WO2020125528A1

WO2020125528A1 - Anchor object detection method and apparatus, electronic device, and storage medium

Info

Publication number: WO2020125528A1
Application number: PCT/CN2019/124678
Authority: WO
Inventors: 刁梁; 梁伟彬
Original assignee: 广东美的白色家电技术创新中心有限公司; 美的集团股份有限公司
Priority date: 2018-12-21
Filing date: 2019-12-11
Publication date: 2020-06-25
Also published as: CN111354038B; CN111354038A

Abstract

The embodiments of the present application disclose an anchor object detection method and apparatus, electronic device, and storage medium. The anchor object detection method comprises: obtaining an image containing anchor object imaging; detecting a straight line in said image; on the basis of the features of the detected straight line, determining the degree of similarity between a detected object and the anchor object; on the basis of the degree of similarity, positioning the anchor object in the image.

Description

Anchor detection method and device, electronic equipment and storage medium

Cross-reference of related applications

This application is based on a Chinese patent application with an application number of 201811575222.0 and an application date of December 21, 2018, and claims the priority of the Chinese patent application. The entire content of the Chinese patent application is hereby incorporated by reference.

Technical field

The present application relates to the field of information technology, and in particular to an anchor detection method and device, electronic equipment, and storage medium.

Background technique

With the development of technology, various visual techniques are usually used to detect anchors. In the related art, a camera is used to collect an image containing an anchor, and then processed based on multiple images such as a deep learning model to track the position of the anchor in the image. However, when using existing anchor tracking methods such as deep learning models, when the tracked object is blocked or the color of the surrounding environment is similar, the detection of the anchor is likely to fail.

Summary of the invention

In view of this, the embodiments of the present application desire to provide an anchor detection method and device, an electronic device, and a storage medium.

The technical solution of this application is implemented as follows:

An anchor detection method, including:

Acquire images that include imaging of anchors;

Detect straight lines in the image;

Determine the similarity between the test object and the anchor based on the characteristics of the detected straight line;

Based on the similarity, the position of the anchor in the image is located.

Based on the above solution, the detection of the straight line in the image includes:

Detect a straight line in the image and obtain at least one first straight line;

Classify straight lines according to the equation of the first straight line;

According to the classification result of the first straight line, a second straight line is formed based on the first straight line belonging to the same class.

Based on the above solution, the determining the similarity between the detected object and the anchor based on the characteristics of the detected straight line includes:

Determine the similarity between the test object and the anchor according to the intersection point, angle and/or length of the second straight line of the at least two second straight lines of the test object The intersection point, the included angle of the second straight line and/or the length of the second straight line.

Based on the above solution, the method further includes:

Perpendicularly projecting the first point corresponding to the ith second straight line to the ith second straight line to obtain the first projected coordinates, wherein the first point is the first straight line corresponding to the second straight line The line between the first and last points and the origin and the point with the largest angle between the preset coordinate axes;

Projecting a second point corresponding to the i-th second line onto the i-th second line to obtain second projected coordinates, wherein the second point is the first corresponding to the second line The line between the beginning and end points of the line and the origin and the point with the smallest angle between the preset coordinate axes;

Determining the distance between the intersection point between the ith second straight line and the jth second straight line and the first projected coordinate and the second projected coordinate;

If the distance is less than the first distance threshold, it is determined that the i-th second straight line and the j-th second straight line intersect;

The length of the ith second straight line is determined based on the coordinates of the intersection point, where the coordinates of the intersection point are the coordinates of the intersection point of the ith second straight line and at least two other second straight lines.

Based on the above solution, the method further includes that: the ith second straight line intersects with the jth second straight line, and then the angle between the ith straight line and the jth second straight line is determined.

Based on the above solution, the straight line classification according to the equation of the first straight line includes:

Determine the rectangular coordinate equation of the first straight line according to the coordinates of the first and last points of the first straight line;

Determine the slope and intercept of the first straight line according to the rectangular coordinate equation;

Determine the paradigm distance between the two first straight lines based on the slope and the intercept;

The two first straight lines whose distance of the normal form is smaller than the second distance threshold are classified into one category.

Match the angle between the x-th second line in the second line set of the test object and other second lines with the angle between the y-th line in the template of the anchor and the other line ;

Based on the matching result of the included angle, determining whether the x-th second straight line and the y-th straight line satisfy the included angle matching condition;

Determining whether the x-th second straight line and the y-th straight line satisfy the length matching condition;

The x-th second straight line and the y-th straight line simultaneously satisfy the included angle matching condition and the length matching condition, according to the coordinates of the head and tail points of the x-th second straight line and the y-th line The coordinates of the first and last points of the line, to determine the affine matrix that transforms the x-th second straight line affine to the y-th straight line;

Transforming the second straight line other than the xth second straight line according to the affine matrix to obtain a second set of straight lines after the affine transformation of the detection object;

Determining the similarity between the second set of straight lines and the straight lines in the template;

According to the similarity, it is determined whether the detection object is the anchor.

Based on the above solution, the determining whether the x-th second straight line and the y-th straight line satisfy the length matching condition includes: determining whether the length difference between the x-th second straight line and the y-th straight line Less than the length threshold; the difference between the length of the x-th second straight line and the y-th straight line is less than the length threshold, it is determined that the x-th second straight line and the y-th straight line satisfy the length matching condition. Based on the above solution, the method further includes:

The x-th second straight line of the test object and the plurality of straight lines in the template simultaneously satisfy the included angle matching condition and the length matching condition to generate a combination set, wherein the combination set includes the x-th Multiple combinations of two straight lines and at least two straight lines;

The determining the similarity between the test object and the anchor based on the characteristics of the detected straight line includes:

Each combination in the combination set is traversed, and the similarity between the test substance and the anchor is determined based on each combination.

Based on the above solution, the method further includes:

Determine the included angle range according to the included angle of all straight lines in the template;

If the angle between the second straight line and the other second straight line where the detection object exists is outside the angle range, the angle between the second straight line and the other second straight line in the detection object is located at the angle The second straight line outside the angular range is eliminated from the second straight line set.

