CN113129397B

CN113129397B - Decoding method of parallelogram coding mark based on graphic geometric relation

Info

Publication number: CN113129397B
Application number: CN202110392348.XA
Authority: CN
Inventors: 朱华炳; 杨霈; 殷玉龙; 杨昭辉
Original assignee: Hefei University of Technology
Current assignee: Hefei University of Technology
Priority date: 2020-12-23
Filing date: 2021-04-13
Publication date: 2022-10-14
Anticipated expiration: 2041-04-13
Also published as: CN113129397A

Abstract

The invention discloses a decoding method of a parallelogram coding mark based on a graph geometric relationship, wherein the parallelogram coding mark is composed of a parallelogram background pattern and a coding pattern, the coding pattern is arranged in the parallelogram background pattern and comprises a positioning pattern, an orientation pattern and a coding combination pattern, and the orientation pattern and the positioning pattern are used for judging the direction of the parallelogram coding mark; the coding combination pattern is used for coding four vertexes of the coding mark for coding. The coding pattern in the parallelogram coding mark provided by the invention can automatically judge the relative rotation direction of the measured target and the camera in the processes of camera calibration, three-dimensional data splicing and the like, thereby improving the flexibility; the positions of the patterns of all the components are distinguished by a method of geometric relationship and vector product of graphs, so that the method has stronger robustness; the decoding method greatly improves the accuracy of decoding; the method has the advantages of high efficiency and high decoding speed, and can realize real-time decoding.

Description

Decoding method of parallelogram coding mark based on graphic geometric relation

Technical Field

The invention relates to the field of vision measurement in computer vision, in particular to a decoding method of a parallelogram coding mark based on a graph geometric relation, which is suitable for the fields of camera calibration, target feature extraction, stereo matching, three-dimensional data splicing and the like.

Background

In visual measurement, feature identification and matching of measured objects are one of the research hotspots in the field. However, under the condition of a high-resolution large field of view or other complex backgrounds, the accuracy of feature extraction and matching cannot meet the precision requirement, and the speed and precision of the current feature extraction and matching can be greatly improved by arranging the coding marks on the surface of the measured object, and meanwhile, the problem of matching of multiple images can be solved. Therefore, the design and identification of coded signs is a key element in visual measurement.

In 1972, russo and Knockeart realized compilation and recognition by using unique characteristics of marker points, and since then, technology for encoding marker points has been rapidly developed. In order to meet the requirements of industrial measurement, there are many types of coding mark points, but most of them are ring, circle or sector, and the coding mark points have the following defects: when the shooting angle changes, part of the codes deform, thereby influencing the precision; the capacity of the code is small, and the code cannot be applied to a target with a large size.

Disclosure of Invention

The invention aims to overcome the defects of the existing coding mark, provides a decoding method of a parallelogram coding mark based on the geometric relation of graphs, is suitable for multi-view stereoscopic vision calibration under a large view field, has good stability on affine transformation images, has large coding capacity, and can also have higher coding precision and accuracy for targets with larger sizes.

In order to realize the effect, the invention adopts the technical scheme that:

a decoding method of a parallelogram coding mark based on a graph geometric relationship is characterized in that the coding mark is a square coding grid, a parallelogram background pattern and a coding pattern are arranged on the surface of the coding grid, the coding pattern is positioned inside the parallelogram background pattern, the coding pattern comprises a positioning pattern, an orientation pattern and a coding combination pattern, the orientation pattern and the positioning pattern are used for judging the direction of the parallelogram coding mark, the coding combination pattern is used for coding each corner point of the parallelogram coding mark, and each coding combination pattern is composed of one coding combination orientation pattern and three coding combination coding patterns. The method for processing the digital image is used for obtaining the coding sequence number and the sub-pixel coordinates of each characteristic corner point in the image which is shot by the camera and contains N parallelogram coding marks, and further completing the decoding of the parallelogram coding marks, and comprises the following steps:

step 1.1, acquiring a coding mark image by utilizing N parallelogram coding marks arranged in a camera shooting space, wherein the coding mark image comprises N parallelogram coding marks;

step 1.2, establishing a pixel coordinate system of an angular point:

in the shot image containing N parallelogram coding marks, taking the upper left corner of the image as the origin o of an angular point pixel coordinate system, taking the upper left corner of the image from left to right as the x-axis direction of the angular point pixel coordinate system, and taking the upper left corner of the image from top to bottom as the y-axis direction of the angular point pixel coordinate system, thereby establishing an angular point pixel coordinate system o-xy;

step 2.1, carrying out 8-bit gray level processing on the coding mark image to obtain a coding mark gray level image P ₁ (ii) a Wherein the code marks the grayscale image P ₁ Is an 8-bit gray scale map;

step 2.2, the grey image P of the coding mark ₁ Copying and backing up to obtain the grey backup image P of the code mark ₁ ′；

Step 2.3, the grey level backup image P of the coding mark ₁ Performing binarization processing to make the background color of each parallelogram coding mark black (i.e. gray value of 0), the orientation pattern and the positioning pattern in each parallelogram coding mark white (i.e. gray value of 255), and the color of the coding unit pattern can be determined according to the coding rule of the parallelogram coding markWhite or black to obtain a binary image P of the coding mark ₂ ；

Step 3, setting a roundness threshold lambda 'of the black connected domain and a pixel number threshold lambda';

step 4.1, calculating a binary image P of the coding mark ₂ Circularity value λ of all black connected components in the image _n And the number of pixel points c _n Wherein the circularity value of each black connected domain can be obtained by equation (1),

wherein l _n Perimeter, s, representing the contour of the corresponding black connected component _n Represents the area corresponding to the black connected domains, n =1,2,3,4;

step 4.2, selecting proper black connected domain roundness threshold value lambda 'and pixel number threshold value lambda' to encode the mark binary image P ₂ Black connected domains of the centers of N directional circular rings corresponding to the N parallelogram coding marks are obtained; respectively recording black connected domains at the centers of the N directional circular rings as a binary image P of a coding mark ₂ Middle ring black connected domain M ₁ Ring black connected domain M ₂ Ring black connected domain M ₃ 823060, circular black connected domain M _N And connecting the ring black connected domain M ₁ Circular ring black connected domain M ₂ Ring black connected domain M ₃ 823060, circular black connected domain M _N Sequentially putting into a ring black connected domain set A', namely a ring black connected domain M ₁ Circular ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M _N The first element, the second element, the third element, the fourth element, the fifth element, the sixth element and the sixth element in the annular black connected domain set A' respectively; thereby obtaining a coded mark binarized image P ₂ Black connected domains inside the directional circular rings in all the N parallelogram coding marks; wherein, the coded mark binarizes the image P ₂ Middle ring black connected domain M ₁ Ring black connected domainM ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M _N Are all smaller than a roundness threshold lambda', and a black connected domain M of a circular ring ₁ Ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M _N The number of the pixel points is larger than the threshold value lambda' of the number of the pixel points;

step 4.3, taking an integer variable i and assigning i =1;

step 4.4, calculating the ith ring black connected domain M in the ring black connected domain set A _i And the coordinates of the centroid pixel are marked as o ″ _d,i (x″ _d,i ,y″ _d,i ) Obtaining the ith ring black connected domain M _i Centroid pixel coordinate o ″) _d,i (x″ _d,i ,y″ _d,i ) The ith element is taken as a centroid set A of the ring black connected domain;

step 4.5, judging whether i is smaller than N, if i is smaller than N, assigning i +1 to i, and returning to the step 4.4 for sequential execution; otherwise, obtaining the binary image P of the coding mark ₂ The center of mass pixel coordinate o' of each ring black connected domain _d,1 (x″ _d,1 ,y″ _d,1 )、o″ _d,2 (x″ _d,2 ,y″ _d,2 )、…、o″ _d,N (x″ _d,N ,y″ _d,N ) Sequentially putting the circular ring black connected domain centroid set A into the circular ring black connected domain centroid set A;

step 5, taking an integer variable zeta and assigning zeta =1;

step 6.1, binarization image P of coding mark ₂ Copying and backing up twice to respectively obtain a 1 st backup binary image P 'of the zeta group' _ζ,1 And the 2 nd backup binary image P 'of the zeta th group' _ζ,2 ；

Step 6.2, utilizing a complex background algorithm for removing the coding mark to carry out binarization on the 1 st backup binary image P 'of the zeta-th group' _ζ,1 Digital image processing is carried out to obtain the zeta th parallelogram binaryzation image P without complex background _ζ,1 ；

At the zeta th non-complex background parallelogram binaryzation image P ″ _ζ,1 In, including centroid pixelCoordinate value (x ″) _d,ζ ,y″ _d,ζ ) The gray value of all pixel points in the parallelogram coding mark is set to be 255 (namely white), and the pixel coordinate value (x ″) containing the centroid is contained _d,ζ ,y″ _d,ζ ) The gray value of all pixel points outside the parallelogram coding mark is set to be 0 (namely black);

step 7.1, coding the mark gray level image P ₁ Copying and backing up again to obtain the zeta th backup gray image P _ζ ″′；

Step 7.2, utilizing a corner detection algorithm to carry out binaryzation on the zeta-th non-complex background parallelogram image P ″ _ζ,1 Extracting sub-pixel coordinates of 4 corner points, and marking as the zeta th non-complex background parallelogram binaryzation image P ″ _ζ,1 4 alternative corner points C ″) _ζ,1 (x″ _ζ,1 ,y″ _ζ,1 )、C″ _ζ,2 (x″ _ζ,2 ,y″ _ζ,2 )、C″ _ζ,3 (x″ _ζ,3 ,y″ _ζ,3 )、C″ _ζ,4 (x″ _ζ,4 ,y″ _ζ,4 )；

Step 7.3, in the zeta th parallelogram binaryzation image P ″, without complex background _ζ,1 Black connected domain corrosion is performed in the process, so that the zeta-th parallelogram binaryzation image P' without complex background _ζ,1 The white connected domain is enlarged, and when the subsequent image multiplication operation is carried out, the complete pixel coordinate value (x') containing the centroid extracted from the operation result image can be ensured _d,ζ ,y″ _d,ζ ) The 4 characteristic angular points of the gray pattern of the parallelogram coding mark are completely stored; and the zeta th parallelogram binaryzation image P' without complex background _ζ,1 The image obtained by the processing in this step is recorded as the ζ -th operation binary image P ″ _ζ,2 ；

Step 7.4, the zeta-th operation binaryzation image P ″ _ζ,2 Setting the value of the part with the middle gray scale value of 255 as 1, and calculating the zeta-th binary image P ″ _ζ,2 Setting the value of the part with the middle gray value of 0 as 0, and then utilizing the zeta-th operation to binarize the image P ″) _ζ,2 And the ζ th backup grayscale image P _ζ "' multiplication operationRecording the obtained result image as the Zeth coverage result image T _ζ (ii) a The ζ th overlay result image T _ζ Contains only the coordinate value (x') of the pixel containing centroid _d,ζ ,y″ _d,ζ ) The image information of the parallelogram coding mark of (1) removes the image information of other parallelogram coding marks;

step 7.5, covering the result image T at the zeta th by utilizing a corner point detection algorithm _ζ Extracting sub-pixel coordinates of all corner points and storing the sub-pixel coordinates into a zeta-th primary selection corner point sub-pixel coordinate set B _ζ Performing the following steps; wherein, the Zeth primary selection corner sub-pixel coordinate set B _ζ In which the ζ th overlay result image T is stored _ζ Contains the coordinate value (x ″) of centroid pixel _d,ζ ,y″ _d,ζ ) The sub-pixel coordinate values of 4 characteristic angular points of the parallelogram coding mark and the sub-pixel coordinate values of interference angular points, wherein the interference angular points are the interference angular points except the characteristic angular points;

step 7.6, taking an integer variable i and assigning i =1;

step 7.7, in the zeta th primary corner point sub-pixel coordinate set B _ζ Finding the coordinate value (x ″) of the distance sub-pixel _ζ,i ,y″ _ζ,i ) Nearest corner sub-pixel coordinate and denoted C' _ζ,i (x′ _ζ,i ,y′ _ζ,i )；

