Circular texture target localization method based on radial difference guidance

Abstract

The present invention discloses a circular texture target positioning method based on radial difference guidance, comprising: using the circular texture target image obtained by template matching as the initial positioning area and calculating its center position coordinates; dividing the correction area into four quadrants, and using the quadrant with the most gentle grayscale change at the edge of the initial positioning area as the search area; delineating candidate areas within the search area; traversing each candidate point and calculating the radial difference of the data features of the candidate points; determining whether the candidate point with the largest radial difference of the data features meets the positioning termination condition; if so, outputting the corresponding candidate point as the center coordinate of the circular texture target; otherwise, increasing the range of the candidate area. The present invention is applied to the field of target recognition and positioning, using the radial difference of the data features of the candidate pixel point at the center of the circular texture target as the main basis for target positioning, making full use of the extensiveness and complexity of circular texture targets, thereby improving the positioning accuracy of circular texture targets.

Description

Technical Field

[0001] The present invention relates to the technical field of target recognition and positioning image processing, and in particular to a circular texture target positioning method based on radial difference guidance. Background Art

[0002] Template matching target recognition and positioning systems play a core role in circular texture target identification and positioning. They can detect texture targets in images and obtain corresponding feature data such as position, shape, and area. Applications include chip positioning and laser marking. However, in actual processing, template matching methods inevitably produce positioning errors. This is due to the processing environment and camera hardware, resulting in various artifacts such as noise and smudges in the obtained circular texture images. Furthermore, the targets themselves are small, so these artifacts can cause positional deviations during the recognition and detection process. Furthermore, the diverse shapes and sizes of targets require a large amount of image data. In actual production, limited software and hardware platforms for image data processing cannot meet these requirements. Although the resulting errors are small, circular texture targets are often small and therefore require high processing precision. Errors even at the micron level can significantly impact the quality of the finished product. Therefore, positional correction of detected targets is crucial.

[0003] To sum up, after target detection and recognition, studying the correction of the target position is of great significance for improving the accuracy of target detection and recognition and enhancing detection efficiency. It can also further strongly support the industrial development of the circular texture target detection industry. Summary of the Invention

[0004] In response to the above-mentioned deficiencies in the prior art, the present invention provides a circular texture target positioning method based on radial difference guidance, which can effectively improve the positioning accuracy of circular texture targets and improve the accuracy of target detection and recognition tasks in the field of fine machining.

[0005] To achieve the above object, the present invention provides a circular texture target positioning method based on radial difference guidance, comprising the following steps:

[0006] Step 1: Using the circular texture target image obtained by template matching as the initial positioning area, and calculating the center position coordinates of the initial positioning area in the original image;

[0007] Step 2: Delimiting a correction region in the original image based on the center coordinates of the initial positioning region and the diameter of the initial positioning region, wherein the center of the correction region and the center of the initial positioning region both coincide with the origin of the coordinate system O-XY;

[0008] Step 3: Divide the correction area into four quadrants along the X-axis and the Y-axis, and use the quadrant where the grayscale change of the edge of the initial positioning area is the most gentle as the search area;

[0009] Step 4: Delineate a rectangular candidate region in the search area, wherein a corner point of the candidate region coincides with the origin of the coordinate system O-XY, and two adjacent sides of the candidate region coincide with the X axis and the Y axis respectively;

[0010] Step 5: Define the pixel points in the candidate area as candidate points, traverse each candidate point, draw a number of rays with each candidate point as the starting point, and calculate the radial difference of the data feature of each candidate point based on the pixel points covered by the ray in the correction area;

[0011] Step 6: Determine whether the candidate point with the largest radial difference of data features meets the positioning termination condition;

[0012] If so, output the corresponding candidate point as the center coordinate of the circular texture target;

[0013] Otherwise, the range of the candidate region is increased and steps 5 to 6 are performed again.