Based on the above solution, the method further includes:

Acquiring the image and obtaining an image containing candidate contour regions imaged by the anchor;

Selecting the contour area with the largest area from the candidate contour areas as the target contour area;

The detecting the straight line in the image includes:

A straight line located in the target contour area is detected.

An anchor detection device, including:

An image module, configured to obtain an image containing an anchor imaging;

A straight line module, configured to detect a straight line in the image;

A similarity module configured to determine the similarity between the detected object and the anchor based on the characteristics of the detected straight line;

The positioning module is configured to position the anchor in the image based on the similarity.

An electronic device includes:

Memory

A processor, connected to the memory, is used to implement the anchor detection method provided by any of the foregoing technical solutions by executing computer-executable instructions located on the memory.

A computer storage medium that stores computer-executable instructions; after the computer-executable instructions are executed, implement the anchor detection method provided by any of the foregoing technical solutions.

The technical solution provided by the embodiment of the present application determines the similarity between the detected object and the anchor based on the characteristics of the detected straight line through the straight line detection in the image; based on the similarity, it is determined whether the currently detected detected object is the anchor If the detection object is very similar to the anchor, then the area where the current detection object is located is the area where the anchor is located, so that the anchor object in the image can be located; even if the detection object is partially blocked, However, as long as the similarity with the anchor is high enough, it will be deemed that the anchor needs to be detected, thereby achieving successful detection of the anchor, which has the characteristics of high detection efficiency and accurate detection.

BRIEF DESCRIPTION

1 is a schematic flowchart of a first anchor detection method provided by an embodiment of the present application;

2 is a schematic diagram of a system for acquiring images provided by an embodiment of the present application;

3 is a schematic flowchart of a straight line detection provided by an embodiment of the present application;

4 is a schematic flowchart of determining the similarity between a detection object and an anchor based on a straight line feature provided by an embodiment of the present application;

5 is a schematic structural diagram of an anchor detection device according to an embodiment of the present application;

6 is a schematic diagram of a process of creating a template provided by an embodiment of this application;

7 is a schematic flowchart of an anchor detection method provided by an embodiment of the present application;

FIG. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

detailed description

The technical solution of the present application will be further elaborated below in conjunction with the drawings and specific embodiments of the specification.

As shown in FIG. 1, this embodiment provides an anchor detection method, including:

Step S110: Acquire an image including the anchor imaging;

Step S120: Detect the straight line in the image;

Step S130: Determine the similarity between the detected object and the anchor based on the characteristics of the detected straight line;

Step S140: Based on the similarity, locate the position of the anchor in the image.

In this embodiment, the image may be an image collected on the anchor. Thus, the imaging of the anchor must be included in the image, but it may be necessary to locate the area where the anchor is located in the entire image.

In this embodiment, the anchor may be an object with a known shape, and the characteristics of the straight line corresponding to the anchor are known. Therefore, in this embodiment, the straight line is directly detected from the image, and then the feature of the detected straight line is obtained. Based on the feature, the similarity between the detected test object and the anchor is determined. If the similarity is sufficiently high, for example, the similarity If it is greater than the similarity threshold, it is determined that the current detection object is the anchor, and the image area where the current detection object is located is the imaging area of the anchor.

In step S110, it may include: receiving the image from the collection device, or collecting the image by itself.

FIG. 2 shows a schematic diagram of obtaining the image in industry. Figure 2 shows the signal light, camera, and pulley production line; the accessory box can be one of the aforementioned anchors, and the accessory box moves with the operation of the pulley production line. When the accessory box moves below the camera and triggers the photoelectric switch, the camera collects the picture and sends it to the processing unit for recognition. The output shows the recognition result. In this way, the processing unit of the device obtains the image collected by the camera. The image obtained by including the accessory box in the image thus obtained.

In step S120, an image is directly detected from the image instead of generating an image based on the points in the image. There are many ways to detect straight lines in the image, and the following provides several optional ways:

Hough transform straight line detection;

Line segment detection (Line Segment Detector, LSD) line detection.

The realization of the Hough transform straight line detection can be as follows:

For any point A(x0, y0) in a rectangular coordinate system, the straight line passing through point A satisfies Y0=k*X0+b (k is the slope and b is the intercept); then the point A(x0, y0) is crossed in the XY plane ) Can be represented by Y0=k*X0+b, but it cannot be represented if the slope of the straight line perpendicular to the X axis is infinite. Therefore, converting the rectangular coordinate system to the polar coordinate system can solve this special situation.

The line segment detection (Line Segment Detector, LSD) line detection can be as follows: first select the origin O of the polar coordinate system as the focus of the image (w/2, h/2). Establish the Cartesian coordinate system; among them is the image coordinate system. The conversion relationship between the polar coordinate system and the Cartesian coordinate system is. Therefore, when the first and last points of a line segment are known, the corresponding one can be solved. For the calculation of the specific angle, please refer to the polar coordinate representation of straight line detection.

There are many specific ways to detect the straight line in the image, but it is not limited to any of the above.

The features of the straight line detected in step S130 include but are not limited to at least one of the following:

The length of a straight line;

Straight angle

Angle between straight lines;

The relative position relationship between the straight lines, for example, the relative position relationship includes: intersection and/or parallel.

The combination of the characteristics of these straight lines directly reflects the characteristics of the detection object, so in step S130, the similarity between the detection object and the anchor can be determined based on the characteristics of the straight line, and then the current detection is determined based on the similarity Whether the object is an anchor.

In short, in this embodiment, straight line detection can be used to detect anchors whose straight line characteristics are known. Compared with the deep learning model, it has the characteristics of simpler and higher accuracy and less resource overhead. At the same time, when the contour to be detected is obscured by an unknown object, both the traditional algorithm and the deep learning algorithm will fail, and this method is still applicable.

