KR101722766B1 - Method for arraying shoes sole picture and distinguishing right-left shoes sole based on recognition of shoes sole surface notice - Google Patents

Abstract

Provided is a method for arraying pictures of shoe soles and distinguishing left and right shoe soles based on the recognition of marks on the shoe sole surface. The method for arraying pictures of shoe soles and distinguishing right-left shoe soles based on the recognition of marks on the shoe sole surface, according to the present invention, allows the pictures of shoe soles to be arrayed by the arraying distribution characteristic analysis of the marks in the currently taken pictures of shoe soles, when the inserting directions of the shoe soles are random during an analysis process of the pictures of the shoe soles for a quality test of the shoe soles such as midsoles. Therefore, a user can check dimensions by comparing the dimensions of the shoe soles with the dimensions of a last, which is a criterion of the quality test of the shoe soles, and can distinguish left and right shoe soles.

Description

A method for identifying a sole of a sole and identifying a left or right foot based on a shoe sole surface marking recognition method.

The present invention relates to a shoe sole image alignment based on shoe sole surface marker recognition and a method for identifying left and right foot, and more particularly, to a shoe sole image analysis for quality inspection of a shoe sole such as a midsole, The shoe sole image is aligned by analyzing the array distribution characteristics of the markers in the shoe sole image that is currently photographed, so that it is possible to analyze the size of the shoe sole compared with the last, which is the standard of shoe sole quality inspection, The present invention relates to a shoe sole image alignment based on shoe sole surface marker recognition, and a method for identifying the left and right foot shoes.

Injection Molding (IP) Injection Molding In the case of midsole, compression molding (CMP) Compression and foaming occur simultaneously for a short period of time compared to midsole, resulting in poor dimensional accuracy and large post strain. Therefore, it is essential to check the quality by measuring the dimensions after the process.

Quality inspection through dimensional measurement is carried out in the form of manual inspection of the inspection personnel by visual inspection. Typically, midsole products produced from 4 to 6 injection machines are moved to a single inspection line, with 10 to 20 inspectors per production line. The inspection line is equipped with a model, dimension and color of different midsole products at the same time, and the inspection staff changes the quality inspection method through dimensional measurement according to each model.

As a result, the fatigue of the inspection personnel is increased and various measurement standards are used for each model because there is no uniform measurement principle. Therefore, it is difficult to maintain the consistency of the quality inspection, and the quality inspection efficiency becomes low. In addition, the cost of quality inspection increases.

In order to improve this, a technique for performing dimensional measurement and quality inspection of a shoe sole using a machine vision is being studied as disclosed in Korean Utility Model Registration No. 20-0292152 entitled " Invei e Shoe Midsole Inspection System " . The conventional shoe sole quality inspection technique based on machine vision is a method of checking the quality of shoe sole by measuring the overall dimensions of the shoe sole product. Accordingly, there has been a limit to apply to a midsole made of a soft foamed material, in which elastic deformation is freely, and in particular, deformation of a toe spring, a sidewall, or the like is bent at an arbitrary angle irrespective of the product dimensions. In the measurement of the size of the inspection personnel by visual manual, the worker judges the permissible deformation of the midsole product and the deformation that causes the failure. It is very difficult to quantify the quality of the sole brush quality inspection technique by machine vision Because. In addition, since it is difficult to manage the databaseized dimension information due to the industrial environment of the footwear sector in a small quantity production type of various types, it has been difficult to smoothly and efficiently carry out the quality inspection of various shoe soles by the technique of shoe sole quality inspection by the conventional machine vision . In other words, it was difficult to mechanically recognize various measurement standards for each model. In particular, it was difficult to update / manage the dimensional information of each shoe sole product by databaseization.

In response to this, Patent Application No. 10-2014-0125019 filed by the present applicant "Real-time quality inspection method of shoe sole" is used to detect local surface expansion state information of a shoe sole portion having almost no distortion, The present invention relates to a technique for improving the accuracy and reliability of a quality inspection by measuring a size of a shoe sole, By quantitatively comparing and analyzing the local surface images of the sole set product and the currently molded shoe sole, it is possible to easily and easily perform the determination of the defective shoe sole with high accuracy without the management of database shoe sole dimension information.

