions, Inc. Systems and methods for processing digital image data representing multiple views of an object of interest
GB202007456D0 (en) * 2020-05-19 2020-07-01 Anatomap Ltd Image processing in evidence collection
CN111860500A (en) * 2020-07-10 2020-10-30 大连海事大学 Shoe print wear area detection and edge tracing method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105005565A (en) * 2014-04-18 2015-10-28 大连恒锐科技股份有限公司 Onsite sole trace pattern image retrieval method
CN107247929A (en) * 2017-05-26 2017-10-13 大连海事大学 A kind of footwear stamp line progressive refinement formula extracting method of combination priori
US20190312985A1 (en) * 2018-04-06 2019-10-10 Motorola Solutions, Inc. Systems and methods for processing digital image data representing multiple views of an object of interest
GB202007456D0 (en) * 2020-05-19 2020-07-01 Anatomap Ltd Image processing in evidence collection
CN111860500A (en) * 2020-07-10 2020-10-30 大连海事大学 Shoe print wear area detection and edge tracing method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
黎向阳;吴敏华;施智平;: "基于Gabor变换域的积分直方图鞋印图像检索", 计算机应用与软件, no. 03 *

Also Published As

Publication number Publication date
CN112907549B (en) 2023-09-05

Similar Documents

Publication Publication Date Title
US20220172348A1 (en) Information processing device, information processing method, and storage medium
WO2018107939A1 (en) Edge completeness-based optimal identification method for image segmentation
Patel et al. Automatic number plate recognition system (anpr): A survey
Chiang et al. A survey of digital map processing techniques
EP0551738B1 (en) Method for connected and degraded text preprocessing
CN109685788B (en) Automatic detection method for floor defect image based on morphological characteristics
CN109697717B (en) Lining crack identification method based on image automatic search
Xu Robust traffic sign shape recognition using geometric matching
CN101114335A (en) Full angle rapid fingerprint identification method
CN113177959B (en) QR code real-time extraction method in rapid movement process
JP3830998B2 (en) Ruled line removal method and character recognition apparatus using the same
CN113379739B (en) Ultrasonic image identification method, device, equipment and storage medium
CN109870458B (en) Pavement crack detection and classification method based on three-dimensional laser sensor and bounding box
CN113240623B (en) Pavement disease detection method and device
CN113689445B (en) High-resolution remote sensing building extraction method combining semantic segmentation and edge detection
CN112508913B (en) Cable section edge detection method based on image detection
Davis et al. Automatic license plate detection using vertical edge detection method
JP2008011484A (en) Apparatus and method for extracting character and graphic string, program for executing the method, recording medium with the program stored therein
Adu-Gyamfi et al. Functional evaluation of pavement condition using a complete vision system
Fang et al. Towards real-time crack detection using a deep neural network with a Bayesian fusion algorithm
US7903862B2 (en) Method for detecting geometrical structures in images
CN110378337B (en) Visual input method and system for drawing identification information of metal cutting tool
CN112907549A (en) Shoe printing pattern fracture characteristic detection and description method and system
Raikar et al. Automatic building detection from satellite images using internal gray variance and digital surface model
CN116843626A (en) Machine vision steel wire rope surface defect detection method based on multi-feature fusion

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