In some embodiments, as shown in FIG. 3, the step S110 may include:

Step S111: Detect the straight line in the image and obtain at least one first straight line;

Step S112: classify straight lines according to the equation of the first straight line;

Step S113: According to the classification result of the first straight line, form a second straight line based on the first straight line belonging to the same class.

Due to the illumination of the image or the error of the straight line detection, sometimes a straight line is detected as multiple straight lines, so in this embodiment, the straight line directly detected from the image is called the first straight line, here The first straight line may be a straight line with a length instead of a straight line with a wireless length that is mathematically defined. The first straight line has an end point, or such an end point is also called an end point. After detecting these first straight lines, the equation of the first straight line can be obtained. This equation can be called a straight line equation. The straight line equation reflects the characteristics of the straight line, for example, the slope and intercept of the straight line.

In this embodiment, straight line classification is performed based on the equation of the first straight line. For example, two straight lines that are sufficiently similar are considered to be classified into the same type, and then a second straight line is obtained based on the first straight line of the same type.

In some embodiments, the step S112 may include at least one of the following:

Compare the slope and intercept of different first straight lines, and classify multiple first straight lines with the same slope and intercept as one category;

Based on the slopes and intercepts of different first straight lines, the paradigm distance between any two first straight lines is calculated. If the paradigm distance is less than the second distance threshold, the two first straight lines can be classified into one category. Specifically, the step S112 may include: determining the rectangular coordinate equation of the first straight line according to the coordinates of the head and tail points of the first straight line; and determining the slope of the first straight line according to the rectangular coordinate equation And intercept; based on the slope and the intercept, determine the paradigm distance between the two first straight lines; the paradigm distance is less than the second distance threshold of the two first straight lines are classified as one type. In this embodiment, the first and last points may be two endpoints of a straight line.

The normal distance is called Manhattan distance, which can be determined by the following formula:

Take the coordinates (X1, Y1) and coordinates (X2, Y2) as an example:

Find the first absolute value of the difference between X1 and X2,

Find the second absolute value of the difference between Y1 and Y2,

The sum of the first absolute value and the second absolute value is calculated to obtain the normal distance of the coordinates (X1, Y1) and the coordinates (X2, Y2).

In short, there are many ways to classify straight lines based on the equations, and are not limited to any of the above.

In step S113, a second straight line connecting the first straight lines of the same type is generated based on the first straight lines classified into the same type. There are various ways to implement step S113, and in this embodiment, two optional ways are provided:

Based on the coordinates of the head and tail points of each first straight line of the same type, soft fitting of the second straight line is performed. In this way, even if there is a certain deviation in the first straight line, the fitting can be successfully and accurately performed. State the second straight line;

Based on the coordinates of the head and tail points of each first straight line of the same type, a hard fitting of the second straight line is performed.

When the soft fitting of the second straight line is obtained, after the coordinates of the head and tail points of multiple second straight lines of the same type are obtained, the head and tail points of all the first straight lines of the same type are located on the second straight line through soft fitting Up or around the second straight line.

When performing the hard fitting of the second straight line, the coordinates of the head and tail points of multiple first straight lines can be directly found, and then the first point and the second point can be obtained. For example, the predetermined coordinate axis may be an x axis or a y axis, the x axis is perpendicular to the y axis, and the x axis and the y axis constitute a two-dimensional coordinate system. Here, the first point is the first and last points of the first straight line connected to the origin, and then looking at the angle between the line and the preset coordinate axis, where the first point is the point corresponding to the maximum included angle, The second point is the point corresponding to the smallest included angle. The first and last points can also be called end points, including: two end points at both ends of the first straight line.

By forming the second straight line, the phenomenon of a large number of repeated straight lines caused by directly detecting the straight line from the image can be reduced, and it is convenient to accurately locate the position of the anchor in the image later.

In some embodiments, the step S130 may include:

The similarity between the test object and the anchor is determined according to the intersection point, the included angle, and/or the length of the second straight line of at least two second straight lines of the test object.

In this embodiment, a picture to be detected will detect multiple contours to be detected, and each contour to be detected may be fitted with multiple second straight lines. The characteristics of the detected straight line may include: the length of the second straight line, the second Angles and intersections between straight lines.

In this embodiment, one or more of the multiple features of the second straight line may be matched with the features of the known straight line of the anchor to determine the similarity between the test object and the test object.

In some embodiments, the method further includes detecting features of the second straight line.

The feature of detecting the second straight line may include: included angle and length. The method for obtaining the angle of the second straight line can be as follows:

Perpendicularly projecting the first point corresponding to the ith second straight line to the ith second straight line to obtain the first projected coordinates, wherein the first point is the first straight line corresponding to the second straight line The line between the first and last points and the origin and the point with the largest angle between the preset coordinate axes

The length of the i-th straight line is determined based on the intersection point coordinates of the i-th second straight line and two other second straight lines that intersect the i-th second straight line.

For example, the ith second straight line and the second straight line A and the second straight line B respectively intersect, and the intersection point of the ith second straight line and the second straight line A is close to the first projected coordinate, and the ith second straight line and the second straight line If the intersection point of the two straight lines B is close to the second projected coordinates, the length of the second straight line may be calculated based on the intersection points between the i-th second straight line and the second straight line A and the second straight line B, respectively.

In some embodiments, the ith second straight line may intersect with more than two second straight lines, and the following method may be used when calculating the length of the ith second straight line:

Select the m1 second straight line with the smallest distance from the first projected coordinate on the ith second straight line;

Select the m2 second straight line with the smallest distance from the first projected coordinate on the ith second straight line;

Calculate the length of the ith second straight line based on the first intersection of the ith straight line and the m1 second straight line, and the second intersection of the ith second straight line and the m2 second straight line; to improve the accuracy of the length .

In some embodiments, the method includes: the ith second straight line intersects with the jth second straight line, and then determining the angle between the ith straight line and the jth second straight line.