7.8, judging whether i is smaller than 4, if i is smaller than 4, assigning i +1 to i and returning to execute the step 7.7; otherwise, the ζ -th overlay result image T has been obtained at this time _ζ Contains the coordinate value (x ″) of the centroid pixel _d,ζ ,y″ _d,ζ ) Of 4 characteristic corner points of the parallelogram-coded logo C' _ζ,1 (x′ _ζ,1 ,y′ _ζ,1 )、C′ _ζ,2 (x′ _ζ,2 ,y′ _ζ,2 )、C′ _ζ,3 (x′ _ζ,3 ,y′ _ζ,3 )、C′ _ζ,4 (x′ _ζ,4 ,y′ _ζ,4 ) And step 8.1 is executed;

step 8.1, backup binarization image P 'at the 2 nd of the zeta group' _ζ,2 Above, will contain the centroid pixel coordinate value (x ″) _d,ζ ,y″ _d,ζ ) The parallelogram coding mark is marked as the zeta th parallelogram coding mark;

step 8.2, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Wherein the sub-pixel coordinate values are (x' _ζ,1 ,y′ _ζ,1 )、(x′ _ζ,2 ,y′ _ζ,2 )、(x′ _ζ,3 ,y′ _ζ,3 )、(x′ _ζ,4 ,y′ _ζ,4 ) The 4 pixel points are used as the corner quadrangle S of the zeta-th parallelogram coding mark _ζ 4 vertices of the Zeta parallelogram code symbol, and connecting the 4 vertices to obtain the corner quadrangle S of the Zeta parallelogram code symbol _ζ ；

Step 9, backup binarization image P 'at 2 nd of the zeta group' _ζ,2 In the above, find the pixel coordinate value (x ″) from the centroid _d,ζ ,y″ _d,ζ ) The nearest white connected domain, and the gray value of the white connected domain is assigned to be 0;

step 10, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Corner quadrilateral S of upper Zeta parallelogram coding sign _ζ Keeping the gray values of all internal pixel points unchanged, and encoding the corner point quadrangle S of the zeta-th parallelogram encoding mark _ζ The gray values of all other pixel points are assigned to be 255;

step 11.1, backup binarization image P 'at 2 nd of zeta group' _ζ,2 In the above, the corner point quadrangle S of the Zeth parallelogram coding mark is extracted _ζ The inner and outer contours in (1) are marked as a contour set D in the zeta-th parallelogram coding mark _ζ ；

Step 11.2, counting the outline set D in the zeta th parallelogram coding mark _ζ The number of pixel points in each contour is recorded as the 2 nd backup binary image P 'of the ζ group' _ζ,2 Positioning circle contour G in upper Zeth parallelogram coding mark _ζ Calculating the coordinates of the centroid pixel of the binary image and recording the coordinates as the 2 nd backup binary image P 'of the zeta group' _ζ,2 Locating circle centroid pixel coordinates o 'within the upper zeta parallelogram encoding logo' _l,ζ (x′ _l,ζ ,y′ _l,ζ )；

Step 12, collecting the outline D in the Zeth parallelogram coding sign _ζ The inner 4 contours with the least number of pixel points are marked as the contours Z of the coding combination orientation pattern _ζ,1 Coded combined orientation pattern profile Z _ζ,2 Coded composite directional pattern profile Z _ζ,3 Coded combined orientation pattern profile Z _ζ,4 And respectively calculate the centroid thereof

Step 13.1, according to the outline set D in the zeta th parallelogram coding mark _ζ In the method, two contours with the largest number of pixel points and 4 contours with the smallest number of pixel points are removed, and the number of remaining contours is kappa _ζ The size of (c) is divided into the following cases:

(1) If κ _ζ =0, then step 14.1 is performed;

(2) If κ _ζ Not equal to 0, then this is κ _ζ The individual outline is the outline of the coding combination coding pattern in the zeta-th parallelogram coding mark and is respectively marked as the coding combination pattern

Coded composite pattern

8230code combined pattern

Step 14.1, giving an integer variable i and an initial value i =1;

step 14.2, backup binarization image P 'at the 2 nd of the zeta group' _ζ,2 Calculating the code pattern

Center of mass of

This step continues after reassigning i +1 to i until i > κ _ζ Finishing; whereby a corresponding coded composite pattern can be obtained

Coded composite pattern

823060A combined coding pattern

Centroid pixel coordinates of

Step 15.1, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 In the above, C is the sub-pixel coordinate of the feature corner point of the ζ -th parallelogram coding mark located in the 1 st coding region, the 2 nd coding region, the 3 rd coding region and the 4 th coding region respectively _ζ,1 (x _ζ,1 ,y _ζ,1 )、C _ζ,2 (x _ζ,2 ,y _ζ,2 )、C _ζ,3 (x _ζ,3 ,y _ζ,3 )、C _ζ,4 (x _ζ,4 ,y _ζ,4 ) (ii) a And the 2 nd backup binary image P 'in the ζ th group' _ζ,2 Wherein the pixel coordinate values are (x' _d,ζ ,y′ _d,ζ ) The pixel point of (1) is recorded as a zeta-group 2 nd backup binary image P' _ζ,2 Oriented Ring centroid o 'of the upper ζ -th parallelogram-encoded logo' _d,ζ (x′ _d,ζ ,y′ _d,ζ ) And the sub-pixel coordinate value is (x' _ζ,1 ,y′ _ζ,1 )、(x′ _ζ,2 ,y′ _ζ,2 )、(x′ _ζ,3 ,y′ _ζ,3 )、(x′ _ζ,4 ,y′ _ζ,4 ) The 4 pixel points are respectively marked as C' _ζ,1 (x″′ _ζ,1 ,y″′ _ζ,1 )、C″′ _ζ,2 (x″′ _ζ,2 ,y″′ _ζ,2 )、C″′ _ζ,3 (x″′ _ζ,3 ,y″′ _ζ,3 )、C″′ _ζ,4 (x″′ _ζ,4 ,y″′ _ζ,4 )；

Step 15.2, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Upper, the direction vector in the ζ -th parallelogram coding index

As is derived from the formula (2),

step 16, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 In the above, 4 pixel points C' _ζ,1 (x″′ _ζ,1 ,y″′ _ζ,1 )、C″′ _ζ,2 (x″′ _ζ,2 ,y″′ _ζ,2 )、C″′ _ζ,3 (x″′ _ζ,3 ,y″′ _ζ,3 )、C″′ _ζ,4 (x″′ _ζ,4 ,y″′ _ζ,4 ) Pixel coordinate o 'of middle distance positioning circle centroid' _l,ζ (x′ _l,ζ ,y′ _l,ζ ) The nearest 2 pixels are respectively marked as C _ζ,1min (x _ζ,1min ,y _ζ,1min ) And C _ζ,2min (x _ζ,2min ,y _ζ,2min )；

Finding out the sub-pixel coordinate values of the characteristic corner points of the 2 nd coding region and the 3 rd coding region in the zeta-th parallelogram coding mark in 4 pixel points and respectively assigning the sub-pixel coordinate values to C _ζ,2 (x _ζ,2 ,y _ζ,2 ) And C _ζ,3 (x _ζ,3 ,y _ζ,3 )；

Calculating the 1 st judgment vector in the ζ th parallelogram coding flag by the formulas (3) and (4)

And 2 nd decision vector

And calculated by the formulas (5) and (6)Dividing the 1 st area into sine values sin alpha _ζ And the 2 nd area division sine value sin beta _ζ ；

The following two cases are performed:

(1) If sin alpha _ζ ＜0,sinβ _ζ Greater than 0, then C _ζ,1min (x _ζ,1min ,y _ζ,1min ) For the characteristic corner of the 3 rd coding region in the ζ -th parallelogram coding index, C _ζ,1min (x _ζ,1min ,y _ζ,1min ) Is assigned to C _ζ,3 (x _ζ,3 ,y _ζ,3 )；C _ζ,2min (x _ζ,2min ,y _ζ,2min ) For the characteristic corner of the 2 nd coding region in the ζ th parallelogram coding index, C _ζ,2min (x _ζ,2min ,y _ζ,2min ) Is assigned to C _ζ,2 (x _ζ,2 ,y _ζ,2 )；

(2) If sin alpha _ζ ＞0,sinβ _ζ Less than 0, then C _ζ,2min (x _ζ,2min ,y _ζ,2min ) For the characteristic corner point of the 3 rd coding region in the ζ th parallelogram coding mark, C _ζ,2min (x _ζ,2min ,y _ζ,2min ) Is assigned to C _ζ,3 (x _ζ,3 ,y _ζ,3 )；C _ζ,1min (x _ζ,1min ,y _ζ,1min ) Coding the Zeta-th parallelogram with a logoFeature corner of No. 2 coding region in the notation, C _ζ,1min (x _ζ,1min ,y _ζ,1min ) Is assigned to C _ζ,2 (x _ζ,2 ,y _ζ,2 )；

Step 17, backup binarization image P 'at 2 nd of the zeta group' _ζ,2 At last, by the above step 16, 4 pixel points C ″ 'have been already existed' _ζ,1 (x″ _ζ,1 ,y″′ _ζ,1 )、C″′ _ζ,2 (x″′ _ζ,2 ,y″′ _ζ,2 )、C″′ _ζ,3 (x″′ _ζ,3 ,y″′ _ζ,3 )、C″′ _ζ,4 (x″′ _ζ,4 ,y″′ _ζ,4 ) Finding out the sub-pixel coordinate values of the characteristic corner points of the 2 nd coding region and the 3 rd coding region in the zeta-th parallelogram coding mark and respectively assigning the sub-pixel coordinate values to the C _ζ,2 (x _ζ,2 ,y _ζ,2 ) And C _ζ,3 (x _ζ,3 ,y _ζ,3 ) Respectively assigning the sub-pixel coordinate values of the rest 2 pixel points to a temporary coordinate value 1 of the zeta-th parallelogram coding mark, and recording as C' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) And temporary coordinate value 2, marked as C' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 )；

Determining the 3 rd judgment vector in the ζ th parallelogram coding flag according to equations (7) and (8)

And 4 th judgment vector

Step 18, backup binarization image P 'at 2 nd of the zeta group' _ζ,2 Root of Chinese Thalic acidAccording to the 3 rd judgment vector

And 4 th judgment vector

The 3 rd region division sine value sin omega is obtained by the formula (9) and the formula (10) _ζ And 4 th area division sine value sin xi _ζ ；

According to sin ω _ζ And sin xi _ζ C is' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Or C' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Is assigned to C _ζ,4 (x _ζ,4 ,y _ζ,4 ) Or C _ζ,1 (x _ζ,1 ,y _ζ,1 )；

The execution is divided into the following two cases:

(1) Sin ω _ζ ＝＝0,sinξ _ζ Not equal to 0, then C' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Namely C 'is the characteristic corner point of the 4 th coding area in the ζ th parallelogram coding mark' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Is assigned to C _ζ,4 (x _ζ,4 ,y _ζ,4 )；C′ _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) C 'is a characteristic corner point of a 1 st coding region in a zeta th parallelogram coding mark' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Is assigned to C _ζ,1 (x _ζ,1 ,y _ζ,1 )；