On the other hand, since the shape of the shoe sole is formed of a curved surface, there is no reference axis, so it is difficult to catch the reference in shoe sole image analysis, and when the shoe sole is inserted in a direction , It is difficult to compare with the last which is the standard of the shoe sole quality inspection, and in order to always put the shoe sole in a constant direction for comparison with the last, a separate device had to be provided. Also, if the left / right shoe sole is inserted without a separate mark, the left and right feet must be distinguished to make comparison with the last.

delete

delete

Korea Registered Utility Model Registration No. 20-0292152 "Injection eve shoe size inspection system" Patent Application No. 10-2014-0125019 "Real-time Quality Inspection Method of Shoe Soles"

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a shoe sole comprising a plurality of first markers arranged squarely at predetermined intervals on a forefoot of a shoe sole, a plurality of first markers formed by a plurality of heating pin holes arranged non- By analyzing the distribution characteristics of the second markings in the shoe sole image through the clustering algorithm and the deviation, distribution, and center coordinate values detection, the shoe sole image can be aligned and the left and right foot can be classified, Shoe sole for quality inspection of the same shoe sole Even when the direction of shoe sole insertion is not constant, it can be compared with the last shoe sole quality inspection standard. A new form of shoe sole that can be performed Surface marker recognition based shoe sole Image alignment and identification of left and right feet And that is the purpose.

According to an aspect of the present invention for achieving the above object, the present invention is characterized in that a plurality of first markers (11) are arranged in a square arrangement at a predetermined interval, a plurality of The image of the shoe sole 10 composed of the rear portion 10b in which the second mark 12 is arranged in a non-square form with the set pattern is shot by the photographing apparatus 100 and input in real time to the analyzing apparatus 200, A photographing step; A global coordinate system is set in the shoe sole image frame (20) by the coordinate setting module (220) of the analyzer (200); A reference straight line 22 passing through the markers arranged in a line is calculated by the analysis module 230 of the analysis apparatus 200 and the specimen rotation A shoe sole closing direction displacement detecting step in which the angle is detected by the analysis module 230 of the analyzer 200; The analysis module 230 of the analysis apparatus 200 divides the mark in the shoe sole image frame 20 into two groups of the forefoot region 24a and the rear region 24b, It is determined whether or not the front portion 24a and the rear portion 24b are arranged in the longitudinal direction in the shoe sole image frame 20 according to the mark distribution characteristic of the rear portion 24b, And a 180 ° rotation correction value is generated when the forefoot portion (24a) is disposed rearward in the longitudinal direction; The shoe sole image frame 20 is detected by the analyzer 200 (200) by an alignment angle obtained by adding 180 ° rotation correction values that can be generated in the shoe sole back and forth inversion correction step to the sample rotation angle calculated in the shoe sole input direction displacement detection step A shoe sole image frame rotation correcting step in which the shoe sole image frame 20 is rotated by the image aligning module 240 of the shoe sole image frame 20, The alignment mark in the shoe sole image frame 20 is divided into two groups of the front group 25a and the rear group 25b by the analysis module 230 of the analysis apparatus 200, And the left and right feet are discriminated by comparing the x-coordinate values of the center coordinate values of the markers belonging to each of the groups (25b). The shoe sole surface image recognition- Identification methods are provided.

In the shoe sole image alignment and shoe sole image recognition based on the shoe sole surface marker recognition method according to the present invention, the shoe sole image frame global coordinate system setting step includes the steps of setting the global coordinate system of the shoe sole image frame, The center of the rear portion 24b of the shoe sole 10 in which the y-axis direction of the global coordinate system is normally inserted is set in the direction of the straight line passing through the first mark 11 arranged in line in the left- In the direction of the straight line passing through the second mark 12 arranged in a row in the longitudinal direction.