[0014] In one embodiment, in step 1, the calculation of the center position coordinates of the initial positioning area in the original image is specifically:

[0015] Performing Gaussian filtering on the circular texture target image corresponding to the initial positioning area;

[0016] According to the circular texture target image after Gaussian filtering, the center position coordinate Core(x,y) of the initial positioning area in the original image is calculated as:

[0017]

[0018] Where (x,y) max The maximum x-coordinate value and y-coordinate value in the initial positioning area, (x, y) min The minimum x-coordinate and y-coordinate values ​​in the initial positioning area.

[0019] In one embodiment, in step 2, the correction area is a rectangular area with a width and height of D, wherein D=ω1·d, d is the diameter of the initial positioning area, and ω1 is an area coefficient greater than 1.

[0020] In one embodiment, in step 4, the width and height of the candidate region are:

[0021] Square size =a·x 3 +b·x2 +c·x+d

[0022] Among them, Square size are the width and height of the candidate area, x is the maximum grayscale difference in the search area, a, b, c are the size coefficients obtained by fitting, and d is the basic size value.

[0023] In one embodiment, in step 5, the calculation process of the radial difference of the data feature is:

[0024] Step 5.1: for any candidate point, draw a number of rays from the candidate point, and extend the rays to the end of the correction area, with the angle between each two adjacent rays being θ;

[0025] Step 5.2, calculating the radial difference of the data characteristic of each ray, including:

[0026] dividing the ray into an inner-edge ray portion and an outer-edge ray portion according to the edge of the initial positioning area;

[0027] Calculating the data characteristic values ​​of the pixel points covered by the inner ray portion of the edge and the outer ray portion of the edge respectively;

[0028] Calculate the difference in data characteristic values ​​between the ray portion inside the edge and the ray portion outside the edge, which is the radial difference in data characteristic of the ray;

[0029] Step 5.3: Based on the radial difference of the data characteristic of each ray, the radial difference of the data characteristic of the candidate point can be obtained.

[0030] In one embodiment, in step 5.1, the angle between two adjacent rays is specifically:

[0031] θ=90° / int(A / T num )

[0032] Where A is the area of ​​the candidate region, T num is the number of pixels in the candidate region.

[0033] In one embodiment, in step 5.2, the data characteristic value is entropy distribution, mode, maximum value, median value or position index.

[0034] In one embodiment, in step 5.3, the calculation process of the radial difference of the data characteristics of the candidate points is specifically as follows:

[0035] Adding the radial difference of the data characteristic of each ray corresponding to the candidate point, thereby obtaining the radial difference of the data characteristic of the ray; or

[0036] The sine values ​​of the radial differences of the data characteristics of each ray corresponding to the candidate point are added together to obtain the radial differences of the data characteristics of the ray.

[0037] In one embodiment, in step 6, the positioning termination condition is specifically:

[0038] Taking the corresponding candidate point as the center coordinate of the circular texture target, calculate the grayscale change values ​​V1, V2, V3, and V4 of the circular texture target edge in the four quadrants respectively;

[0039] When there is When , it is determined that the positioning termination condition is not met, otherwise it meets the positioning termination condition, where i = 1 or 2 or 3 or 4, ω2 is a determination coefficient greater than 1.

[0040] In one embodiment, in step 6, the step of increasing the range of the candidate region is specifically:

[0041] Square size-1 =ω3·Square size-0

[0042] Among them, Square size-1 The width and height of the candidate region after the range is enlarged, Square size-0 is the width and height of the candidate region before range enlargement, and ω3 is an enlargement coefficient greater than 1.

[0043] Compared with the prior art, the present invention has the following beneficial technical effects:

[0044] 1. This invention uses the radial difference of the data features of the candidate pixel points at the center of a circular texture target as the primary basis for target positioning. This fully utilizes the universality and complexity of circular texture targets, effectively improving the positioning accuracy of circular texture targets and thus enhancing the accuracy of target detection and recognition tasks in the field of fine machining.

[0045] 2. The present invention determines the search area for target positioning by the smoothness of the grayscale change at the edge of the initial positioning area of ​​the circular texture target, and then searches for the target center by demarcating the candidate area within the search area. At the same time, when the target center search fails, the candidate area is gradually increased in an iterative manner, thereby effectively reducing the computational complexity of the candidate area search and improving the positioning efficiency of the circular texture target. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0047] Figure 1 Flowchart of a circular texture target positioning method based on radial difference guidance in an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a correction area in an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of the division of quadrant areas in an embodiment of the present invention;

[0050] Figure 4 Schematic diagram of drawing out rays in an embodiment of the present invention.