(2) Sin omega _ζ ≠0,sinξ _ζ =0, then C' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Namely C 'is the characteristic corner point of the 4 th coding area in the ζ th parallelogram coding mark' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Is assigned to C _ζ,4 (x _ζ,4 ,y _ζ,4 )；C′ _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) C 'is a characteristic corner point of a 1 st coding region in a zeta th parallelogram coding mark' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Is assigned to C _ζ,1 (x _ζ,1 ,y _ζ,1 )；

To this end, the binary image P 'is backed up at the 2 nd backup of the ζ group' _ζ,2 The sub-pixel coordinate C of the characteristic corner point of the 1 st coding region in the ζ -th parallelogram coding mark is found _ζ,1 (x _ζ,1 ,y _ζ,1 ) Sub-pixel coordinate C of characteristic corner point of No. 2 coding region _ζ,2 (x _ζ,2 ,y _ζ,2 ) 3. Sub-pixel coordinate C of characteristic corner point of the 3 rd coding region _ζ,3 (x _ζ,3 ,y _ζ,3 ) And the sub-pixel coordinate C of the characteristic corner of the 4 th coding region _ζ,4 (x _ζ,4 ,y _ζ,4 )；

Step 19, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Will be composed of characteristic corner points C _ζ,1 (x _ζ,1 ,y _ζ,1 ) Angular point C of orientation feature _ζ,4 (x _ζ,4 ,y _ζ,4 ) Is recorded as a vector

And is derived from equation (11):

taking an integer variable i and assigning i =1;

step 20, backup binarization image P 'at 2 nd of the zeta group' _ζ,2 According to the coding combination, the pattern centroid is oriented

(Vector)

And direction vector

The i-th code combination decision vector is calculated by the equations (12), (13) and (14)

Coding combination judgment cosine value

And

the following four cases are performed:

(1) If it is

And is provided with

Then

To orient the pattern centroids for the encoded combination of the 3 rd encoded region

Value to o' _z ^ζ,3 (x′ _z ^ζ,3 ,y′ _z ^ζ,3 )；

(2) If it is

And is provided with

Then

To orient the pattern centroids for the encoded combination of the 2 nd encoded region, will

Value to o' _z ^ζ,2 (x′ _z ^ζ,2 ,y′ _z ^ζ,2 )；

(3) If it is

And is provided with

Then the

To orient the pattern centroids for the code combinations of the 1 st code region

Assigned to o' _z ^ζ,1 (x′ _z ^ζ,1 ,y′ _z ^ζ,1 )；

(4) If it is

And is provided with

Then the

To orient the pattern centroids for the encoded combination of the 4 th encoded region

Assigned to o' _z ^ζ,4 (x′ _z ^ζ,4 ,y′ _z ^ζ,4 )；

Judging whether i meets the condition that i is less than 4, if i is less than 4, assigning i +1 to i, and returning to execute the step 20; otherwise, the code combined orientation pattern centroid o 'of the 1 st code area in the ζ th parallelogram code mark is obtained' _z ^ζ,1 (x′ _z ^ζ,1 ,y′ _z ^ζ,1 ) Coded combined directional pattern centroid o 'of 2 nd coded region' _z ^ζ,2 (x′ _z ^ζ,2 ,y′ _z ^ζ,2 ) And the center of mass o' of the code combined orientation pattern in the No. 3 code area _z ^ζ,3 (x′ _z ^ζ,3 ,y′ _z ^ζ,3 ) And a coded combined directional pattern centroid o 'of a 4 th coded region' _z ^ζ,4 (x′ _z ^ζ,4 ,y′ _z ^ζ,4 ) And continuing to execute step 21;

step 21, assigning the integer variable i with i =1 again;

step 22, defining 4 two-dimensional floating-point type numerical groups

Initializing all elements in the 4 two-dimensional arrays, and assigning a value of-1; taking 4 integer variables

Initialization

Step 23, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Calculating the contour of the code combination code pattern mark circle in the Zeth parallelogram code mark

Centroid pixel coordinates of

And the positioning circle center o' _l,ζ The ith difference vector of the formed zeta th group

Calculated by the following equations (16) and (17)

The method is executed in four cases:

(1) If it is

And is provided with

Then make it give

Will be provided with

Is assigned to

If i > k _ζ 1, executing step 24, otherwise assigning i +1 to i, and returning to execute step 23;

(2) If it is

And is

Then order

Will be provided with

Is assigned to

If i > k _ζ 1, executing step 24, otherwise, assigning i +1 to i, and returning to execute step 23;

(3) If it is

And is

Then order

Will be provided with

Is assigned to

(4) If it is

And is provided with

Then make it give

Will be provided with

Is assigned to

step 24.1, assigning the integer variable i with the value i =1 again;

step 24.2, according to

Taking integer variables

And

and assigning values;

the following cases are divided into execution:

(1) If it is

Then take the integer variable

And

and assign a value

(2) If it is

Will array

Elements in the same row which are not-1 are respectively marked as

And

and are respectively assigned to _ζ,1 (x _ζ,1 ,y _ζ,1 ) And the abscissa and the ordinate of the computer, and calculating the sine value

And (4) judging:

1) If it is

Then get the integer variable

And

and assign values

2) If it is

Then take the integer variable

And

and assign a value

3) If it is

Then take the integer variable

And

and assign values

(3) If it is

Will array

Elements in the same row which are not-1 are respectively marked as

And

and

and are respectively assigned to _ζ,1 (x _ζ,1 ,y _ζ,1 ) Abscissa and ordinate and o _ζ,2 (x _ζ,2 ,y _ζ,2 ) The abscissa and the ordinate of the computer, while calculating the sine value

And

and judging:

1) If it is

Or

Then take the integer variable

And

and assign values

2) If it is

Or

Then get the integer variable

And

and assign values

3) If it is

Or

Then get the integer variable

And

and assign values

(4) If it is

Then get the integer variable

And

and assign values

If i is less than 4, assigning i +1 to i and returning to execute the step 24.2; otherwise, continuing to execute step 25;

step 25, obtaining the serial number W of the zeta th parallelogram coding mark through the formula (21) _ζ ：

W _ζ ＝V _ζ ^T ·U (21)

Wherein, the column vector U = (2) ⁰ ,2 ¹ ,2 ² ,...2 ¹¹ ) ^T Column vector of

Step 26, recording the code number of the characteristic corner point belonging to the sigma-th coding region (wherein sigma =1,2,3, 4) in the zeta-th parallelogram coding mark as

Wherein W _ζ For calibrating angular points

Coding of the coding squaresThe value of the number and the sigma represents a calibration angular point

The sigma-th coding region;

thus, the coding numbers and the corresponding sub-pixel coordinate values of the 4 feature corners in the ζ -th parallelogram coding mark are obtained:

coded number is

The sub-pixel coordinate value of the characteristic corner is (x) _ζ,1 ,y _ζ,1 )；

The code number is

The sub-pixel coordinate value of the characteristic corner is (x) _ζ,2 ,y _ζ,2 )；

Coded number is

The sub-pixel coordinate value of the characteristic corner is (x) _ζ,3 ,y _ζ,3 )；

Coded number is

The sub-pixel coordinate value of the characteristic corner is (x) _ζ,4 ,y _ζ,4 )；

Step 27, judging whether zeta is smaller than N, if yes, reassigning zeta +1 to zeta, and returning to step 6.1 for sequential execution; if not, the sub-pixel coordinate values and the corresponding coding numbers of the 4 characteristic angular points on each parallelogram coding mark are found.

Further, in step 6.1, obtaining a 1 st backup binary image P 'of the zeta th group' _ζ,1 And the Zeth circular ring black connected domain centroid pixel coordinate value (x ″) in the circular ring black connected domain centroid set A _d,ζ ,y″ _d,ζ ) Removing the 1 st backup binary image P 'of the zeta-th group by using a complex background algorithm for removing the coding mark as an input condition' _ζ,1 And finally obtaining the zeta th parallelogram binaryzation image P' without complex background _ζ The steps are as follows:

step 6.1.1, taking an integer variable i and assigning i =1;

step 6.1.2, backup binary image P 'at the 1 st backup of the ζ th group' _ζ,1 Find the distance pixel coordinate value (x ″) _d,ζ ,y″ _d,ζ ) The nearest white connected domain is recorded as the 1 st backup binary image P 'of the zeta th group' _ζ,1 The ith circular ring white connected domain M' _i And the ith circular ring white connected domain M' _i All the gray values are assigned to be 0;

step 6.1.3, judging whether i is smaller than N, if i is smaller than N, assigning i +1 to i, and returning to the step 6.1.2 for sequential execution; otherwise, the 1 st backup binary image P 'of the ζ group is obtained' _ζ,1 Setting the orientation circular rings in all the parallelogram coding marks to be black, and setting the 1 st backup binary image P 'of the zeta-th group' _ζ,1 The image processed in this step is recorded as the zeta-th black background extraction image P' _ζ,b And executing step 6.1.4;

step 6.1.4, extracting image P 'from the zeta-th black background' _ζ,b In all the black connected domains, find the coordinate value (x ″) of the distance pixel _d,ζ ,y″ _d,ζ ) The nearest black connected domain is denoted as the ζ -th black background extraction image P' _ζ,b The ζ th parallelogram background black connected domain Ω _ζ ；

Step 6.1.5, extracting image P 'from the zeta-th black background' _ζ,b Upper extraction of zeta-th parallelogram background black connected domain omega _ζ Obtaining the zeta-th parallelogram background black connected domain omega _ζ Inner and outer contour set V of _ζ ；

Step 6.1.6, compare contour set V _ζ The perimeter of the maximum perimeter is recorded as the zeta-th parallelogram background black connected domain omega _ζ Outer contour E of _ζ ；

Step 6.1.7, extract image P 'at the ζ th black background' _ζ,b In the above, willOuter contour E _ζ The gray value of all the pixels within is assigned to 255, and the outer contour E _ζ Except for the gray values of all the pixel points, the gray values are assigned to be 0, and therefore the zeta th parallelogram binaryzation image P' without the complex background is obtained _ζ (ii) a Leading the parallelogram to be a binary image P ″' without complex background at the zeta th _ζ In (1), the pixel coordinate value (x ″) of the centroid is included _d,ζ ,y″ _d,ζ ) The gray value of all pixel points in the parallelogram coding mark is set to be 255 (namely white), and the gray value contains the coordinate value (x ″) of the centroid pixel _d,ζ ,y″ _d,ζ ) The gray values of all the pixel points outside the parallelogram coding mark are set to be 0 (namely black).

There is also provided a computer readable storage medium comprising a computer program for use in conjunction with an electronic device having image processing functionality, said computer program being executable by a processor to perform the decoding method.

Compared with the prior art, the invention has the following beneficial effects:

1. the coding pattern in the parallelogram coding mark provided by the invention is composed of a positioning pattern, an orientation pattern and a coding combination pattern, wherein the positioning pattern and the orientation pattern can automatically judge the relative rotation direction of a measured target and a video camera in the processes of camera calibration, three-dimensional data splicing and the like, so that the flexibility is improved;

2. the parallelogram coding mark provided by the invention comprises four coding combination patterns, and each coding combination pattern comprises three coding combination coding patterns, wherein 4096 different codes can be generated by the parallelogram coding mark provided by the invention, so that the capacity is expanded, and the three-dimensional data splicing can be completed aiming at a target with a larger size;

3. in the parallelogram coding mark provided by the invention, the positions of the orientation pattern, the positioning pattern and each coding combination pattern and the positions of each coding combination coding pattern are distinguished by a method of graphic geometric relationship and vector product, so that the parallelogram coding mark has stronger robustness and higher precision for affine transformation images;

4. the decoding method of the parallelogram coding mark based on the geometric relationship of the graph can effectively remove the interference of noise, and the method for removing the complex background greatly improves the accuracy of decoding;

5. the decoding method of the parallelogram coding mark based on the graph geometric relationship has high efficiency and high decoding speed, and can realize real-time decoding.