The angle here can be directly solved by the straight line equation of the two second straight lines, or can be solved by other methods. For example, and find the projection point of the first point Pfi = (x _fi , y _fi ) on the line i Ptfi = (xt _fi , yt _fi ), and the second point Pni = (x _ni , y _ni ) on the line i The projection point Ptni on the line = (xt _ni , yt _ni ), calculate the point Pfi (or Pni) j≠i which is the closest distance between the Pfi on the straight line i and the start and end points of other straight lines, and find the intersection coordinate of the straight lines i and j PRfi = (xr _fi , Yr _fi ), the angle between i and j is:

The preset coordinate axis may be an x axis or a y axis. If the coordinate distance between the end points of the two second straight lines in the anchor profile is less than the set threshold, the two second straight lines are considered to intersect, and the angle between the straight lines can be obtained by the straight line equation.

By repeating the above steps, the angle between any two second straight lines of the outline and the length of the second straight line can be anchored.

A plurality of second straight lines can be detected in the anchoring profile, each second straight line has its own head and tail points, and then find other second straight lines that are closest to and farthest from the head and tail points of each second straight line. It is further determined whether two second straight lines intersect. In some embodiments, each second straight line will find two second intersecting straight lines, and the length of the second straight line can be confirmed by cutting the line.

In some embodiments, as shown in FIG. 4, the step S130 may include:

Step S131: the angle between the x-th second straight line and the other second straight lines in the second set of straight lines of the detection object and the y-th straight line and other straight lines in the template of the anchor Angle matching;

Step S132: Based on the matching result of the included angle, determine whether the x-th second straight line and the y-th straight line in the template satisfy the included angle matching condition;

Step S133: Determine whether the x-th second straight line and the y-th straight line satisfy the length matching condition;

Step S134: the x-th second straight line and the y-th straight line satisfy the angle matching condition and the length matching condition at the same time, according to the coordinates of the head and tail points of the x-th second straight line and the The coordinates of the first and last points of the yth line determine the affine matrix that transforms the second affine line of the xth line to the yth line;

Step S135: Transform the second straight line other than the x-th second straight line according to the affine matrix to obtain a second set of straight lines after the affine transformation of the test object;

Step S136: Determine the similarity between the second set of straight lines and the straight lines in the template;

Step S137: Determine whether the detection object is the anchor according to the similarity.

Since the angle of the anchor relative to the camera will change to some extent, the imaging of the anchor in the image may change in translation and the like. In this embodiment, for accurate matching, the angle and the length between the straight lines are matched to determine the similarity between the detected straight line and the straight line in the anchor template.

Both x and y are positive integers, where x is less than or equal to the total number of second straight lines, and y is less than or equal to the total number of straight lines in the template.

In some embodiments, the template may be pre-generated by the electronic device and stored in the electronic device. In some embodiments, the template may also be pre-received by the electronic device from other devices.

The angle between a second straight line and any straight line in the template satisfies the angle similarity condition and the length similarity condition, which means that the detected second line is similar to the corresponding line in the template. Displacement or the like causes the coordinates of two similar or matching straight lines to be inconsistent. Thus, by changing the coordinates of the head and tail points of two similar or matching straight lines, the affine matrix can be solved. Using these affine matrices for the detected affine changes of other second straight lines, you can quickly get the affine changed second straight lines, and then match the straight lines in the template, you can get the detected straight lines and the template The first ratio between the number of matching lines and the total number of lines, or the second ratio between the number of matching lines and the number of unmatching lines. The first ratio and/or the second ratio here can be directly used as a parameter for evaluating the degree of similarity between the test substance and the anchor substance. If the first ratio and/or the second ratio is greater than the ratio threshold, it may be considered that the detection object and the anchor are the same, and the detection object is the imaging region of the anchor in the image area.

The determining whether the x-th second straight line and the y-th straight line satisfy the length matching condition includes: determining whether a length difference between the x-th second straight line and the y-th straight line is less than a length threshold;

If the length difference between the xth second straight line and the yth straight line is less than the length threshold, it is determined that the xth second straight line and the yth straight line satisfy the length matching condition.

In some embodiments, the method further includes: the picture to be detected can detect multiple contours, and the contours need to be traversed; thereby, an image with an anchor imaging is found from the multiple images, and the imaging of the anchor is obtained The position in the image.

In some embodiments, the method may further include: the x-th second straight line of the test object and the multiple straight lines in the template satisfy the angle matching condition and the length matching condition at the same time, generating a combination Set, wherein the combination set includes a plurality of combinations of the x-th second straight line and at least two straight lines; the step S130 may include: traversing each combination in the combination set and determining the based on each combination The similarity between the test object and the anchor.

For example, in the step S130, it may include: traversing the combination set to determine the affine matrix corresponding to each combination; based on the second straight line detected by the transformation of each affine matrix solved, the transformed first The second straight line is matched with the straight line in the template, and according to the degree of similarity or matching degree between the transformed second straight line and the straight line, it is determined whether the currently detected test object is the anchor.

If the anchor is rectangular, trapezoid, etc., there may be multiple parallel sides, then at this time, an x-th second straight line may be similar to or match with multiple straight lines in the template. In this embodiment, Consider a set of matching two straight lines as a combination to form a combination set. Then traverse each combination in the combination set, solve the affine matrix of each combination, and determine the similarity between the test object and the anchor based on each combination, thereby determining whether the test object and the anchor are similar. In some embodiments, the method further includes:

Once the angle between each straight line in the template and other straight lines is determined, the angle range between multiple straight lines in the template is determined. In this embodiment, in order to reduce the amount of subsequent calculations, the angle range may be determined based on the angle between any two straight lines in the template in advance. Then, using the included angle range, an abnormal second straight line is proposed from the second straight line set.

In this embodiment, the method further includes:

Obtain the image;

The obtained image contains candidate contour regions imaged by the anchor;

The step S120 may include detecting a straight line located in the target contour area.

In this embodiment, pixel-level segmentation can be used to segment the area where the object is imaged from the background area to obtain one or more candidate contour areas.

Currently, only one candidate contour area is obtained, and it is determined that the candidate contour area is the target contour area.

Currently, if multiple candidate contour regions are obtained, one or more target contour regions can be selected from the candidate contour regions. For example, a candidate contour region with the largest current area is selected as the target contour region, and straight line detection is performed in step S110.