[0051] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0053] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0054] In addition, the terms "first," "second," and so on, used in this disclosure are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referenced. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this disclosure, "plurality" means at least two, such as two or three, unless otherwise specifically defined.

[0055] In the present invention, unless otherwise specified or limited, the terms "connection" and "fixation" should be understood in a broad sense. For example, "fixation" can mean fixed connection, detachable connection, or integration; it can mean mechanical connection, electrical connection, physical connection, or wireless communication connection; it can mean direct connection or indirect connection through an intermediate medium; it can mean internal communication between two elements or interaction between two elements, unless otherwise specified. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0056] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in this field can implement it. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0057] For the detection, identification and positioning of circular texture targets, the commonly used template matching method is affected by factors such as the data set, noise interference and uneven illumination. The final target positioning result often has a certain position deviation and cannot meet more refined work requirements. In addition, due to the differences in the development of various industries, the software and hardware platform resources are limited. Even with a huge data set, the computing efficiency will be reduced due to resource limitations. Therefore, after identification and positioning, the position of the target is corrected, which can make full use of resources and improve the efficiency of the project while improving the accuracy of the project. In view of this, the present embodiment discloses a circular texture target positioning method based on radial difference guidance, which uses the radial difference of the data features of the candidate pixel points at the center of the circular texture target as the main basis for target positioning, making full use of the extensiveness and complexity of the circular texture target, and can effectively improve the positioning accuracy of the circular texture target, thereby improving the accuracy of the target detection and recognition tasks in the field of fine machining.

[0058] The core definition of template matching technology is to perform edge detection or feature point detection in the target image for a known feature pattern, so as to successfully identify and locate the shape target corresponding to the template in the target image. Therefore, in the template matching of a circular texture pattern, the circular shape in the target image can be guaranteed, and when correcting the circular texture target, there is no need to correct its shape. Therefore, during the correction process, there is no need to adjust the shape of the target. The recognition and positioning correction of the target can be mapped to the recognition and positioning of the target center. Figure 1 In this embodiment, the circular texture target positioning method based on radial difference guidance specifically includes the following steps:

[0059] Step 1: Use the circular texture target image obtained by template matching as the initial positioning area, and calculate the center position coordinates of the initial positioning area in the original image;

[0060] Step 2: Delineating a correction region in the original image based on the center coordinates of the initial positioning region and the diameter of the initial positioning region, wherein the center of the correction region and the center of the initial positioning region both coincide with the origin of the coordinate system O-XY;

[0061] Step 3: Divide the correction area into four quadrants along the X-axis and Y-axis, and use the quadrant where the grayscale change of the edge of the initial positioning area is the most gentle as the search area;

[0062] Step 4: Delineate a rectangular candidate region in the search area, wherein a corner point of the candidate region coincides with the origin of the coordinate system O-XY, and two adjacent sides of the candidate region coincide with the X axis and the Y axis respectively;

[0063] Step 5: Define the pixels in the candidate area as candidate points, traverse each candidate point, draw several rays with each candidate point as the starting point, and calculate the radial difference of the data feature of each candidate point based on the pixel points covered by the ray in the correction area;

[0064] Step 6: Determine whether the candidate point with the largest radial difference of data features meets the positioning termination condition;

[0065] If so, output the corresponding candidate point as the center coordinate of the circular texture target;

[0066] Otherwise, increase the range of the candidate region and perform steps 5 to 6 again.

[0067] After obtaining the circular texture target image after template matching, in order to reduce the influence of noise and uneven illumination, it is necessary to perform Gaussian filtering on the obtained target image in the frequency domain to obtain a filtered image. Therefore, the specific implementation method for calculating the center position coordinates of the initial positioning area in the original image in step 1 is:

[0068] First, perform Gaussian filtering on the circular texture target image corresponding to the initial positioning area, which is:

[0069]

[0070] Among them, h(x,y) is the circular texture target image after Gaussian filtering;

[0071] Then, according to the circular texture target image after Gaussian filtering, the center position coordinate Core(x,y) of the initial positioning area in the original image is calculated as:

[0072]

[0073] Where (x,y)max The maximum x-coordinate value and y-coordinate value in the initial positioning area, (x, y) min The minimum x-coordinate and y-coordinate values ​​in the initial positioning area.