Drawings

FIG. 1 is a schematic diagram of a parallelogram encoding flag of the present invention;

FIG. 2 is a schematic diagram of arrangement of each vertex, each corner, and each coding pattern in each coding combination in the parallelogram coding flag according to the present invention;

FIG. 3 is a coded logo grayscale image P ₁ A display diagram of;

FIG. 4 shows a binary image P of a coded mark ₂ A display diagram of;

FIG. 5 shows a binary image P with a coded mark ₂ Extracting a display diagram of a circular black connected domain centroid set A;

FIG. 6 is the No. 1 parallelogram binaryzation image P ″, without complex background _1,1 A display diagram of;

FIG. 7 is a diagram illustrating a method for extracting a 1 st parallelogram binaryzation image P ″, which has no complex background, by using Harris checkpoint detection algorithm _1,1 A sub-pixel coordinate result display diagram of the center corner point;

FIG. 8 is the No. 1 parallelogram binarized image P without complex background ₁ "and 1 st backup grayscale image P ₁ After image multiplication, the 1 st overlay result image T is obtained ₁ ；

FIG. 9 is a diagram of the 1 st overlay result image T extracted by Harris checkpoint detection algorithm ₁ A sub-pixel coordinate result display diagram of the center corner point;

FIG. 10 is a 1 st binarized image P without complex background coding marks ₁ "is shown;

FIG. 11 is a flowchart illustrating a decoding method according to the present invention.

Detailed Description

The following detailed description of the preferred embodiments of the present invention, taken in conjunction with the accompanying drawings, will make the advantages and features of the present invention more comprehensible to those skilled in the art, and will thus provide a clear and concise definition of the scope of the present invention.

Referring to fig. 1, a decoding method for a parallelogram coding mark based on a geometric graph relationship is disclosed, the coding mark is a square coding square, a parallelogram background pattern and a coding pattern are arranged on the surface of the coding square, the coding pattern is positioned inside the parallelogram background pattern, the coding pattern comprises a positioning pattern, an orientation pattern and a coding combination pattern, the orientation pattern and the positioning pattern are used for judging the direction of the parallelogram coding mark, the coding combination pattern is used for coding 4 angular points of the parallelogram coding mark, and each coding combination pattern is composed of a coding combination orientation pattern and three coding combination coding patterns.

The positioning pattern, the orientation pattern, all the code combination orientation patterns and all the code combination code patterns in the parallelogram code sign are all non-overlapping and non-communicating.

Marking one characteristic corner of any parallelogram coding mark as a vector to determine a first vertex o ″) ₁ Intersecting each other to form a vector defining a first vertex o ″' on the parallelogram-shaped coded symbol ₁ Determining a first edge N by recording any one edge of the first edge as a vector _v1 Determining a first edge N in the vector _v1 Marking the vertex of the upper parallelogram coding mark as a vector to determine a second point o ″) ₂ Wherein the vector determines the second point o ″) ₂ Determining the first vertex o ″' with the vector ₁ Are 2 points which are not coincident with each other, and the vector is recorded

To specify a vector

The positional relationship of the positioning pattern and the orientation pattern within the parallelogram-shaped coded marker is as follows: oriented pattern within parallelogram-encoded symbolsThe center points to the direction and the specified vector of the center of mass of the positioning pattern

In the same direction.

Marking the plane marked by the parallelogram coding as a mark plane P _t Determining the first vertex o' by the vector ₁ Making a prescribed vector for the starting point

The unit vector in the same direction is denoted as the 1 st predetermined unit vector

When the person looks at the parallelogram coding mark, the first vertex o' is determined by the vector ₁ At the mark plane P as the center of rotation _t Defining the 1 st unit vector

Counterclockwise rotation by an angle beta '(0 DEG < beta' < 90 DEG) to obtain a 2 nd prescribed unit vector

The first vertex o' is determined in space as a vector ₁ As a starting point, an

The unit vectors with the same direction are recorded as positive vectors

Respectively marking two characteristic angular points which are closest to an orientation pattern in the parallelogram coding mark as 1 st temporary vertex o ″ ₃ And the 2 nd temporary vertex o ″ ₄ (ii) a If vector

Cross-product specified vector

Direction of the resulting vector and the forward vector

Are in the same direction, they will be recorded as vectors

Auxiliary vector

If vector

Cross product specified vector

Direction of the obtained vector and forward vector

Are in the same direction, then vector is set

Is recorded as an auxiliary vector

The color of a parallelogram background pattern in the parallelogram coding mark is recorded as color 1, and the colors of an orientation pattern, a positioning pattern, all coding combination orientation patterns and all coding combination coding patterns in the parallelogram coding mark are color 2, wherein the color 1 is obviously different from the color 2. In this embodiment, the color of the parallelogram background pattern in the parallelogram coding mark is black (color 1), and the colors of the orientation pattern, the positioning pattern, all the coding combination orientation patterns and all the coding combination coding patterns in the parallelogram coding mark are white (color 2).

In the parallelogram coding mark, the minimum area in the areas of all the coding combination coding patterns is recorded as the minimum area of the coding combination coding patterns, and the sizes of the areas of the orientation patterns, all the coding combination orientation patterns and all the coding combination coding patterns in the parallelogram coding mark satisfy the following relations: the areas of all the coding combination coding patterns inside the parallelogram coding mark are smaller than the area of the orientation patterns inside the parallelogram coding mark, and the areas of all the coding combination orientation patterns inside the parallelogram coding mark are smaller than the minimum area of the coding combination coding patterns. In this embodiment, the orientation pattern is a circular connected domain with a diameter of 12 mm, the encoding combination orientation pattern is a circular connected domain with a diameter of 3 mm, the encoding combination encoding pattern is a circular connected domain with a diameter of 6 mm, and the positioning pattern is a circular connected domain with an outer circle diameter of 12 mm and an inner circle diameter of 5 mm, as shown in fig. 1.

The positions of the orientation pattern and the positioning pattern need to meet the following requirements: let the line segment connecting the geometric center of the orientation pattern and the geometric center of the orientation pattern be l ₁ Then line segment l ₁ Must be parallel to any one of the sides of the parallelogram lattice and line segment l ₁ Midpoint p of ₁ Near the geometric center point of the parallelogram lattice. The positions of the four code combinations need to satisfy the following requirements: segment l ₁ Midpoint p of ₁ The four line segments are respectively connected with the middle points of the four sides of the parallelogram grid to form four line segments, and the obtained four line segments divide the whole parallelogram grid into four areas. Then, the four encoded combined orientation patterns are located inside the four regions, respectively, and the geometric center of each encoded combined orientation pattern is located near the geometric center point of the respective region. And, among the four vertexes of the parallelogram grid, each vertex has and only has one coding combination orientation pattern closest to the vertex, namely, among the straight line distances between each vertex and the geometric centers of the four coding combination orientation patterns, there is and only one coding combination orientation pattern with the shortest straight line distance.

Meanwhile, of the four encoded combined orientation patterns, two encoded combined orientation patterns are significantly closer to the orientation pattern than to the positioning pattern. Likewise, the other two coded combined orientation patterns are significantly more distant from the positioning pattern than from the orientation patternAnd (4) approaching. As shown in FIG. 2, the positions of the coding patterns of 3 coding combinations in each coding combination should satisfy the following requirements: marking the orientation pattern as A ₁ With a geometric center of a ₁ The positioning pattern is A ₂ With a geometric center of a ₂ From a to a ₁ Starting from lead to a ₂ The directed line segment of (2) is called the normal vector α. In four code combinations, the Y code combination orientation pattern is marked as C _Y With a geometric center of c _Y Within the Yth coding combination, the 1 st coding pattern of the Yth coding combination is D _Y1 With a geometric center of d _Y1 The Y-th code is combined with the 2 nd code pattern to be D _Y2 With a geometric center of d _Y2 The Yth code combines the 3 rd code pattern into D _Y3 With a geometric center of d _Y3 . Combining orientation patterns C by the Yth code _Y Geometric center c of _Y Starting from the first code pattern D, the first code pattern D is respectively directed to the Y-th code pattern _Y1 Geometric center d of _Y1 The Y code combines the 2 nd coding pattern D _Y2 Geometric center d of _Y2 The Yth code combines the 3 rd coding pattern D _Y3 Geometric center d of _Y3 Are called vectors y _Y1 Vector gamma _Y2 Vector gamma _Y3 . Simultaneous vector gamma _Y1 Vector gamma _Y2 Vector gamma _Y3 The vector products gamma of the respective standard vectors alpha _Y1 ×α、γ _Y2 ×α、γ _Y3 The result of x α is η _Y1 、η _Y2 、η _Y3 As shown in the following formula:

then η _Y1 、η _Y2 、η _Y3 There is a need for significant differences in die length or direction; wherein Y =1,2,3,4.

As shown in fig. 11, the digital image processing method is used to obtain the coding sequence number and the sub-pixel coordinates of each feature corner point in the image containing N (in this embodiment, N is 10) parallelogram coding marks captured by the camera, and further complete the decoding of the parallelogram coding marks as follows:

step 1.1, utilizing 10 parallelogram coding marks placed in a camera shooting space to obtain a coding mark image, wherein the coding mark image comprises 10 parallelogram coding marks;

step 1.2, establishing a pixel coordinate system of an angular point: in the shot image containing 10 parallelogram coding marks, taking the upper left corner of the image as the origin o of an angular point pixel coordinate system, taking the upper left corner of the image as the x-axis direction of the angular point pixel coordinate system from left to right, and taking the upper left corner of the image as the y-axis direction of the angular point pixel coordinate system from top to bottom, thereby establishing an angular point pixel coordinate system o-xy;

step 2.1, carrying out gray level processing on the coding mark image to obtain a coding mark gray level image P ₁ (ii) a Wherein the code marks a grayscale image P ₁ An 8-bit gray scale map, as shown in FIG. 3;

step 2.2, the grey image P of the coding mark ₁ Copying and backing up to obtain a grey level backup image P of the coding mark ₁ ′；

Step 2.3, the grey level backup image P of the coding mark ₁ Performing binarization processing to make the background color of each parallelogram coding mark be black (namely the gray value is 0), the orientation pattern and the positioning pattern in each parallelogram coding mark be white (namely the gray value is 255), the color of the coding unit pattern can be white or black according to the coding rule of the parallelogram coding mark, and further obtaining a coding mark binary image P ₂ As shown in fig. 3;

step 3, setting a roundness threshold lambda '=1 of the black connected domain and a pixel point number threshold lambda' =30;

step 4.1, calculating a binary image P of the coding mark ₂ The circularity value of all black connected components in the black matrix _n And the number of pixel points c _n Wherein the circularity value of each black connected domain can be obtained by equation (1),

wherein，l _n Perimeter, s, representing the contour of the corresponding black connected component _n Represents the area corresponding to the black connected domain, n =1,2.., 10;

step 4.2, the black connected domain roundness threshold λ' =1 and the pixel number threshold λ =30 selected in this embodiment may be implemented on the coding mark binary image P ₂ Obtaining black connected domains of the centers of 10 directional circular rings corresponding to the 10 parallelogram coding marks respectively; respectively recording black connected domains at the centers of 10 directional circular rings as a binary image P of a coding mark ₂ Middle ring black connected domain M ₁ Ring black connected domain M ₂ Ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ And connecting the ring black connected region M ₁ Ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ Sequentially putting into a ring black connected domain set A', namely a ring black connected domain M ₁ Circular ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ Respectively being the 1 st element in the ring black connected domain set A' element No. 2, \ 8230, element No. 10; thereby obtaining a coded mark binarized image P ₂ The black connected domain inside the orientation circle in each parallelogram encoding sign in (1); wherein, the coded mark binarizes the image P ₂ Middle ring black connected domain M ₁ Circular ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ Are all less than 1, and the black connected domain M of the ring ₁ Ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ The number of the pixel points is more than 30;

step 4.3, taking an integer variable i and assigning i =1;

step 4.4, calculating the ith ring black connected domain M in the ring black connected domain set A _i The coordinates of the centroid pixel are marked as o ″ _d,i (x″ _d,i ,y″ _d,i ) The obtained black connected domain of the ith ringM _i Centroid pixel coordinate o ″) _d,i (x″ _d,i ,y″ _d,i ) Putting the circular ring black connected domain centroid set A as the ith element of the circular ring black connected domain centroid set A;

step 4.5, judging whether i is smaller than N, if i is smaller than 10, assigning i +1 to i, and returning to the step 4.4 for sequential execution; otherwise, obtaining the binary image P of the coding mark ₂ Middle ring black connected domain M ₁ Ring black connected domain M ₂ Circular ring black connected domain M ₃ 823060, circular black connected domain M ₁₀ Centroid pixel coordinate o ″) _d,1 (775,135)、o″ _d,2 (x″ _d,2 ,y″ _d,2 )、…、o″ _d,10 (x″ _d,10 ,y″ _d,10 ) And sequentially putting the circular ring black connected domain centroid sets A into a circular ring black connected domain centroid set A, as shown in FIG. 5;

step 5, an integer variable zeta is taken and assigned zeta =1;

step 6.1, binarization image P of coding mark ₂ Performing copy backup twice to obtain a 1 st backup binary image P 'of a 1 st group' _1,1 And a 1 st group of 2 nd backup binary image P' _1,2 ；