In the shoe sole image alignment and shoe sole image recognition based on shoe sole surface marker recognition according to the present invention, the shoe soles input direction displacement detection step may include a KNN clustering algorithm A plurality of clustering units 21 are formed by dividing the landmarks by groups according to any one direction selected from the x-axis direction and the y-axis direction by a K-Nearest Neighbor clustering algorithm 231 Wow; The reference clustering unit 21 having the largest number of landmarks belonging to the clustering unit 21 and having the smallest deviation of the landmark coordinate values is obtained from the plurality of clustering units 21 by the analysis module 230 of the analysis apparatus 200 Selecting a reference clustering unit to be selected; A reference straight line calculating step in which a reference straight line 22 passing through the mark belonging to the reference clustering unit 21 is calculated by the analysis module 230 of the analyzer 200; The rotation angle of the specimen formed by the reference line 22 and one axis selected from the x-axis and the y-axis in the global coordinate system of the shoe sole image frame 20 is measured by the analysis module 230 of the analyzer 200 And an angle calculating step.

In the shoe sole image alignment and shoe solver identification method based on the shoe sole surface marker recognition method according to the present invention, the shoe sole back and forth inversion correction step may include a KNN clustering algorithm K (k) included in the analysis module 230 of the analyzer 200 The shoe sole markers 2 divided into two groups of the first group 23a and the second group 23b on the basis of the density of the markings in the shoe sole image frame 20 by the Neighbor Neighbor clustering algorithm 231 Dog groups; Group variance value calculating step of calculating an x-axis variance value of the markers belonging to each of the first group 23a and the second group 23b; A full-length zone designation step of designating a group having a large variance value among the first group 23a and the second group 23b as a full-length zone 24a and a group having a small dispersion value as a rear zone 24b; The average of the y-axis coordinate values of the markers belonging to the group designated as the full-length zone 24a and the average of the y-axis coordinate values of the marks belonging to the group designated as the rear zone 24b are transmitted to the analysis module 230 of the analyzer 200 And when the average of the y-axis coordinate values of the markers belonging to the group designated as the fore part 24a is smaller than the average of the y-axis coordinate values of the markers belonging to the group designated as the rear zone 24b, And a 180 ° rotation correction value grant decision step of generating a 180 ° rotation correction value.

In the shoe sole image alignment and the left and right foot identification method based on shoe sole surface marker recognition according to the present invention, the left and right foot identification steps are performed using a KNN clustering algorithm (K-Nearest The shoe sole alignment mark is divided into two groups, that is, the front group 25a and the rear group 25b based on the density of the alignment marks in the shoe sole image frame 20 by the neighbor clustering algorithm 231. [ A group segmentation step; A center coordinate value of each of the front and rear groups of the front group 25a and the rear group 25b is calculated by the analysis module 230 of the analyzer 200; The analysis module 230 of the analysis apparatus 200 compares the coordinate values of the x-axis of the center coordinate values calculated in the center coordinate value calculation step of the pre / tributary group to determine the left and right contingencies, When the x-axis coordinate value of the center coordinate value is larger than the rear group 25b, the left and right feet are discriminated. When the x-axis coordinate value of the center coordinate value is larger than the forefoot group 25b in the front group 25a, And a determination step.

In the shoe sole image alignment and shoe soles identification method based on the shoe sole surface marker recognition method according to the present invention, the plurality of clustering unit calculation steps may include a first marker 11 and a second marker 12, And the specimen rotation angle calculating step calculates the specimen rotation angle of the shoe sole image frame 20 based on the rotation of the specimen rotated by the x-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22 So that the angle can be calculated by the analysis module 230 of the analysis apparatus 200.

Alternatively, the calculating of the plurality of clustering units may be performed such that the widthwise clustering in which the plurality of clustering units 21 are formed while the second markers 12 are divided in groups along the x-axis direction, The analysis module 230 of the analysis apparatus 200 can calculate the rotation angle of the specimen formed by the y-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22.

According to the shoe sole image alignment and the left and right foot identification method based on shoe sole surface marker recognition according to the present invention, the shoe sole image is shoe sole quality inspection It is possible to perform dimensional analysis as compared with the reference last, and it is also possible to divide the left and right feet. It is possible to compare the shoe sole brush for quality inspection of shoe sole such as midsole so that it can be compared with last which is the standard of shoe sole brush quality inspection even if the direction of shoe sole is not constant. Is accurately and smoothly performed.