[0074] After determining the center position of the initial positioning area, it is unrealistic to perform positioning correction on the circular texture target on the entire original image due to the large amount of data and the involvement of many unnecessary calculation operations. Therefore, in the actual correction process, this embodiment delineates the correction area of ​​the circular texture target in the original image, and ensures that the area of ​​the area is as small as possible while ensuring that the actual position of the circular texture target is included in the correction area. Generally speaking, in template matching, due to the significance of the features and target edges of the circular texture pattern, the difference between the initial positioning area and the actual area obtained will be within a smaller limited range. Therefore, in step 2, a rectangular area with a width and height of D=ω1·d is delineated as the correction area, where d is the diameter of the initial positioning area, and ω1 is a regional coefficient greater than 1. In the specific implementation process, the regional coefficient is preferably ω1=1.5, that is Figure 2 shown.

[0075] In the process of positioning and correcting the center of the circular texture target area, since some images have very high resolution, if the entire image is to be considered for calculation, more rays need to be drawn, resulting in a large amount of data in the area to be calculated. In order to save hardware resources and improve efficiency, this embodiment uses the area division method in step 3 to perform a preliminary judgment. The specific operations are:

[0076] First, the entire correction area is divided into four equal parts along the X-axis and the Y-axis to form four quadrants, namely Figure 3 The four quadrants of upper left, lower left, upper right and lower right;

[0077] Secondly, on the circular texture target outline corresponding to the initial positioning area, the Canny operator is used to detect the edge position, and the gray value change trend of the edge position is counted, that is, the average difference in the gray value of the pixels outside the circular texture target outline and inside the circular texture target outline;

[0078] Comparing the grayscale value change trends in the four quadrants, when approaching the actual circular texture target position, the edge grayscale change of the initially positioned target is gentle, let it be gray mild Therefore, we only need to find the area with the most gentle edge grayscale change, and the corresponding quartered area is the correction area that needs to be shifted, that is, the quadrant area with the smallest average difference in the grayscale value of the pixel points outside the circular texture target outline and inside the circular texture target outline is used as the search area A. orien ,for:

[0079] gray mild ∈A orien∈A all

[0080] Among them, A all Indicates the correction area;

[0081] Since the position error of the circular texture target area can be mapped to the position error of its center, the correction of the circular texture target position can be converted into the radial difference of the center of the initial positioning area to guide the correction. orien Set a square candidate area in the box, the specific border size is Square size , and one corner of the candidate area coincides with the origin of the coordinate system O-XY, and the two adjacent sides of the candidate area coincide with the X axis and the Y axis respectively. In order to include the center coordinates of the actual position of the circular texture target in the candidate area, the calculation formula of its border size is set as:

[0082] Square size =a•x 3 +b·x 2 +c·x+d

[0083] Among them, Square size are the width and height of the candidate area, x is the maximum grayscale difference in the search area, a, b, c are the size coefficients obtained by fitting, and d is the basic size value, usually d = 1.

[0084] In the specific implementation process, the fitting process of size coefficients a, b, and c is:

[0085] First, a series of data is selected, which include the actual center coordinates of at least three circular texture targets;

[0086] Secondly, the distance between the center of the initial positioning area of ​​the circular texture target and the actual center of the circular texture target is used as the width and height of the candidate area to ensure that the actual center of the circular texture target can be included in the candidate area;

[0087] Finally, the selected data is substituted into the above calculation formula to determine the values ​​of the three size coefficients a, b, and c, and further obtain the width and height sizes of the candidate area.