Step 6.2, utilizing a complex background algorithm for removing coding marks to carry out binarization on the 1 st backup binary image P 'of the 1 st group' _1,1 Digital image processing is carried out, and then a 1 st parallelogram binaryzation image P ″, which has no complex background, is obtained _1,1 (ii) a The 1 st parallelogram binaryzation image P ″, which has no complex background ₁ In 1, the gray values of all the pixels inside the parallelogram coding mark containing the centroid pixel coordinate values (775, 135) are set to be 255 (i.e. white), and the gray values of all the pixels outside the parallelogram coding mark containing the centroid pixel coordinate values (775, 135) are set to be 0 (i.e. black);

step 7.1, coding the mark gray level image P ₁ Performing copy backup to obtain the 1 st backup grayscale image P ₁ ″′；

Step 7.2, utilizing Harris corner detection algorithm to binarize an image P' on the 1 st parallelogram without complex background _1,1 Extract 4 corner pointsAnd is marked as the 1 st parallelogram binaryzation image P ″, which has no complex background _1,1 Upper 4 alternative corner points C ″) _1,1 (0.6904×10 ³ ,0.1910×10 ³ )、C″ _1,2 (0.6899×10 ³ ,0.0238×10 ³ )、C″ _1,3 (0.8575×10 ³ ,0.0237×10 ³ )、C″ _1,4 (0.8573×10 ³ ,0.1908×10 ³ )；

Step 7.3, in the No. 1 complex background-free parallelogram binaryzation image P ″ _1,1 The first parallelogram binaryzation image P ″, which has no complex background and is made of black connected domain corrosion _1,1 The white connected domain in the image is enlarged, so that when image multiplication operation is carried out subsequently, the complete gray pattern of the parallelogram coding mark containing the coordinate values (775, 135) of the centroid pixel can be extracted from the operation result image, and 4 characteristic angular points of the gray pattern are completely stored; and the first parallelogram binaryzation image P' without complex background _1,1 The image obtained after the processing in this step is recorded as the 1 st operation binary image P ″ _1,2 ；

Step 7.4, calculating the 1 st binary image P ″ _1,2 Setting the value of the part with the middle gray scale value of 255 as 1, and calculating the 1 st binary image P ″ _1,2 Setting the value of the part with the middle gray scale value of 0 as 0, and then utilizing the 1 st operation to binarize the image P ″) _1,2 And 1 st backup grayscale image P ₁ ' multiply, and record the result image as the 1 st overlay result image T ₁ As shown in fig. 8; the 1 st overlay result image T ₁ Only contains the image information of the parallelogram coding mark containing the coordinate value (775, 135) of the centroid pixel, and removes the image information of other parallelogram coding marks;

step 7.5, covering the result image T at the 1 st by using Harris corner detection algorithm ₁ Extracting sub-pixel coordinates of all corner points and storing the sub-pixel coordinates into a 1 st initially selected corner point sub-pixel coordinate set B ₁ Performing the following steps; wherein, the 1 st initially selected corner sub-pixel coordinate set B ₁ Storing the 1 st overlay result image T ₁ Contains the parallel of the centroid pixel coordinate values (775, 135)Sub-pixel coordinate values of 4 characteristic angular points of the quadrilateral coding mark and sub-pixel coordinate values of interference angular points, wherein the interference angular points are interference angular points except the characteristic angular points;

step 7.6, taking an integer variable i and assigning i =1;

7.7, performing sub-pixel coordinate set B on the 1 st initially selected corner point ₁ Finding the coordinate value (x ″) of the distance sub-pixel _1,i ,y″ _1,i ) Nearest corner sub-pixel coordinate is recorded as C' _1,i (x′ _1,i ,y′ _1,i )；

7.8, judging whether i is smaller than 4, if i is smaller than 4, assigning i +1 to i and returning to execute the step 7.7; otherwise, the 1 st overlay result image T is already obtained at this time ₁ Sub-pixel coordinates C 'of the 4 characteristic corner points of the parallelogram-coded logo comprising centroid pixel coordinate values (775, 135) thereon' _1,1 (0.6927×10 ³ ,0.1889×10 ³ )、C′ _1,2 (0.6926×10 ³ ,0.0256×10 ³ )、C′ _1,3 (0.8559×10 ³ ,0.0257×10 ³ )、C′ _1,4 (0.8558×10 ³ ,0.1890×10 ³ ) As shown in fig. 9, and step 8.1 is performed;

step 8.1, backing up the binary image P 'at the 2 nd backup of the 1 st group' _1,2 Marking the parallelogram coding mark containing the centroid pixel coordinate values (775, 135) as a 1 st parallelogram coding mark;

step 8.2, in the 2 nd backup binary image P 'of the 1 st group' _1,2 The sub-pixel coordinate values are (0.6904 × 10) ³ ,0.1910×10 ³ )、(0.6899×10 ³ ,0.0238×10 ³ )、(0.8575×10 ³ ,0.0237×10 ³ )、(0.8573×10 ³ ,0.1908×10 ³ ) 4 pixel points as corner quadrangle S of the 1 st parallelogram coding mark ₁ 4 vertices of the parallelogram, and connecting the 4 vertices to obtain the corner point quadrangle S of the 1 st parallelogram coding mark ₁ ；

Step 9, in the 2 nd backup binarization image P 'of the 1 st group' _1,2 In the upper step, the centroid set of the black connected domain of the circular ring is searchedA, a white connected domain with the nearest 1 st orientation circular ring centroid pixel coordinate value (775, 135) is assigned with the gray value of the white connected domain as 0;

step 10, backing up binary image P 'in the 2 nd backup of the 1 st group' _1,2 Upper, the corner point quadrangle S of the 1 st parallelogram code mark is kept ₁ The gray values of all the internal pixel points are unchanged, and the corner point quadrangle S of the 1 st parallelogram coding mark is coded ₁ The gray values of all other pixel points are assigned to be 255;

step 11.1, backing up the binary image P 'at the 2 nd backup of the 1 st group' _1,2 Extracting the corner point parallelogram S of the first parallelogram code mark ₁ All the inner and outer contours in (1) are marked as the contour set D of the 1 st parallelogram coding mark ₁ ；

Step 11.2 statistics of the contour set D in the 1 st parallelogram encoding flag ₁ The number of pixel points contained in each contour is recorded as the positioning circle contour G in the 1 st parallelogram coding mark ₁ Calculating the coordinates of the centroid pixel of the binary image and recording the coordinates as a No. 1 set of No. 2 backup binary image P' _1,2 Localizer circle centroid pixel coordinate o 'within the top 1 st parallelogram encoding landmark' _l,1 (775,81)；

Step 12, counting the outline set D in the 1 st parallelogram coding mark ₁ The number of pixel points contained in each contour is recorded as the 2 nd backup binary image P 'of the 1 st group by 4 contours containing the minimum number of pixel points' _1,2 Coded combined orientation pattern profile Z in the upper 1 st parallelogram-coded logo _1,1 Coded composite directional pattern profile Z _1,2 Coded composite directional pattern profile Z _1,3 Coded composite directional pattern profile Z _1,4 And respectively calculate the coordinates of the centroid pixel

Step 13.1 encoding the set D of contours in the landmark from the 1 st parallelogram ₁ Removing two contours with the largest number of pixel pointsAnd 4 contours with the least number of pixel points, the number of remaining contours k ₁ =1; due to kappa ₁ Not equal to 0, the 1 contour is the contour of the coded combined coded pattern in the 1 st parallelogram coded mark and is marked as the coded combined pattern

Step 14.1, giving an integer variable i and an initial value i =1;

step 14.2, backing up the binary image P 'at the 2 nd backup of the 1 st group' _1,2 Up-computing coded composite pattern

Of centroid pixel coordinate o' ₁ ⁱ (x′ ₁ ⁱ ,y′ ₁ ⁱ ) Assigning i +1 to i again and continuing to execute the step until i > 1; whereby a corresponding coded combination pattern can be obtained

Coded composite pattern

8230code combined pattern

Centroid pixel coordinates of

In this embodiment, the code pattern can be obtained

Of is o' ₁ ¹ (728,45)；

Step 15.1, backup binarization image P 'at No. 1 group No. 2' _1,2 In the above, the characteristic corner points of the 1 st parallelogram coding mark located in the 1 st coding region, the 2 nd coding region, the 3 rd coding region and the 4 th coding region are recordedRespectively has a sub-pixel coordinate of C _1,1 (x _1,1 ,y _1,1 )、C _1,2 (x _1,2 ,y _1,2 )、C _1,3 (x _1,3 ,y _1,3 )、C _1,4 (x _1,4 ,y _1,4 ) (ii) a And the binary image P 'is copied at the 2 nd backup of the 1 st group' _1,2 In the above, the pixel points with pixel coordinate values of (775, 135) are recorded as the 1 st set of 2 nd backup binary image P' _1,2 Locating ring centroid o 'of upper 1 parallelogram coding mark' _d,1 (775, 135), the sub-pixel coordinate value is (0.6927 × 10) ³ ,0.1889×10 ³ )、(0.6926×10 ³ ,0.0256×10 ³ )、(0.8559×10 ³ ,0.0257×10 ³ )、(0.8558×10 ³ ,0.1890×10 ³ ) The 4 pixel points are respectively marked as C _1，1 (0.6927×10 ³ ,0.1889×10 ³ )、C″′ _1,2 (0.6926×10 ³ ,0.0256×10 ³ )、C″′ _1,3 (0.8559×10 ³ ,0.0257×10 ³ )、C″′ _1,4 (0.8558×10 ³ ,0.1890×10 ³ )；

Step 15.2, in the 1 st group, the 2 nd backup binary image P' _1,2 Direction vector in the 1 st parallelogram code flag

As can be derived from the equation (2),

step 16, backing up binary image P 'in the 2 nd backup of the 1 st group' _1,2 In the above, 4 pixel points C' _1,1 (0.6927×10 ³ ,0.1889×10 ³ )、C″′ _1,2 (0.6926×10 ³ ,0.0256×10 ³ )、C″′ _1,3 (0.8559×10 ³ ,0.0257×10 ³ )、C″′ _1,4 (0.8558×10 ³ ,0.1890×10 ³ ) Pixel coordinate o 'of middle distance positioning circle centroid' _l,1 (775, 81) the nearest 2 pixels, and respectively marked as C _1,1min (x _1,1min ,y _1,1min ) And C _1,2min (x _1,2min ,y _1,2min ) In this embodiment, it is C _1,1min (0.6926×10 ³ ,0.0256×10 ³ ) And C _1,2min (0.8559×10 ³ ,0.0257×10 ³ )。