Even if the marking of the square array is displayed on the forefoot portion of the forming mold of the midsole, the effect of the present invention that can be applied to the IP midsole can be represented by the fact that the linear alignment pattern In the midsole molded product is slightly disturbed and the straight line angle between the marks becomes different. Therefore, when the reference alignment for dimensional inspection is performed using a straight line connecting the center points of each mark arranged in a square, or when the reference alignment for dimensional inspection is performed using a separate origin mark, Lt; / RTI > In addition, separate image processing for finding the origin markers is necessary for sorting using separate origin markers, which requires a large amount of resources and time, thereby increasing the inspection time.

According to the present invention, there is provided a shoe sole image alignment and shoe soles identification method based on shoe sole surface marker recognition, which solves such a problem. The alignment process using the coordinates of the markers obtained through a single image processing, And even if the actual clustered points do not form a straight line, the average straight line of the group of densely populated points hardly changes, so that the accuracy of alignment can be increased.

1 is a block diagram of a method of shoe sole image alignment and left and right foot identification based on shoe sole surface marker recognition according to an embodiment of the present invention;
FIG. 2 is a block diagram of an analysis apparatus applied to a method of shoe sole image alignment and left and right foot identification based on shoe sole surface marker recognition according to an embodiment of the present invention;
3 is an illustration of a shoe sole image frame in shoe sole image alignment and left and right foot identification based on shoe sole surface marker recognition according to an embodiment of the present invention;
4 is a diagram illustrating a step of setting a shoe sole image frame global coordinate system according to an embodiment of the present invention;
5 is a sequence block diagram of a shoe sole input direction displacement detecting step according to an embodiment of the present invention;
FIG. 6 is a diagram illustrating a step of calculating a plurality of clustering units by longitudinal clustering in the step of detecting the shoe sole insertion direction displacement according to the embodiment of the present invention; FIG.
FIG. 7 is a view showing a reference clustering unit selecting step, a reference straight line calculating step, and a sample rotation angle calculating step of the shoe sole insertion direction displacement detecting step according to the embodiment of the present invention by longitudinal clustering;
8 is a diagram illustrating a step of calculating a plurality of clustering units by widthwise clustering in the shoe sole insertion direction displacement detecting step according to the embodiment of the present invention;
FIG. 9 is a view showing a reference clustering unit selecting step, a reference straight line calculating step, and a sample rotation angle calculating step of the shoe soles insertion direction displacement detecting step according to the embodiment of the present invention by the width direction clustering;
FIG. 10 is a sequence block diagram of a shoe sole before and after reversal correction step according to an embodiment of the present invention; FIG.
11 is a view showing a step of correcting the inverse of the shoe sole according to the embodiment of the present invention;
FIG. 12 is a view illustrating an example of a shoe sole screen frame alignment according to an embodiment of the present invention; FIG.
FIG. 13 is a sequence block diagram of a left and right foot identifying step according to an embodiment of the present invention, according to an embodiment of the present invention; FIG.
FIG. 14 is a view illustrating a left and right foot identification step according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings 1 to 14. Meanwhile, in the drawings and detailed description, a general shoe sole, a midsole, a shoe soles pre-shoe, a shoe soles rear shoe, a shoe sole rear heating hole, a global coordinate system, an image alignment, a K-nearest neighbor clustering algorithm To simplify or omit the illustration and the description of the constitution and operation that can be easily understood by those skilled in the art. In the drawings and specification, there are shown in the drawings and will not be described in detail, and only the technical features related to the present invention are shown or described only briefly. Respectively.

The shoe sole image alignment based on shoe sole surface marker recognition according to an exemplary embodiment of the present invention includes steps of shoe sole imaging, shoe sole image frame global coordinate system setting, shoe sole input direction displacement detection step , A shoe sole back and forth reversing correction step, a shoe sole image frame rotation correction step, and a left and right foot identification step.