If a detection object similar to or matching the anchor is detected in the target contour area, it can be considered that the anchor is successfully positioned, and the detection can be stopped.

If no anchor is detected in the current target contour area, the candidate contour area with the next largest area is used as the target contour area to continue the detection, so until the anchor is detected, or until all candidate contour areas are regarded as the target contour The area has been tested.

In some embodiments, the imaging size of the anchor may also be determined directly according to the acquisition parameters of the camera and the size of the anchor, and one or more of the candidate contour regions may be selected as the target contour based on the imaging size Area, parallel detection of each target contour area to speed up the positioning rate of the anchor. The acquisition parameters may include but are not limited to parameters related to image acquisition such as focal length and/or depth of field.

As shown in FIG. 5, this embodiment also provides an anchor detection device, including:

The image module 110 is configured to obtain an image including the imaging of the anchor;

The straight line module 120 is configured to detect straight lines in the image;

The similarity module 130 is configured to determine the similarity between the detected object and the anchor based on the characteristics of the detected straight line;

The positioning module 140 is configured to position the anchor in the image based on the similarity.

In some embodiments, the image module 110, the straight line module 120, the similarity module 130, and the positioning module 140 may all be program modules. After the program modules are executed by the processor, image acquisition, straight line detection, and similarity The determination of the degree and the positioning of the anchor in the image.

In some embodiments, the straight line module 120 is configured to detect a straight line in the image and obtain at least one first straight line; classify straight lines according to the equation of the first straight line; according to the first straight line The classification result of the line forms a second straight line based on the first straight line belonging to the same class.

In some embodiments, the similarity-based module 130 is configured to determine the detection based on the intersection point, the included angle, and/or the length of the second straight line of at least two second straight lines of the test object The similarity between the object and the anchor.

In some embodiments, the device further includes:

A first projection module configured to vertically project the first point corresponding to the ith second straight line onto the ith second straight line to obtain a first projection coordinate, wherein the first point is the second straight line The line between the beginning and end points of the first straight line corresponding to the straight line and the origin and the point with the largest angle between the preset coordinate axes;

A second projection module configured to project a second point corresponding to the ith second straight line to the ith second straight line to obtain second projected coordinates, wherein the second point is the first The line between the beginning and end points of the first straight line corresponding to the two straight lines and the origin and the point with the smallest angle between the preset coordinate axes;

A length module configured to determine the distance between the intersection point between the ith second straight line and the jth second straight line and the first projected coordinate and the second projected coordinate; the distance is less than the first distance threshold , It is determined that the i-th second straight line intersects with the j-th second straight line; based on the coordinates of the intersection point, the length of the i-th second straight line is determined, wherein the coordinate of the intersection point is the i-th second straight line The coordinates of the intersection point of the second straight line and at least two other second straight lines.

In some embodiments, the device further includes:

The first angle module is configured such that the i-th second straight line intersects with the j-th second straight line, and then determines the angle between the i-th straight line and the j-th second straight line.

In some embodiments, the straight line module 120 is configured to determine the rectangular coordinate equation of the first straight line according to the coordinates of the first and last points of the first straight line; determine the first coordinate according to the rectangular coordinate equation The slope and the intercept of the straight line; based on the slope and the intercept, determine the paradigm distance between the two first straight lines; the two firsts whose distance of the paradigm is less than the second distance threshold Straight lines are grouped together.

In some embodiments, the similarity module 130 is configured to determine the angle between the x-th second straight line in the second straight line set of the detected object and other second straight lines, and the template of the anchor The angle match between the y-th straight line and other straight lines in the inside; based on the matching result of the angle, determine whether the x-th second straight line and the y-th straight line in the template satisfy the angle matching Conditions; determine whether the x-th second straight line and the y-th straight line satisfy the length matching condition; if the x-th second straight line and the y-th straight line simultaneously satisfy the included angle matching condition and all The length matching condition, based on the coordinates of the head and tail points of the xth second line and the coordinates of the head and tail points of the yth line, determining the affine transformation of the xth second line to the y An affine matrix of straight lines; according to the affine matrix, transform the second straight line other than the x-th second straight line to obtain a set of second straight lines after the affine transformation of the test object; determine the The similarity between the second set of straight lines and the straight lines in the template; according to the similarity, it is determined whether the detection object is the anchor.

In some embodiments, the device further includes:

A combination module, configured such that the x-th second straight line of the detection object and the multiple straight lines in the template simultaneously satisfy the included angle matching condition and the length matching condition to generate a combined set, wherein the combined set Including multiple combinations of the xth second straight line and at least two straight lines;

The similarity module 130 is configured to traverse each combination in the combination set and determine the similarity between the detection object and the anchor based on each combination.

In some embodiments, the device further includes:

The included angle range module is configured to determine the included angle range according to the included angle of all straight lines in the template;

The rejection module is configured such that the angle between the second straight line and the other second straight line where the detected object exists is outside the included angle range, and then the angle between the detected object and the other second straight line is included The second straight line whose angle is outside the included angle range is removed from the second straight line set.

In some embodiments, the device further includes:

A candidate contour region module configured to acquire the image and obtain a candidate contour region imaged by the anchor including the anchor;

A target contour area module configured to select a contour area with the largest area from the candidate contour area as the target contour area;

The straight line module 120 is configured to detect a straight line located in the target contour area.

The following provides a specific example in combination with any of the above embodiments:

Example 1:

Hardware part and system algorithm part.

As shown in Figure 2, the hardware part includes: a belt production line, a photoelectric sensor, and a camera directly above the belt production line. When the accessory box moves below the camera and triggers the photoelectric switch, the camera collects the picture and sends it to the processing unit Recognize, the most output shows the recognition results.

The first part, the template is established, the picture as a template needs to have the following standards: (1) the background of the picture is pure (2) the contour of the anchor to be detected is clear and unobstructed (2) the outer contour of the anchor to be detected is the largest and contains A template anchor.