[0088] After template matching, the circular texture target can obtain an initial positioning center position (i.e., the center position coordinates of the initial positioning area), which will traverse each pixel value in the candidate area. The specific traversal idea is: starting from the initially positioned circular texture center, set as the seed point; starting to grow from the pixel points in the eight neighborhoods around the seed point, merge the points in the candidate area and continue to grow outward until it touches the edge of the candidate area, thus completing the growth of the candidate area. During the growth process, each time it extends to a pixel candidate point position, the coordinates of the current candidate point are used as the starting point of the ray, and the ray begins to be drawn out, and the ray is extended to the end of the correction area, that is, Figure 4 As shown. Then calculate and store the radial difference of the data characteristics of the pixel points falling on the ray, and then obtain the radial difference of the data characteristics corresponding to each candidate point. Finally, extract the candidate point with the largest radial difference of the data characteristics as the backup point of the actual center point of the circular texture target. Therefore, the specific implementation process of calculating the radial difference of the data characteristics of the candidate points is as follows:

[0089] Step 5.1: For any candidate point, draw several rays from the candidate point, and extend the rays to the correction area. The angle between each two adjacent rays is θ;

[0090] Step 5.2, calculate the radial difference of the data characteristics of each ray, including:

[0091] According to the edge of the initial positioning area, the rays are divided into the inner ray part and the outer ray part;

[0092] Calculate the data characteristic values ​​of the pixel points covered by the ray part inside the edge and the ray part outside the edge respectively;

[0093] The difference in data characteristic values ​​between the ray portion inside the edge and the ray portion outside the edge is calculated, which is the radial difference in data characteristic of the ray;

[0094] In step 5.3, based on the radial difference of the data characteristics of each ray, the radial difference of the data characteristics of the candidate point can be obtained.

[0095] In step 5.1, due to the diversity and resolution differences of circular texture target images, in order to ensure that the circular texture target can obtain sufficient data guidance to achieve the purpose of correction, the angle of each two rays is determined by the data characteristic density of the current candidate area, that is:

[0096] θ=90° / int(A / T num )

[0097] Where A is the area of ​​the candidate region, T numis the number of pixels in the candidate region. When the density of pixels in the candidate region is high, the angle of the ray can be set wider; when the density of pixels in the candidate region is low, more rays need to be set to ensure that the rays in the quadrant area can significantly represent the data changes in the region.

[0098] In step 5.2, the data characteristic value of the pixel point can be a series of parameters that can characterize the data characteristics of the pixel point, such as entropy distribution, grayscale mode, grayscale maximum, grayscale median and position index. In step 5.3, the radial difference of the data characteristics of the candidate point corresponding to each ray can be added, that is, the radial difference of the data characteristics of the ray can be obtained. For example, such as using the change of entropy to guide the correction of the circular texture target, for any candidate point in the candidate area, the radial difference of the entropy distribution of the pixel point in the target area and the correction area on each ray of the candidate point derived from the current position is counted, and then the radial difference of the entropy distribution corresponding to each ray corresponding to the candidate point is added to obtain the radial difference of the entropy distribution corresponding to the candidate point; the radial difference of the entropy distribution corresponding to each candidate point is saved in the storage module M, and finally the radial difference of the entropy distribution corresponding to all candidate points in the storage module M is compared to extract the largest radial difference I of the entropy distribution. max , the corresponding candidate point is the actual center point of the circular texture target, that is, I max =MAX[M].

[0099] It is worth noting that when calculating on rays, due to the randomness of the data characteristics of the pixel points falling on the rays, some radial difference singular values ​​will cause the calculated difference on the ray to be too small or too large, thereby ignoring the direction of the ray, but the corresponding ray may be in the direction that needs to be corrected, thus causing misjudgment. Based on this, since the sine function has the effect of amplifying the minimum value and suppressing the maximum value within [0,2π], the sine function is used in this embodiment to perform numerical correction on the radial difference P of the data characteristics on each ray. The numerical correction process is as follows:

[0100] P new =sin(P)

[0101] Among them, P is the radial difference of the ray data characteristics, P new is the sine of the radial difference of the ray's data characteristics.

[0102] That is, in step 5.3, the sine values ​​of the radial differences of the data characteristics of the candidate points corresponding to each ray are preferably added together to obtain the radial differences of the data characteristics of the ray.