The shoe sole image photographing step includes a front part 10a in which a plurality of first marks 11 are arranged squarely at a set interval and a plurality of second marks 12 formed in a plurality of heating pin holes 12a, An image of the shoe sole 10 made up of a square arrayed rear region 10b is taken by the photographing apparatus 100 and input to the analyzing apparatus 200 in real time as shown in FIG. The shoe sole 10 according to the embodiment of the present invention is not limited to the midsole.

The image of the shoe sole 10 input in real time to the analyzer 200 is stored in the memory 210 as a shoe sole image frame 20. [

The plurality of first marks 11 are horizontally / vertically arranged squarely in the front part 10a having a large markable area as shown in Fig. The plurality of heating pin holes 12a used as the second marking 12 are arranged in a non-square manner in the rear region 10b since they are holes for molding. The pin holes 12a are vertically aligned.

The shoe sole image frame global coordinate system setting step is a step in which the global coordinate system is set in the shoe sole image frame 20 by the coordinate setting module 220 of the analyzer 200 as shown in FIG. The step of setting the shoe sole image frame global coordinate system according to the embodiment of the present invention includes the steps of arranging the first marker 11 aligned in the left and right direction on the fore part 24a of the shoe sole 10 in which the x- And a straight line passing through the second mark 12 arranged in the longitudinal direction in the center of the rear region 24b of the shoe sole 10 in which the y-axis direction of the global coordinate system is normally entered, As shown in FIG.

The shoe sole insertion direction displacement detection step is performed in such a manner that a reference straight line 22 passing through the markers arranged in a row is calculated by the analysis module 230 of the analysis apparatus 200 and the global coordinate system of the shoe sole image frame 20, And the test piece rotation angle formed by the straight line 22 is detected by the analysis module 230 of the analyzer 200.

Here, the shoe sole insertion direction displacement detecting step according to the embodiment of the present invention is performed through a plurality of clustering unit calculating step, a reference clustering unit selecting step, a reference straight line calculating step, and a sample rotation angle calculating step as shown in FIG.

The multiple clustering unit calculation step may be performed by a KNN clustering algorithm 231 included in the analysis module 230 of the analysis apparatus 200 so that the landmark is selected from any one of x- And a plurality of clustering units 21 are formed while being divided by groups on the basis of the direction. 6, the longitudinal clustering in which the plurality of clustering units 21 are formed while the first landmarks 11 and the second landmarks 12 are divided along the y axis direction .

The reference clustering unit selecting step selects the reference clustering unit 21 having the largest number of landmarks belonging to the clustering unit 21 and having the smallest deviation of the landmark coordinate values from the plurality of clustering units 21 Is selected by the module 230. In the reference clustering unit selecting step after the multiple clustering unit calculating step by the longitudinal clustering, the reference clustering unit 21 belongs to the fore part 24a. At this time, the widest part of the left and right width is selected as the reference clustering unit 21 do.

The reference straight line calculating step is a step in which the reference straight line 22 passing through the marker belonging to the reference clustering unit 21 is calculated by the analysis module 230 of the analysis apparatus 200 as shown in FIG. Here, the markers through which the reference straight line 22 passes may not be straight lines.

The sample rotation angle calculating step may be performed such that the sample rotation angle formed by one of the x-axis and the y-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22 is determined by the analysis module 230 of the analyzer 200, , The x-axis and the reference straight line 22 of the global coordinate system of the shoe sole image frame 20 are calculated as shown in Fig. 7 in the sample rotation angle calculating step after the multiple clustering unit calculating step by longitudinal clustering And the analysis module 230 of the analysis apparatus 200 calculates the rotation angle of the specimen.

Alternatively, the plurality of clustering unit calculating steps of the shoe sole insertion direction displacement detecting step may include a step of calculating a plurality of clustering units 21 by dividing the second landmarks 12 along the x- Can be performed. In the reference clustering unit selecting step after the multiple clustering unit calculating step by the width direction clustering, the reference clustering unit 21 is included in the rear region 24b. At this time, the central vertical line portion is selected as the reference clustering unit 21 do. 9, the specimen rotation angle calculation step may be performed by the analysis module 230 of the analysis apparatus 200, such that the specimen rotation angle formed by the y-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22 .