The specific template creation method can refer to Figure 6, including the following steps:

Select the contour extraction template picture, such as: pixel segmentation, set pixel threshold, extract candidate contour regions in the image by pixel segmentation; specifically, compare the area of each contour candidate region, and retain the largest area to filter out all others area;

Binary image;

Extract the outline from the binarized picture;

Further extract the outline with the largest area;

The straight line detection algorithm is used for straight line detection, for example, the straight line detection algorithm is used to extract straight lines for the area contour. The algorithm used in the specific embodiment is the Hough straight line detection or LSD straight line detection algorithm, which detects all straight lines in the contour;

Calculate the rectangular coordinate equations of all straight lines. The straight line detection can get the starting and ending coordinates of each line segment. Let the coordinate of the first point of the i-th line be Psi=(x _si , y _si ) and the coordinate of the tail point Pei=( x _ei , y _ei ), then the equation of a straight line i is:

At this time, the slope a _{i of the} straight line i and the intercept b _i are obtained. After all the slopes and intercepts are obtained, the slopes and intercepts are clustered, and the number of cluster centers is the number of straight lines of the anchor contour to be inspected .

After the clustering is completed, straight line fitting is performed on all the head and tail points of the line i belonging to the same class, the angle between the original detection line from all points to the image origin and the x axis is calculated, and the first point of the maximum angle is obtained Pfi = (x _fi , y _fi ) projection point on line i Ptfi = (xt _fi , yt _fi ), and the second point corresponding to the smallest angle Pni = (x _ni , y _ni ) on line i Projection point Ptni = (xt _ni , yt _ni ), the projection calculation formula takes the farthest as an example, if the straight line i passes through the origin, then Pfi is the farthest point from the origin, and Pni is the closest point to the origin:

Calculate the point Pfi (or Pni) j≠i which is the closest distance between the Pfi on the straight line i and the start and end points of other straight lines, and find the coordinate of the intersection point of the straight lines i and j PRfi=(xr _fi , yr _fi ). for

The method of obtaining the same and other linear Pni nearest point Pni (or Pfi) Pfi straight line i and j is calculated intersection coordinates _{_{PRni = (xr ni, yr ni}} ), i having a length at this time is calculated as:

Record the length L _{i of} each line, the two included angles θ _fi and θ _ni , and finally detect that the anchor in the template contains C _line lines.

Record the maximum value of all included angles as θ _min and the minimum value as θ _max .

After the template is established, the test can be carried out. When the accessory box blocks the photoelectric switch and the camera collects the picture to be inspected, referring to FIG. 7, the inspection may include the following operations:

Image binarization, extraction of contours, and selection of contours, which can include: pixel segmentation, and extraction of candidate contours through pixel thresholds; preliminary filtering of contours by area thresholds, filtering out areas smaller than anchors, and obtaining N _area candidate contours; use the straight line detection algorithm for the first contour (N _area =1) to detect all straight lines in the contour.

Use straight line detection algorithm, for example, next calculate the rectangular coordinate equations of all straight lines, through straight line detection can get the starting and ending coordinates of each line segment, set the first coordinate of the i-th straight line as Psi = (x _si , y _si ), tail The coordinates are Pei = (x _ei , y _ei ), using formula (1) to obtain the slope a _{i of the} straight line _i , and the intercept is b _i , after obtaining all the slopes and intercepts, calculate the paradigm distance of each two sets of straight lines, The formula is as follows:

L _d = a _i -a _j +b _i -b _j (j≠i) (5)

At this time, it is necessary to set the second distance threshold Td for clustering. When L _d ≤ Td, the anchor line is classified into one class, and finally to the N _line class line. Due to the use of the point slope equation, the slope of the line tends to be infinite. (i.e. a straight line perpendicular to the x-axis), then set the slope threshold T _a, the straight line i, or if _a _i ≥T a straight line perpendicular to the x-axis is set when _a _i = T a.

The projection coordinates Ptfi = (xt _fi , yt _fi ) and Ptni = (xt _ni , yt _ni ) are calculated using the same method as in template creation (5).

I is calculated straight line with all the other angle, to obtain an angle of N _D, the angle of the primaries, retain all _{_{_{θ min ≤θ d ≤θ max (d∈}}} [1, N D]), to give N ' _D angles. Determine the second straight line of the M1th line so that the distance between a certain projection point of the second straight line of the M1th line and a certain projection coordinate of the second straight line of the ith line is the smallest; The second intersection point coordinate of the intersection point of the second straight line and the second straight line of the M1th line, and determine the angle between the two; determine the second straight line of the M2th line, such that a certain projection of the second straight line of the M1th line The distance between the point and a certain projection coordinate of the second straight line of the i-th line is the smallest; the coordinate of the third intersection point of the intersection point of the second straight line of the i-th line and the second straight line of the M2th line is determined, and the second Angle; determine the length of the ith second straight line on the anchor based on the coordinates of the second intersection point and the coordinates of the third intersection point.

Calculate the similarity between the straight line i and the template straight line using the selected angle, and the formula for calculating the similarity between the straight line i and j is as follows:

Obtain at this time: β _i =min{β _ij }(j∈[1,C _line ]) (7)

When j = v (v ∈ [1, C _line ]), and β _i takes the minimum value, it means that the straight line i matches v in the template the most.

Set the similarity threshold T _β , when β _i ≤ T _β , it is judged that the straight line i in the test sample may be similar to the template straight line v, and all straight lines can be preliminarily screened and classified by this method. The angle between i and the two straight lines matched by the above template is j _n and j _f . When the stepped polygon of parallel lines is selected as the anchor, several straight lines that match the matching will be obtained by angle filtering. At this time, v is not unique It is _assumed that there are N _linei 'straight lines matching the i straight line in the template. At this time, j _n and j _{f are also} not unique, there are multiple sets of combinations, and there are a total of N _ci combinations.