Calculating the 1 st judgment vector in the 1 st parallelogram coding flag by the formulas (3) and (4)

And 2 nd decision vector

And calculating the area judgment sine value 1sin alpha by the formula (5) and the formula (6) ₁ Sum area judgment sine value 2sin beta ₁ ；

In this embodiment:

sinα ₁ ＞0,sinβ ₁ if < 0, then C _1,2min (0.8559×10 ³ ,0.0257×10 ³ ) For the characteristic corner point of the 3 rd coding region in the 1 st parallelogram coding mark, C _1,2min (0.8559×10 ³ ,0.0257×10 ³ ) Is assigned to C _1,3 (x _1,3 ,y _1,3 )；C _1,1min (0.6926×10 ³ ,0.0256×10 ³ ) Is composed ofFeature corner of the 2 nd coding region in the 1 st parallelogram coding index, C _1,1min (0.6926×10 ³ ,0.0256×10 ³ ) Is assigned to C _1,2 (x _1,2 ,y _1,2 )；

Step 17, 1 st group of 2 nd backup binary image P' _1,2 Through the above step 16, at 4 pixel points C ″' _1,1 (0.6927×10 ³ ,0.1889×10 ³ )、C″′ _1,2 (0.6926×10 ³ ,0.0256×10 ³ )、C″′ _1,3 (0.8559×10 ³ ,0.0257×10 ³ )、C″′ _1,4 (0.8558×10 ³ ,0.1890×10 ³ ) Finding out the sub-pixel coordinate values of the characteristic angular points of the 2 nd coding region and the 3 rd coding region in the 1 st parallelogram coding mark and respectively assigning the sub-pixel coordinate values to the C _1,2 (x _1,2 ,y _1,2 ) And C _1,3 (x _1,3 ,y _1,3 ) (ii) a Respectively assigning the sub-pixel coordinate values of the rest 2 pixel points to the 1 st temporary coordinate value of the 1 st parallelogram coding mark, and recording as C' _1,5 (0.6927×10 ³ ,0.1889×10 ³ ) And 2 nd temporary coordinate value of C' _1,6 (0.8558×10 ³ ,0.1890×10 ³ ) (ii) a Determining the 3 rd judgment vector in the 1 st parallelogram coding flag according to the formulas (7) and (8)

And 4 th judgment vector

Step 18, in the 2 nd backup binary image P 'of the 1 st group' _1,2 According to step 17The 3 rd judgment vector calculated in the step

And 4 th judgment vector

The 3 rd region judgment sine value sin omega can be obtained through the formula (9) and the formula (10) ₁ And 4. Area judgment sine value sin xi ₁ ；

In the present embodiment, a parallel vector sine threshold g =0.01,sin ξ is set in advance ₁ If < g, the judgment is regarded as the 4 th judgment vector

And direction vector

Parallel: namely sin omega ₁ ≠0,sinξ ₁ 0 is added, then C' _1,6 (0.8558×10 ³ ,0.1890×10 ³ ) Namely the characteristic corner point of the 4 th coding region in the 1 st parallelogram coding mark, and C' _1,6 (0.8558×10 ³ ,0.1890×10 ³ ) Is assigned to C _1,4 (x _1,4 ,y _1,4 )；C′ _1,5 (0.6927×10 ³ ,0.1889×10 ³ ) C 'is a characteristic corner point of a 1 st coding region in a 1 st parallelogram coding mark' _1,5 (0.6927×10 ³ ,0.1889×10 ³ ) Is assigned to C _1,1 (x _1,1 ,y _1,1 )；；

Step 19, backing up binary image P 'in the 2 nd backup of the 1 st group' _1,2 Will be defined by the characteristic corner point C _1,1 (0.6927×10 ³ ,0.1889×10 ³ ) Angular point C of orientation feature _1,4 (0.8558×10 ³ ,0.1890×10 ³ ) Is recorded as a vector

And is derived from equation (11):

taking an integer variable i and assigning i =1;

step 20, backing up binary image P 'in the 2 nd backup of the 1 st group' _1,2 According to the coding combination, the pattern centroid is oriented

(Vector)

And direction vector

Determining cosine values in combination with coding

And

in this embodiment, according to the above formula, the specific calculation result is:

due to the fact that

And is

Then

To orient the pattern centroid for the code combination of code region 1, will

Value to o' _z ^1,1 (x′ _z ^1,1 ,y′ _z ^1,1 )；

Due to the fact that

And is provided with

Then

Value to o' _z ^1,2 (x′ _z ^1,2 ,y′ _z ^1,2 )；

Due to the fact that

And is

Then the

Value to o' _z ^1,3 (x′ _z ^1,3 ,y′ _z ^1,3 )；

Due to the fact that

And is provided with

Then

Assigned to o' _z ^1,4 (x′ _z ^1,4 ,y′ _z ^1,4 )；

Step 21, assigning the integer variable i with i =1 again;

step 22, defining 4 two-dimensional floating-point type arrays Cr ₁ ¹ [3][2]、Cr ₁ ² [3][2]、Cr ₁ ³ [3][2]、Cr ₁ ⁴ [3][2]Initializing all elements in the 4 two-dimensional arrays to be assigned as-1; taking 4 integer variables b ₁ ¹ ,b ₁ ² ,b ₁ ³ ,b ₁ ⁴ Initialization of b ₁ ¹ ＝0,b ₁ ² ＝0,b ₁ ³ ＝0,b ₁ ⁴ ＝0；

Step 23, in the 1 st group of 2 nd backup binary image P' _1,2 Calculating the circular contour S of the coded and combined coded pattern mark in the 1 st parallelogram code mark ₁ ¹ Centroid pixel coordinate o' ₁ ¹ (728, 45) respectively locate the center of mass o 'of the circle' _l,1 Formed 1 st group 1 st discrimination vector

Calculated by the following equations (16) and (17)

Due to the fact that

And is

Reassign value to b ₁ ² =1, and i > 0, step 24.1 is performed directly;

step 24.1, assigning the integer variable i with the value i =1 again;

step 24.2, the results obtained in this example are as follows:

(1)

get integer variable

And

and assign values

(2)

Get the set Cr ₁ ² [3][2]The elements in the same row are not Cr of-1 ₁ ² [0][0]And Cr ₁ ² [0][1]And respectively assigned to _1,1 (x _1,1 ,y _1,1 ) And the abscissa and the ordinate of the sine value sin phi are calculated ₁ ²

Due to sin phi ₁ ² =0, take the integer variable

And

and assign values

(3)

Get integer variable

And

and assign values

(4)

Get integer variable

And

and assign values

Step 25, obtaining the number W of the 1 st parallelogram coding mark through the formula (21) ₁ ：

W ₁ ＝V ₁ ^T ·U＝8 (21)

Wherein, the column vector U = (2) ⁰ ,2 ¹ ,2 ² ,...2 ¹¹ ) ^T Column vector V ₁ ＝(0,0,0,1,0,0,0,0,0,0,0,0) ^T 。

Step 26, marking the 1 st parallelogram coding mark belonging to the sigma-th coding region (wherein sigma)Code number L of the calibration corner points of =1,2,3,4) ₈ ^σ Wherein the lower subscript 8 is the calibration corner point L ₈ ^σ The value of the upper corner mark sigma represents the calibrated corner point L ₈ ^σ The sigma-th coding region;

in this embodiment, the sub-pixel coordinate values of 4 characteristic corners and the corresponding corner coding numbers on the found parallelogram coding sign with the coding number of 8 are as follows:

the code number is L ₈ ¹ The sub-pixel coordinate value of the characteristic corner is C _1,1 (0.6927×10 ³ ,0.1889×10 ³ )；

The code number is L ₈ ² The sub-pixel coordinate value of the characteristic corner is C _1,2 (0.6926×10 ³ ,0.0256×10 ³ )；

The code number is L ₈ ³ The sub-pixel coordinate value of the characteristic corner is C _1,3 (0.8559×10 ³ ,0.0257×10 ³ )；

The code number is L ₈ ⁴ The sub-pixel coordinate value of the characteristic corner is C _1,4 (0.8558×10 ³ ,0.1890×10 ³ )；

Step 27, judging whether zeta is smaller than 8, if zeta +1 is smaller than 8, reassigning zeta +1 to zeta, and returning to step 6.1 to execute in sequence; if not, finding the sub-pixel coordinate values and the corresponding coding numbers of the 4 characteristic angular points on each parallelogram coding mark.

The 1 st backup binary image P 'of the 1 st group obtained in the step 6.1 is processed' _1,1 And taking the 1 st circular ring black connected domain centroid pixel coordinate value (775, 135) in the circular ring black connected domain centroid set A as an input condition, and removing the 1 st group 1 st backup binary image P 'by using a coding mark removing complex background algorithm' _1,1 And finally obtaining a 1 st parallelogram binaryzation image P without the complex background ₁ "the steps are as follows:

step 6.1.1, taking an integer variable i and assigning i =1;

step 6.1.2, backup binary image P 'at the No. 1 st group' _1,1 Searching a white connected domain nearest to the pixel coordinate values (775, 135) and marking as a 1 st backup binary image P 'of the 1 st group' _1,1 The ith circular ring white connected domain M' _i And the ith circular ring white connected domain M' _i All the pixel values of (a) are assigned to 0;

step 6.1.3, judging whether i is smaller than 10, if i is smaller than 10, assigning i +1 to i, and returning to the step 6.1.2 for sequential execution; otherwise, the 1 st backup binary image P 'of the 1 st group has been obtained' _1,1 The directional rings in all the parallelogram coding marks are black; and 1 st backup binary image P 'of the 1 st group' _1,1 The image obtained after the processing of the present step 6.1.3 was designated as the 1 st black background extracted image P' _1,b And performing step 6.1.4;

step 6.1.4, extracting image P 'from the 1 st black background' _1,b Of all the black connected components on the image plane, the black connected component closest to the pixel coordinate values (775, 135) is found and is marked as the 1 st black background extraction image P' _1,b The 1 st parallelogram background black connected domain omega ₁ ；

Step 6.1.5, extracting image P 'from the 1 st black background' _1,b Extracting the 1 st parallelogram background black connected domain omega ₁ All the inner and outer contours of the first parallelogram to obtain the 1 st black connected domain omega of the parallelogram background ₁ Inner and outer contour set V of ₁ ；

Step 6.1.6, compare contour set V ₁ The perimeter of all the contours in the graph is recorded as the 1 st parallelogram background black connected domain omega ₁ Outer contour E of ₁ ；

Step 6.1.7, extracting image P 'from the 1 st black background' _1,b Method for distinguishing black connected domain omega of parallelogram background by ray method ₁ Outer contour E of ₁ All other pixels and the outline E ₁ All the inner pixel points and the outer contour E ₁ The gray value of all the pixels within is assigned to 255, and the outer contour E ₁ Except that the gray values of all the pixel points are assigned to be 0, and then a 1 st parallelogram binary image without complex background is obtainedLike P ₁ "; binary image P of No. 1 parallelogram without complex background ₁ "in this case, all pixel values inside the parallelogram-shaped code flag containing the centroid pixel coordinate values (775, 135) are set to 255, and all pixel values outside the parallelogram-shaped code flag containing the centroid pixel coordinate values (775, 135) are set to 0, as shown in fig. 10.