The inversion of the shoe sole is performed by the analysis module 230 of the analyzer 200 to divide the mark in the shoe sole image frame 20 into two groups of the front part 24a and the rear part 24b, It is judged whether or not the front part 24a and the rear part 24b are arranged in the longitudinal direction in the shoe sole image frame 20 according to the mark distribution characteristics of the front part 24a and the rear part 24b, And the 180 DEG rotation correction value is generated when the forefoot portion 24b is disposed in the longitudinal direction and the forefoot portion 24a is disposed in the longitudinal direction.

Here, the step of correcting the inverse of the shoe soles according to the embodiment of the present invention may be performed by dividing the shoe soles into two groups, calculating the variance value for the two groups, Deg.] Rotation correction value grant decision step.

The shoe sole-marked two-group dividing step is performed by a K-nearest neighbor clustering algorithm (KNN clustering algorithm) 231 provided in the analysis module 230 of the analyzer 200, The first group 23a and the second group 23b are divided into two groups.

The two-group variance value calculation step is a step of calculating the x-axis variance values of the markers belonging to the first group 23a and the second group 23b.

The forelever-toe rear region designation step is a step of designating the front part 24a as a group having a large variance value among the first group 23a and the second group 23b and the rear group 24b as a group having a small variance value . This is because the front part of the forefoot part 24a is larger in the x-axis direction (width direction) dispersion value than the rear part 24b because of the wide front part due to the characteristics of the shoe shape.

The 180 ° rotation correction value imparting step determines the average of the y-axis coordinate values of the markers belonging to the group designated by the full-length part 24a and the average of the y-axis coordinate values of the marks belonging to the group designated by the rear zone 24b 200), and the average of y-axis coordinate values of the markers belonging to the group designated by the forefoot portion 24a is compared with the average of the y-axis coordinate values of the markers belonging to the group designated by the rear region 24b And generates a 180-degree rotation correction value when it is smaller. 11 (a), when the shoe sole 10 is in a state without front-to-rear reversal, a 180 ° rotation correction value is not given, and as shown in Fig. 11 (b) A 180 ° rotation correction value is given.

The shoe sole image frame rotation correction step may include a step of calculating a shoe sole image frame (20) by an alignment angle obtained by adding a 180 ° rotation correction value that can be generated in the step of correcting the shoe sole before and after the shoe sole to the sample shoe rotation angle calculated in the shoe sole insertion direction displacement detection step Is rotated by the image aligning module 240 of the analyzer 200 and the shoe sole image in the shoe sole image frame 20 is aligned to the set position and the mark is also arranged. Fig. 12 illustrates the alignment of the shoe sole image frame 20.

The left and right foot identification steps are performed by the analysis module 230 of the analysis apparatus 200 such that the alignment mark in the shoe sole image frame 20 is divided into two groups of the front group 25a and the rear group 25b, Axis coordinate values of the center coordinate values of the markers belonging to the rear group 25a and the rear group 25b are compared with each other to discriminate the left foot and the accident.

Here, the left and right foot discrimination step according to the embodiment of the present invention is performed by dividing the shoe soles alignment mark before / after the shoe soles, calculating the central coordinate value for each of the groups, do.

The step of dividing the shoe sole alignment mark before / after toughening is performed by the K-nearest neighbor clustering algorithm (KNN clustering algorithm) 231 provided in the analysis module 230 of the analyzer 200, Is divided into two groups, the front group 25a and the rear group 25b on the basis of the density of the markers.

The calculation of the center coordinate value for each of the former and latter groups is performed by the analysis module 230 of the analysis apparatus 200 in the center coordinate values of the marks belonging to the front group 25a and the rear group 25b.