Composition is selected from _{_{n ci = 1 (n ci ∈}} [1, N ci]), this time with a straight line intersecting i and j _n ¹ ¹ j _f, i and j _n is calculated as ^an intersection PTRni = (xtr _{_ni,} ytr _ni ), and the intersection of i and j _f ¹ is PTRfi = (xtr _fi , ytr _fi ), calculate the length LP _t , set the threshold Tl, and when |LP _i -L _v |≤Tl, perform subsequent matching.

Assuming that n _ci =1 (n _ci ∈[1,N _ci ]) meets the length screening, and assuming that the straight line corresponds to the straight line v _m ( _m∈ [1,N _linei ']) in the template, let m=1, at this time v _m has PRfv = (xr _fv , yr _fv ), PRnv = (xr _nv , yr _nv ), at this time, it is assumed that PRni corresponds to PRnv, and in actual industrial field applications, the camera height will be fixed, so the anchor in the test picture The anchor and anchor in the picture of the fixed object and the template only undergo translation and rotation transformation.

First translate the line i so that the PRni and PRnv pairs coincide, the translation affine matrix can be:

among them

Let PTRni' after affine transformation PTRni' = PTRnv = (xtr _ni ', ytr _ni '), PTRfi after affine transformation PTRfi' = (xtr _fi ', ytr _fi '), then rotate θ around PTRni The shooting matrix is

among them

Finally, the global affine transformation matrix is

After the affine matrix is calculated, the remaining N _area -1 straight lines obtained in step (4) are affine transformed by the affine matrix. Affine transform the first and last coordinates of all straight lines to obtain

Find the point-slope straight line equations of Ps” and Pf” to obtain the corresponding straight line slope a′ and intercept b′. At this time, the angle matching result of the remaining N _area -1 straight lines is obtained through step (5). Match the straight line j in the template, then calculate the straight line similarity

L _d = a′-a _j +b′-b _j (14)

When L _d ≤ Td, it is judged that the straight line is consistent, the same method is used to judge the similarity of all straight lines, and finally the C′ _line is matched with the matched straight line. At this time, there is an anchor matching similarity:

Set the matching threshold Tnl, when

, It is determined that the detection is successful, and M _{p→r is} used as the affine matrix of the entire picture. At this time, all points on the test image can be matched on the template image through the affine matrix.

If all combinations have been traversed but still cannot be matched, let m=m+1, with v _m ( _m∈ [1,N _linei ']), traverse all N _linei ', and repeat steps (7) to (9).

If the match fails, let n _ci =n _ci +1(n _ci ∈[1,N _ci ]) traverse the other combinations in n _ci , and repeat steps (6) to (9).

If the matching of line i fails, traverse the remaining N _line straight lines and repeat steps (4) to (10)

If the contour N _area = 1 matching fails, let N _area = N _area +1 and repeat (3) to (10) until a matching contour is found.

In short, by finding all the second straight lines on the contour to be detected, the angle between the second straight line on the contour to be detected and all other straight lines and the length of the cut line, and generating a list of angles and a list of the length of the cut line, and The angle and length of the second straight line generated by the anchor contour are matched, and the matching situation of the i-th second straight line and the second line of the anchor contour is preliminarily predicted; it can quickly determine whether each current candidate contour area is an anchor Imaging area. Generally, the candidate contours are sorted, and the contour with the largest area is selected first, thereby saving detection time.

The detection speed is fast, and the accuracy is much higher than the pure deep learning algorithm, which can achieve pixel-level point positioning.

Anti-occlusion, even if the anchor is blocked by objects of similar colors, it is still extremely robust.

Strong versatility, suitable for all polygonal anchors; and extremely fast, for non-stepped symmetric anchors, the affine matrix can be positioned efficiently at one time.

As shown in FIG. 8, this embodiment provides an electronic device, including:

Memory

A processor, connected to the memory, is configured to implement a method for detecting a method anchor provided by any of the foregoing technical solutions by executing computer-executable instructions located on the memory, for example, FIG. 1, FIG. 3, and FIG. 4, One or more of the method anchor detection methods shown in FIGS. 6 and 7.

The memory may be various types of memory, such as random access memory, read-only memory, flash memory, etc. The memory can be used for information storage, for example, storing computer-executable instructions. The computer executable instructions may be various program instructions, for example, target program instructions and/or source program instructions.

The processor may be various types of processors, for example, a central processor, a microprocessor, a digital signal processor, a programmable array, a digital signal processor, an application specific integrated circuit, or an image processor.

The processor may be connected to the memory through a bus. The bus may be an integrated circuit bus or the like.

In some embodiments, the electronic device may further include: a communication interface, and the communication interface may include: a network interface, for example, a local area network interface, a transceiver antenna, and the like. The communication interface is also connected to the processor and can be used for information transmission and reception.

In some embodiments, the electronic device further includes a human-machine interaction interface. For example, the human-machine interaction interface may include various input and output devices, such as a keyboard, a touch screen, and the like.

This embodiment provides a computer storage medium that stores computer-executable instructions; after the computer-executable instructions are executed, the method for detecting a method anchor provided by any of the foregoing technical solutions is implemented, for example, FIG. 1. One or more of the method anchor detection methods shown in FIGS. 3 and 4, 6 and 7.

The computer storage medium may include various recording media having a recording function, for example, various storage media such as a CD, a floppy disk, a hard disk, a magnetic tape, an optical disk, a U disk, or a mobile hard disk. The optional computer storage medium may be a non-transitory storage medium, and the computer storage medium may be read by a processor, so that computer-executable instructions stored on a computer storage mechanism are acquired and executed by the processor to implement any of the foregoing An anchor detection method provided by a technical solution, for example, executes an anchor detection method applied in a terminal device or an anchor detection method in an application server.

In the several embodiments provided in this application, it should be understood that the disclosed device and method may be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a division of logical functions. In actual implementation, there may be other division methods, such as: multiple units or components may be combined, or Can be integrated into another system, or some features can be ignored, or not implemented. In addition, the displayed or discussed components are coupled to each other, or directly coupled, or the communication connection may be through some interfaces, and the indirect coupling or communication connection of the device or unit may be electrical, mechanical, or other forms of.