The decoding method of parallelogram coding mark based on graph geometry relation provided by the invention needs to compile corresponding computer program and execute the program on the computer to realize the corresponding operation processing and logic control function, therefore, the invention also provides a computer readable storage medium, comprising the computer program used in combination with the electronic device with image processing function, the computer program can be executed by the processor to realize the decoding method.

The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes performed by the present specification and drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims

1. A decoding method of a parallelogram coding mark based on a graph geometric relationship is characterized in that the coding mark is a square coding square, a parallelogram background pattern and a coding pattern are arranged on the surface of the coding square, the coding pattern is positioned inside the parallelogram background pattern, the coding pattern comprises a positioning pattern, an orientation pattern and a coding combination pattern, the orientation pattern and the positioning pattern are used for judging the direction of the parallelogram coding mark, the coding combination pattern is used for coding each corner point of the parallelogram coding mark, and each coding combination pattern is composed of one coding combination orientation pattern and three coding combination coding patterns, and the decoding method is characterized in that: the method for processing the digital image is used for obtaining the coding sequence number and the sub-pixel coordinates of each characteristic corner point in the image which is shot by the camera and contains N parallelogram coding marks, and further the steps for decoding the parallelogram coding marks are as follows:

step 1.2, establishing a pixel coordinate system of an angular point;

step 2.1, carrying out gray level processing on the coding mark image to obtain a coding mark gray level image P ₁ ；

Step 2.2, 8 bits of gray level image P of coding mark ₁ Copying and backing up to obtain a grey level backup image P of the coding mark ₁ ′；

Step 2.3, the grey level backup image P of the coding mark ₁ ' carry on the binarization processing to get the binary image P of the coding mark ₂ ；

step 4.1, calculating a binary image P of the coding mark ₂ Circularity value λ of all black connected components in the image _n And the number of pixel points c _n Wherein the circularity value of each black connected domain is given by equation (1),

step 4.2, selecting proper black connected domain roundness threshold lambda 'and pixel point number threshold lambda' to encode the mark binary image P ₂ Black connected domains of the centers of N directional circular rings corresponding to the N parallelogram coding marks are obtained and are placed into a circular ring black connected domain set A';

wherein, the coded mark binarizes the image P ₂ The circularity value of each circular ring black connected domain is smaller than the circularity threshold lambda', and the number of pixel points of each circular ring black connected domain is largeA threshold value lambda' of the number of pixels is set;

step 4.3, taking an integer variable i and assigning i =1;

step 4.4, calculating the ith ring black connected domain M in the ring black connected domain set A _i The coordinates of the centroid pixel are marked as o ″ _d,i (x″ _d,i ,y″ _d,i ) Obtaining the ith ring black connected domain M _i Centroid pixel coordinate o ″) _d,i (x″ _d,i ,y″ _d,i ) The ith element is taken as a circular black connected domain centroid set A;

step 4.5, judging whether i is smaller than N, if i is smaller than N, assigning i +1 to i, and returning to the step 4.4 to execute in sequence; otherwise, obtaining the binary image P of the coding mark ₂ The center of mass pixel coordinate o' of each ring black connected domain _d,1 (x″ _d,1 ,y″ _d,1 )、o″ _d,2 (x″ _d,2 ,y″ _d,2 )、…、o″ _d,N (x″ _d,N ,y″ _d,N ) Sequentially putting the circular black connected domain mass center sets into a circular black connected domain mass center set A;

step 5, taking an integer variable zeta and assigning zeta =1;

step 6.1, binarizing the image P with the coding mark ₂ Performing copy backup twice to respectively obtain a 1 st backup binary image P 'of the zeta group' _ζ,1 And the 2 nd backup binary image P 'of the zeta th group' _ζ,2 ；

Step 6.2, carrying out binarization on the 1 st backup binary image P 'of the zeta th group by using a complex background algorithm for removing coding marks' _ζ,1 Digital image processing is carried out to obtain the zeta th parallelogram binaryzation image P ″, without complex background _ζ,1 ；

Step 7.1, coding the mark gray level image P ₁ Copying and backing up again to obtain a zeta-th backup gray level image P' _ζ ；

Step 7.2, utilizing a corner detection algorithm to perform binarization on the zeta th parallelogram binary image P' without complex background _ζ,1 Extracting sub-pixel coordinates of 4 corner points, and marking as the zeta-th complex background-free parallelogram binaryzation image P ″ _ζ,1 Upper 4 alternative corner points C ″) _ζ,1 (x″ _ζ,1 ,y″ _ζ,1 )、C″ _ζ,2 (x″ _ζ,2 ,y″ _ζ,2 )、C″ _ζ,3 (x″ _ζ,3 ,y″ _ζ,3 )、C″ _ζ,4 (x″ _ζ,4 ,y″ _ζ,4 )；

Step 7.3, obtaining a parallelogram binaryzation image P ″' without a complex background at the zeta th _ζ,1 Black connected domain corrosion is performed in the process, so that the zeta-th parallelogram binaryzation image P' without complex background _ζ,1 White connected domain in (1) is enlarged; and recording the processed image as the zeta-th operation binary image P ″ _ζ,2 ；

Step 7.4, the zeta-th operation binaryzation image P ″ _ζ,2 Setting the value of the part with the middle gray scale value of 255 as 1, and calculating the zeta-th binary image P ″ _ζ,2 Setting the value of the part with the middle gray value of 0 as 0, and then utilizing the zeta-th operation to binarize the image P ″) _ζ,2 And the ζ th backup grayscale image P' _ζ Performing multiplication, and recording the obtained result image as the Zeta-th coverage result image T _ζ ；

Step 7.5, covering the result image T at the zeta th by utilizing a corner point detection algorithm _ζ Extracting sub-pixel coordinates of all corner points and storing the sub-pixel coordinates into a zeta-th primary corner point sub-pixel coordinate set B _ζ Performing the following steps;

step 7.6, taking an integer variable i and assigning value i =1;

step 7.7, in the zeta th primary corner point sub-pixel coordinate set B _ζ Finding the coordinate value (x ″) of the distance sub-pixel _ζ,i ,y″ _ζ,i ) Nearest corner sub-pixel coordinate is recorded as C' _ζ,i (x′ _ζ,i ,y′ _ζ,i )；

7.8, judging whether i is smaller than 4, if i is smaller than 4, assigning i +1 to i and returning to execute the step 7.7; otherwise, the ζ -th overlay result image T has been obtained at this time _ζ Contains the coordinate value (x ″) of centroid pixel _d,ζ ,y″ _d,ζ ) Of 4 characteristic corner points of the parallelogram-coded logo C' _ζ,1 (x′ _ζ,1 ,y′ _ζ,1 )、C′ _ζ,2 (x′ _ζ,2 ,y′ _ζ,2 )、C′ _ζ,3 (x′ _ζ,3 ,y′ _ζ,3 )、C′ _ζ,4 (x′ _ζ,4 ,y′ _ζ,4 ) And step 8.1 is executed;

step 8.2, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Wherein the sub-pixel coordinate values are (x' _ζ,1 ,y′ _ζ,1 )、(x′ _ζ,2 ,y′ _ζ,2 )、(x′ _ζ,3 ,y′ _ζ,3 )、(x′ _ζ,4 ,y′ _ζ,4 ) The 4 pixel points are used as the corner quadrangle S of the zeta-th parallelogram coding mark _ζ 4 vertices of (a), connecting the 4 vertices to obtain a corner quadrangle S of the zeta-th parallelogram coding mark _ζ ；

Step 9, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 In the above, the pixel coordinate value (x ″) from the centroid is found _d,ζ ,y″ _d,ζ ) The nearest white connected domain, and the gray value of the white connected domain is assigned to be 0;

step 10, backing up a binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Corner quadrilateral S of upper Zeta parallelogram coding sign _ζ Keeping the gray values of all internal pixel points unchanged, and making the corner quadrangle S of the zeta-th parallelogram coding mark _ζ The gray values of all other pixel points are assigned to be 255;

step 11.1, backup binarization image P 'at 2 nd of zeta group' _ζ,2 In the above, the corner quadrangle S of the ζ -th parallelogram coding mark is extracted _ζ The inner and outer contours in (1) are marked as a contour set D in the zeta-th parallelogram coding mark _ζ ；

Step 11.2, counting the outline set D in the zeta-th parallelogram coding mark _ζ The number of pixel points included in each contour is recorded as the zeta-th groupNo. 2 backup binary image P' _ζ,2 Positioning circle contour G in upper Zeth parallelogram coding mark _ζ Calculating the coordinates of the centroid pixel of the binary image and recording the coordinates as the 2 nd backup binary image P 'of the zeta group' _ζ,2 Locating circle centroid pixel coordinates o 'within the upper zeta parallelogram encoding logo' _l,ζ (x′ _l,ζ ,y′ _l,ζ )；

Step 12, collecting the outline D in the zeta-th parallelogram coding mark _ζ The 4 contours with the least number of pixel points are recorded as the contour Z of the coding combination orientation pattern _ζ,1 Coded composite directional pattern profile Z _ζ,2 Coded composite directional pattern profile Z _ζ,3 Coded combined orientation pattern profile Z _ζ,4 And respectively calculate the centroid thereof

Step 13.1, according to the outline set D in the zeta th parallelogram coding mark _ζ In the method, two contours with the largest number of pixel points and 4 contours with the smallest number of pixel points are removed, and the number of remaining contours kappa is determined _ζ Is divided into the following cases: if κ _ζ =0, then step 14.1 is executed; otherwise this κ _ζ The profiles are the profiles of the coded combined coding patterns in the zeta-th parallelogram coding mark and are respectively marked as the coded combined patterns

Coded composite pattern

823060A combined coding pattern

Step 14.1, giving an integer variable i and an initial value i =1;

step 14.2, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Calculating the code pattern

Center of mass of

This step continues after i +1 is reassigned to i until i > κ _ζ Ending; whereby a corresponding coded composite pattern can be obtained

Coded composite pattern

823060A combined coding pattern

Centroid pixel coordinates of

Step 15.1, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 In the above, the sub-pixel coordinates of the characteristic corner points of the ζ -th parallelogram coding mark located in the 1 st coding region, the 2 nd coding region, the 3 rd coding region and the 4 th coding region are respectively C _ζ,1 (x _ζ,1 ,y _ζ,1 )、C _ζ,2 (x _ζ,2 ,y _ζ,2 )、C _ζ,3 (x _ζ,3 ,y _ζ,3 )、C _ζ,4 (x _ζ,4 ,y _ζ,4 ) (ii) a And backup binary image P 'at 2 nd of the zeta group' _ζ,2 The pixel coordinate value is (x ″) _d,ζ ,y″ _d,ζ ) The pixel point of (1) is recorded as a zeta-group 2 nd backup binary image P' _ζ,2 Directional ring centroid o 'of upper zeta-th parallelogram coding mark' _d,ζ (x′ _d,ζ ,y′ _d,ζ ) And seating the sub-pixelsThe standard value is (x' _ζ,1 ,y′ _ζ,1 )、(x′ _ζ,2 ,y′ _ζ,2 )、(x′ _ζ,3 ,y′ _ζ,3 )、(x′ _ζ,4 ,y′ _ζ,4 ) The 4 pixel points are respectively marked as C' _ζ,1 (x″′ _ζ,1 ,y″′ _ζ,1 )、C″′ _ζ,2 (x″′ _ζ,2 ,y″′ _ζ,2 )、C″′ _ζ,3 (x″′ _ζ,3 ,y″′ _ζ,3 )、C″′ _ζ,4 (x″′ _ζ,4 ,y″′ _ζ,4 )；

Step 15.2, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 The direction vector in the above, ζ th parallelogram code flag