In the left and right foot determination step, the analysis module 230 of the analysis apparatus 200 compares the x-axis coordinate values of the center coordinate values calculated in the center coordinate value calculation step for the former / the latter groups, . The left and right foot discrimination step according to the embodiment of the present invention determines that the entire group 25a is left-footed when the x-axis coordinate value of the central coordinate value is larger than the rear group 25b, Axis coordinate value of the center coordinate value is larger than the center coordinate value of the center coordinate value. This is because the shape of the shoe is asymmetrical in the anteroposterior direction and the forefoot region is bent inward.

The shoe sole image recognition and shoe sole image recognition based on shoe sole surface marker recognition according to the embodiment of the present invention configured as described above includes a plurality of first marks arranged in a square shape at a predetermined interval on the fore part 24a of the shoe sole 10, The arrangement distribution characteristics of a plurality of second marks 12 formed of a plurality of heating pin holes 12a arranged in a non-square shape in a set pattern on the shoe sole image 11 and the rear area 24b are shown in the clustering algorithm and marks The shoe sole image can be aligned and the left and right foot shoes can be separated by analyzing the deviation, distribution and center coordinate value of the shoe sole 10, Even if the throwing direction of the sole 10 is not constant, it is possible to perform a dimensional inspection as compared with the last, which is a standard of the shoe sole quality inspection, so that the quality inspection of the shoe sole can be performed accurately and smoothly The.

Although the shoe sole image alignment and shoe soles identification method based on shoe sole surface marker recognition according to the embodiment of the present invention has been described above with reference to the above description and drawings, It will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.

10: Shoes brush
10a: Whole site
10b: posterior region
11: First mark
12: Second mark
12a: heating pin hole
20: shoe sole picture frame
21: Clustering unit
22: Reference straight line
23a: the first group
23b: the second group
24a:
24b: rear region
25a: The whole group
25b: Tribe Group
100:
200: Analyzer
210: memory
220: Coordinate setting module
230: Analysis module
231: KNN Clustering Algorithm
240: image alignment module

Claims (7)