The above-mentioned units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units; Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, the functional units in the embodiments of the present application may all be integrated into one processing module, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above integration The unit can be implemented in the form of hardware, or in the form of hardware plus software functional units.

Persons of ordinary skill in the art may understand that all or part of the steps to implement the above method embodiments may be completed by program instructions related hardware. The foregoing program may be stored in a computer-readable storage medium, and when the program is executed, Including the steps of the above method embodiments; and the foregoing storage media include: mobile storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks, etc. A medium that can store program codes.

The above is only the specific implementation of this application, but the scope of protection of this application is not limited to this, any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in this application. It should be covered by the scope of protection of this application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

An anchor detection method, including:

Acquire images that include imaging of anchors;

Detect straight lines in the image;

Determine the similarity between the detected object and the anchor based on the characteristics of the detected straight line;

Based on the similarity, the position of the anchor in the image is located.
The method according to claim 1, wherein

The detecting the straight line in the image includes:

Detect a straight line in the image and obtain at least one first straight line;

Classify straight lines according to the equation of the first straight line;

According to the classification result of the first straight line, a second straight line is formed based on the first straight line belonging to the same class.
The method according to claim 2, wherein the determining the similarity between the test object and the anchor object based on the characteristics of the detected straight line includes:

The similarity between the test object and the anchor is determined according to the intersection point, the included angle, and/or the length of the second straight line of the at least two second straight lines of the test object.
The method of claim 3, wherein the method further comprises:

Perpendicularly projecting the first point corresponding to the ith second straight line to the ith second straight line to obtain the first projected coordinates, wherein the first point is the first straight line corresponding to the second straight line The line between the first and last points and the origin and the point with the largest angle between the preset coordinate axes;

Projecting a second point corresponding to the i-th second line onto the i-th second line to obtain second projected coordinates, wherein the second point is the first corresponding to the second line The line between the beginning and end points of the line and the origin and the point with the smallest angle between the preset coordinate axes;

Determining the distance between the intersection point between the ith second straight line and the jth second straight line and the first projected coordinate and the second projected coordinate;

If the distance is less than the first distance threshold, it is determined that the i-th second straight line and the j-th second straight line intersect;

The length of the ith second straight line is determined based on the coordinates of the intersection point, where the coordinates of the intersection point are the coordinates of the intersection point of the ith second straight line and at least two other second straight lines.
The method of claim 3, wherein the method further comprises:

The intersection of the i-th second straight line and the j-th second straight line determines the angle between the i-th second straight line and the j-th second straight line.
The method according to claim 2, wherein

The straight line classification according to the equation of the first straight line includes:

Determine the rectangular coordinate equation of the first straight line according to the coordinates of the first and last points of the first straight line;

Determine the slope and intercept of the first straight line according to the rectangular coordinate equation;

Determine the paradigm distance between the two first straight lines based on the slope and the intercept;

The two first straight lines whose distance of the normal form is smaller than the second distance threshold are classified into one category.
The method according to claim 2, wherein the determining the similarity between the test object and the anchor object based on the characteristics of the detected straight line includes:

Match the angle between the x-th second line in the second line set of the test object and other second lines with the angle between the y-th line in the template of the anchor and the other line ;

Based on the matching result of the included angle, determining whether the x-th second straight line and the y-th straight line satisfy the included angle matching condition;

Determining whether the x-th second straight line and the y-th straight line satisfy the length matching condition;

The x-th second straight line and the y-th straight line simultaneously satisfy the included angle matching condition and the length matching condition, according to the coordinates of the head and tail points of the x-th second straight line and the y-th line The coordinates of the first and last points of the line, determine the affine matrix that transforms the x-th second straight line affine to the y-th straight line;

Transforming the second straight line other than the xth second straight line according to the affine matrix to obtain a second set of straight lines after the affine transformation of the detection object;

Determining the similarity between the second set of straight lines and the straight lines in the template;

According to the similarity, it is determined whether the detection object is the anchor.
The method according to claim 7, wherein the method further comprises:

The x-th second straight line of the test object and the plurality of straight lines in the template simultaneously satisfy the included angle matching condition and the length matching condition to generate a combination set, wherein the combination set includes the x-th Multiple combinations of two straight lines and at least two straight lines;

The determining the similarity between the test object and the anchor based on the characteristics of the detected straight line includes:

Each combination in the combination set is traversed, and the similarity between the test object and the anchor is determined based on each combination.
The method of claim 7, wherein the method further comprises:

Determine the included angle range according to the included angle of all straight lines in the template;

If the angle between the second straight line and the other second straight line where the detection object exists is outside the angle range, the angle between the second straight line and the other second straight line in the detection object is located at the angle The second straight line outside the angular range is eliminated from the second straight line set.
The method according to claim 7, wherein

The determining whether the x-th second straight line and the y-th straight line satisfy the length matching condition includes:

Determining whether the length difference between the x-th second straight line and the y-th straight line is less than a length threshold;

If the length difference between the xth second straight line and the yth straight line is less than the length threshold, it is determined that the xth second straight line and the yth straight line satisfy the length matching condition.
The method according to any one of claims 1 to 10, wherein the method further comprises:

Acquiring the image and obtaining an image containing candidate contour regions imaged by the anchor;

Selecting the contour area with the largest area from the candidate contour areas as the target contour area;

The detecting the straight line in the image includes:

A straight line located in the target contour area is detected.
An anchor detection device, including:

An image module, configured to obtain an image containing an anchor imaging;

A straight line module, configured to detect a straight line in the image;

A similarity module configured to determine the similarity between the detected object and the anchor based on the characteristics of the detected straight line;

The positioning module is configured to position the anchor in the image based on the similarity.
An electronic device, including:

Memory

A processor, connected to the memory, is configured to implement the method provided in any one of claims 1 to 11 by executing computer-executable instructions located on the memory.
A computer storage medium that stores computer-executable instructions; after the computer-executable instructions are executed, the method provided in any one of claims 1 to 11 is implemented.