As is derived from the formula (2),

step 16, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 On top, find 4 pixels C' _ζ,1 (x″′ _ζ,1 ,y″′ _ζ,1 )、C″′ _ζ,2 (x″′ _ζ,2 ,y″′ _ζ,2 )、C″′ _ζ,3 (x″′ _ζ,3 ,y″′ _ζ,3 )、C″′ _ζ,4 (x″′ _ζ,4 ,y″′ _ζ,4 ) Pixel coordinate o 'of middle-distance positioning circle centroid' _l,ζ (x′ _l,ζ ,y′ _l,ζ ) The nearest 2 pixels are respectively marked as C _ζ,1min (x _ζ,1min ,y _ζ,1min ) And C _ζ,2min (x _ζ,2min ,y _ζ,2min )；

By the formula (3) And (4) calculating the 1 st judgment vector in the ζ th parallelogram coding mark

And 2 nd decision vector

And calculating the 1 st area dividing sine value sin alpha according to the formulas (5) and (6) _ζ And 2 nd area division sine value sin beta _ζ ；

Step 17, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 And respectively assigning the sub-pixel coordinate values of the rest 2 pixel points to the temporary coordinate value 1 of the zeta-th parallelogram coding mark, and recording the temporary coordinate value as C' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) And temporary coordinate value 2, marked as C' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 )；

And 4 th judgment vector

Step 18, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 According to the 3 rd judgment vector

And 4 th judgment vector

The 3 rd region division sine value sin omega is obtained through the formula (9) and the formula (10) _ζ And 4 th area dividing sine value sin xi _ζ ；

Step 19, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 Will be composed of characteristic corner points C _ζ,1 (x _ζ,1 ,y _ζ,1 ) Directing deviceSymbolized corner point C _ζ,4 (x _ζ,4 ,y _ζ,4 ) Is recorded as a vector

And is derived from equation (11):

taking an integer variable i and assigning i =1;

step 20, backing up the binary image P 'at the 2 nd backup of the zeta group' _ζ,2 According to the coding combination, the pattern centroid is oriented

(Vector)

And direction vector

Coding combination judgment cosine value

And

the value taking condition of (1):

judging whether i meets the condition that i is less than 4, if i is less than 4, assigning i +1 to i, and returning to execute the step 20; otherwise, the center of mass of the coded combined orientation pattern of the 1 st coding region in the ζ th parallelogram coding mark is obtained

Coded combined orientation pattern centroid for coding region 2

Coded combined orientation pattern centroid for 3 rd coded region

And the coded combined orientation pattern centroid of the 4 th coded region

And proceeds to step 21;

step 21, assigning the integer variable i with i =1 again;

step 22, defining 4 two-dimensional floating-point type numerical groups

Initialization

Step 23, backup binarization image P 'at 2 nd of the zeta group' _ζ,2 In the above-mentioned manner,calculating the circular outline of the coded combined coded pattern mark in the Zeth parallelogram coded mark

Centroid pixel coordinates of

And the center o of the positioning circle' _l,ζ The ith difference vector of the formed ζ -th group

Calculated by the following equations (16) and (17)

Step 24.1, assigning the integer variable i with the value i =1 again;

step 24.2, according to

Taking integer variables

And

and assigning values;

if i is less than 4, assigning i +1 to i and returning to execute the step 24.2; otherwise, continuing to execute the step 25;

W _ζ ＝V _ζ ^T ·U (21)

Wherein, the column vector U = (2) ⁰ ,2 ¹ ,2 ² ,...2 ¹¹ ) ^T Column vector

Step 26, marking the code number of the characteristic corner point belonging to the sigma-th coding region in the zeta-th parallelogram coding mark as

Wherein W _ζ For calibrating angular points

The number of the coding square grid, the value of sigma represents the calibration angular point

(ii) a σ -encoding region to which σ =1,2,3,4;

thus, the coding numbers of the 4 feature corners in the ζ -th parallelogram coding mark and the corresponding sub-pixel coordinate values thereof are obtained:

coded number is

Coded number is

The code number is

Coded number is

Step 27, judging whether zeta is smaller than N, if yes, reassigning zeta +1 to zeta, and returning to step 6.1 to execute in sequence; if not, the sub-pixel coordinate values and the corresponding coding numbers of the 4 characteristic angular points on each parallelogram coding mark are found.

2. The decoding method according to claim 1, wherein: in step 6.1, obtaining a 1 st backup binary image P 'of the zeta th group' _ζ,1 And the Zeth circular ring black connected domain centroid pixel coordinate value (x ″) in the circular ring black connected domain centroid set A _d,ζ ,y″ _d,ζ ) Removing the 1 st backup binary image P 'of the zeta th group by using a complex background algorithm for removing a coding mark as an input condition' _ζ,1 And finally obtaining the zeta th parallelogram binaryzation image P' without complex background _ζ The steps are as follows:

step 6.1.1, taking an integer variable i and assigning i =1;

step 6.1.2, backup binary image P 'at the 1 st backup of the ζ th group' _ζ,1 Upper search distance pixel coordinate value (x ″) _d,ζ ,y″ _d,ζ ) The nearest white connected domain is recorded as the 1 st backup binary image P 'of the ζ th group' _ζ,1 The ith circular ring white connected domain M' _i And the ith circular ring white connected domain M' _i All the gray values are assigned to 0;

step 6.1.3, judging whether i is smaller than N, if i is smaller than N, assigning i +1 to i, and returning to the step 6.1.2 for sequential execution; if not, then,the zeta th group 1 st backup binary image P' _ζ,1 Setting the orientation circular rings in all the parallelogram coding marks to be black, and setting the 1 st backup binary image P 'of the zeta-th group' _ζ,1 The image processed in this step was recorded as the ζ -th black background extraction image P' _ζ,b And performing step 6.1.4;

Step 6.1.5, extracting image P 'from the ζ -th black background' _ζ,b Upper extraction of zeta-th parallelogram background black connected domain omega _ζ The zeta th parallelogram background black connected domain omega is obtained _ζ Inner and outer contour set V of _ζ ；

Step 6.1.6, compare contour set V _ζ The perimeter of the longest perimeter of all the contours in the drawing is recorded as the zeta-th parallelogram background black connected domain omega _ζ Outer contour E of _ζ ；

Step 6.1.7, extracting image P 'from the zeta-th black background' _ζ,b Upper, the outer contour E _ζ The gray value of all the pixels within is assigned to 255, and the outer contour E _ζ Except for the gray values of all the pixel points, the gray values are assigned to be 0, and therefore the zeta th parallelogram binaryzation image P' without the complex background is obtained _ζ (ii) a Leading the parallelogram to be a binary image P ″' without complex background at the zeta th _ζ In (1), the pixel coordinate value (x ″) of the centroid is included _d,ζ ,y″ _d,ζ ) The gray value of all pixel points in the parallelogram coding mark is set to be 255, and the gray value contains the coordinate value (x ″) of the centroid pixel _d,ζ ,y″ _d,ζ ) The gray values of all the pixel points outside the parallelogram coding mark are set to be 0.

3. The decoding method according to claim 1, wherein: in step 16, execution is performed in two cases:

(1) Sin alpha _ζ ＜0,sinβ _ζ If greater than 0, then C _ζ,1min (x _ζ,1min ,y _ζ,1min ) For the characteristic corner point of the 3 rd coding region in the ζ th parallelogram coding mark, C _ζ,1min (x _ζ,1min ,y _ζ,1min ) Is assigned to C _ζ,3 (x _ζ,3 ,y _ζ,3 )；C _ζ,2min (x _ζ,2min ,y _ζ,2min ) For the characteristic corner of the 2 nd coding region in the ζ th parallelogram coding index, C _ζ,2min (x _ζ,2min ,y _ζ,2min ) Is assigned to C _ζ,2 (x _ζ,2 ,y _ζ,2 )；

(2) If sin alpha _ζ ＞0,sinβ _ζ If < 0, then C _ζ,2min (x _ζ,2min ,y _ζ,2min ) For the characteristic corner of the 3 rd coding region in the ζ -th parallelogram coding index, C _ζ,2min (x _ζ,2min ,y _ζ,2min ) Is assigned to C _ζ,3 (x _ζ,3 ,y _ζ,3 )；C _ζ,1min (x _ζ,1min ,y _ζ,1min ) For the characteristic corner of the 2 nd coding region in the ζ th parallelogram coding index, C _ζ,1min (x _ζ,1min ,y _ζ,1min ) Is assigned to C _ζ,2 (x _ζ,2 ,y _ζ,2 )。

4. The decoding method according to claim 1, wherein: in step 18, it is performed in two cases:

(1) Sin omega _ζ ＝＝0,sinξ _ζ Not equal to 0, then C' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Namely C 'is the characteristic corner point of the 4 th coding area in the ζ th parallelogram coding mark' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Is assigned to C _ζ,4 (x _ζ,4 ,y _ζ,4 )；C′ _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) For the characteristic corner of the 1 st coding region in the ζ -th parallelogram coding indexC's' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Is assigned to C _ζ,1 (x _ζ,1 ,y _ζ,1 )；

(2) Sin ω _ζ ≠0,sinξ _ζ =0, then C' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Namely C 'is the characteristic corner point of the 4 th coding area in the ζ th parallelogram coding mark' _ζ,6 (x′ _ζ,6 ,y′ _ζ,6 ) Is assigned to C _ζ,4 (x _ζ,4 ,y _ζ,4 )；C′ _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) C 'is a characteristic corner point of a 1 st coding region in a zeta th parallelogram coding mark' _ζ,5 (x′ _ζ,5 ,y′ _ζ,5 ) Is assigned to C _ζ,1 (x _ζ,1 ,y _ζ,1 )。

5. The decoding method according to claim 1, wherein: in step 20, the execution is divided into four cases:

(1) If it is

And is

Then

Assign to

(2) If it is

And is

Then

To orient the pattern centroids for the coded combinations of the 2 nd coded region, will

Is assigned to

(3) If it is

And is

Then

To orient the pattern centroids for the encoded combination of the 1 st encoded region

Is assigned to

(4) If it is

And is provided with

Then the

Is assigned to

6. The decoding method according to claim 1, wherein: in step 23, the execution is divided into four cases:

(1) If it is

And is

Then make it give

Will be provided with

Is assigned to

(2) If it is

And is provided with

Then order

Will be provided with

Is assigned to

(3) If it is

And is

Then make it give

Will be provided with

Is assigned to

(4) If it is

And is

Then order

Will be provided with

Is assigned to

If i > k _ζ 1, then step 24 is executed, otherwise i +1 is assigned to i, and the step 23 is returned to.

7. The decoding method according to claim 1, wherein: in step 24.2, execution is divided into the following cases:

(1) If it is

Then take the integer variable

And

and assign values

(2) If it is

Will array

Elements in the same row which are not-1 are respectively marked as

And

and respectively assigned to _ζ,1 (x _ζ,1 ,y _ζ,1 ) And the abscissa and the ordinate of the computer, and calculating the sine value

And (4) judging:

1) If it is

Then get the integer variable

And

and assign values

2) If it is

Then take the integer variable

And

and assign values

3) If it is

Then take the integer variable

And

and assign values

(3) If it is

Will array

Elements in the same row which are not-1 are respectively marked as

And

and

and respectively assigned to _ζ,1 (x _ζ,1 ,y _ζ,1 ) Abscissa and ordinate of and o _ζ,2 (x _ζ,2 ,y _ζ,2 ) The abscissa and the ordinate of the computer, while calculating the sine value

And

and judging:

(1) If it is

Or

Then take the integer variable

And

and assign values

(2) If it is

Or

Then take the integer variable

And

and assign values

(3) If it is

Or

Then take the integer variable

And

and assign values

(4) If it is

Then take the integer variable

And

and assign values

8. A computer-readable storage medium comprising a computer program for use in conjunction with an electronic device having image processing functionality, the computer program being executable by a processor to perform the decoding method of claim 1.