A front part (10a) in which a plurality of first marks (11) are arranged squarely at a set interval, a rear part (12a) in which a plurality of second marks (12) formed by a plurality of heating fin holes (12a) 10b is shot by a photographing apparatus 100 and input to the analyzing apparatus 200 in real time;
A global coordinate system is set in the shoe sole image frame (20) by the coordinate setting module (220) of the analyzer (200);
A reference straight line 22 passing through the markers arranged in a line is calculated by the analysis module 230 of the analysis apparatus 200 and the specimen rotation A shoe sole closing direction displacement detecting step in which the angle is detected by the analysis module 230 of the analyzer 200;
The analysis module 230 of the analysis apparatus 200 divides the mark in the shoe sole image frame 20 into two groups of the forefoot region 24a and the rear region 24b, It is determined whether or not the front portion 24a and the rear portion 24b are arranged in the longitudinal direction in the shoe sole image frame 20 according to the mark distribution characteristic of the rear portion 24b, And a 180 ° rotation correction value is generated when the forefoot portion (24a) is disposed rearward in the longitudinal direction;
The shoe sole image frame 20 is detected by the analyzer 200 (200) by an alignment angle obtained by adding 180 ° rotation correction values that can be generated in the shoe sole back and forth inversion correction step to the sample rotation angle calculated in the shoe sole input direction displacement detection step A shoe sole image frame rotation correcting step in which the shoe sole image frame 20 is rotated by the image aligning module 240 of the shoe sole image frame 20,
The alignment mark in the shoe sole image frame 20 is divided into two groups of the front group 25a and the rear group 25b by the analysis module 230 of the analysis apparatus 200, And the left and right feet are discriminated by comparing the x-coordinate values of the center coordinate values of the markers belonging to each of the groups (25b). The shoe sole surface image recognition- Identification method. The method according to claim 1,
The shoe sole image frame global coordinate system setting step may include:
The x-axis direction of the global coordinate system is set in the direction of a straight line passing through the first mark 11 arranged in a line in the left-right direction on the forefoot portion 24a of the shoe sole 10 normally inserted, Is set in the direction of a straight line passing through the second marking (12) arranged in a row in the longitudinal direction at the center of the rear region (24b) of the normally inserted shoe sole (10) Image alignment and identification of left and right feet. 3. The method of claim 2,
Wherein the shoe sole input direction displacement detecting step comprises:
The K-nearest neighbor clustering algorithm 231 included in the analysis module 230 of the analysis apparatus 200 determines whether any one of the x-axis direction and the y-axis direction is used as a reference And a plurality of clustering units (21) are formed while being divided into groups according to the groups;
The reference clustering unit 21 having the largest number of landmarks belonging to the clustering unit 21 and having the smallest deviation of the landmark coordinate values is obtained from the plurality of clustering units 21 by the analysis module 230 of the analysis apparatus 200 Selecting a reference clustering unit to be selected;
A reference straight line calculating step in which a reference straight line 22 passing through the mark belonging to the reference clustering unit 21 is calculated by the analysis module 230 of the analyzer 200;
The rotation angle of the specimen formed by the reference line 22 and one axis selected from the x-axis and the y-axis in the global coordinate system of the shoe sole image frame 20 is measured by the analysis module 230 of the analyzer 200 Wherein the shoe sole surface mark recognition based shoe sole image alignment and the left and right foot identification are included. The method of claim 3,
Wherein the step of correcting the inverse of the shoe sole comprises:
A K-nearest neighbor clustering algorithm 231 included in the analysis module 230 of the analysis apparatus 200 determines whether or not the first group 23a ) And a second group (23b); dividing the shoe soles into two groups;
Group variance value calculating step of calculating an x-axis variance value of the markers belonging to each of the first group 23a and the second group 23b;
A full-length zone designation step of designating a group having a large variance value among the first group 23a and the second group 23b as a full-length zone 24a and a group having a small dispersion value as a rear zone 24b;
The average of the y-axis coordinate values of the markers belonging to the group designated as the full-length zone 24a and the average of the y-axis coordinate values of the marks belonging to the group designated as the rear zone 24b are transmitted to the analysis module 230 of the analyzer 200 And when the average of the y-axis coordinate values of the markers belonging to the group designated as the fore part 24a is smaller than the average of the y-axis coordinate values of the markers belonging to the group designated as the rear zone 24b, And a 180 ° rotation correction value generation step of generating a 180 degree rotation correction value for each shoe sole. 5. The method of claim 4,
Wherein the step of identifying the left and right hands comprises:
Nearest neighbor clustering algorithm 231 included in the analysis module 230 of the analysis apparatus 200 determines whether or not the alignment marks in the shoe sole image frame 20 are concentrated on the entire group 25a ) And a tribe group (25b).
A center coordinate value of each of the front and rear groups of the front group 25a and the rear group 25b is calculated by the analysis module 230 of the analyzer 200;
The analysis module 230 of the analysis apparatus 200 compares the coordinate values of the x-axis of the center coordinate values calculated in the center coordinate value calculation step of the pre / tributary group to determine the left and right contingencies, When the x-axis coordinate value of the center coordinate value is larger than the rear group 25b, the left and right feet are discriminated. When the x-axis coordinate value of the center coordinate value is larger than the forefoot group 25b in the front group 25a, Wherein the shoe sole surface mark recognition based shoe sole image alignment and the left and right foot identification method are characterized by including a discriminating step. The method of claim 3,
The calculating of the plurality of clustering units causes the longitudinal clustering in which a plurality of clustering units 21 are formed while the first landmarks 11 and the second landmarks 12 are divided in groups along the y axis direction,
The sample rotation angle calculating step is such that the sample rotation angle formed by the x-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22 is calculated by the analysis module 230 of the analyzer 200 A method of shoe sole image alignment and identification of left and right foot based on shoe sole surface marker recognition. The method of claim 3,
The plurality of clustering unit calculating steps may be performed such that the widthwise clustering in which the plurality of clustering units 21 are formed while the second landmarks 12 are divided in groups along the x-
The sample rotation angle calculating step is such that the sample rotation angle formed by the y-axis of the global coordinate system of the shoe sole image frame 20 and the reference straight line 22 is calculated by the analysis module 230 of the analyzer 200 A method of shoe sole image alignment and identification of left and right foot based on shoe sole surface marker recognition.

Images