[0103] In step 5, the candidate point with the largest radial difference in the data features is only a backup point. This is because the candidate area is formed by data fitting. When an image with large positional deviations is encountered, it may be impossible to include the actual center of the circular texture target in the candidate area, and in this case, it will be impossible to obtain an accurate correction result. Therefore, this embodiment designs a feedback mechanism. When a backup point for the center of the circular texture target is obtained in step 5, it is determined whether the backup point meets the conditions for positioning termination:

[0104] Return to the step in S002 and calculate the grayscale change value V of the edge pixel value of the target outline at the center of the current circular texture target. When the grayscale change value V of the edge pixel value of one area in the four directional areas is greater than twice the grayscale value average value Vavg of the other three areas, it means that the pixel point selected in S003 is not the correct target area center. At this time, expand the original candidate area size by 1.5 times to obtain a new candidate area size Square' size .Right now:

[0105] If satisfied, output the corresponding candidate point as the center coordinate of the circular texture target;

[0106] Otherwise, increase the range of the candidate area and perform steps 5 to 6 again. The specific range of increasing the candidate area is:

[0107] Square size-1 =ω3·Square size-0

[0108] Among them, Square size-1 The width and height of the candidate region after the range is enlarged, Square size-0 The width and height of the candidate region before range enlargement, ω3 is an enlargement coefficient greater than 1, for example, it can be set to ω3=1.5.

[0109] In the specific implementation process, the positioning termination conditions are as follows:

[0110] Taking the corresponding candidate point as the center coordinate of the circular texture target, calculate the grayscale change values ​​V1, V2, V3, and V4 of the circular texture target edge in the four quadrants respectively. The grayscale change value is the difference between the mean grayscale values ​​of the pixels inside and outside the circular texture target edge.

[0111] When the edge grayscale change value of a quadrant area is greater than the edge grayscale change value of the other three quadrant areas by a multiple, the positioning termination condition is not met, that is, if there is When , it is determined that the positioning termination condition is not met, otherwise the positioning termination condition is met, where i = 1 or 2 or 3 or 4, ω2 is a determination coefficient greater than 1, for example, it can be set to ω2 = 2.

[0112] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the inventive concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A circular texture target positioning method based on radial difference guidance, characterized in that: The steps include: Step 1: Using the circular texture target image obtained by template matching as the initial positioning area, and calculating the center position coordinates of the initial positioning area in the original image; Step 2: Delimiting a correction region in the original image based on the center coordinates of the initial positioning region and the diameter of the initial positioning region, wherein the center of the correction region and the center of the initial positioning region both coincide with the origin of the coordinate system O-XY; Step 3: Divide the correction area into four quadrants along the X-axis and the Y-axis, and use the quadrant where the grayscale change of the edge of the initial positioning area is the most gentle as the search area; Step 4: Delineate a rectangular candidate region in the search area, wherein a corner point of the candidate region coincides with the origin of the coordinate system O-XY, and two adjacent sides of the candidate region coincide with the X axis and the Y axis respectively; Step 5: Define the pixel points in the candidate area as candidate points, traverse each candidate point, draw a number of rays with each candidate point as the starting point, and calculate the radial difference of the data feature of each candidate point based on the pixel points covered by the ray in the correction area. The calculation process is as follows: Step 5.1: For any candidate point, draw a number of rays from the candidate point, and extend the rays to the end of the correction area. The angle between each two adjacent rays is ; Step 5.2, calculating the radial difference of the data characteristic of each ray, including: dividing the ray into an inner-edge ray portion and an outer-edge ray portion according to the edge of the initial positioning area; Calculating the data characteristic values ​​of the pixel points covered by the inner ray portion of the edge and the outer ray portion of the edge respectively; Calculate the difference in data characteristic values ​​between the ray portion inside the edge and the ray portion outside the edge, which is the radial difference in data characteristic of the ray; Step 5.3, based on the radial difference of the data characteristic of each ray, the radial difference of the data characteristic of the candidate point can be obtained; Step 6: Determine whether the candidate point with the largest radial difference of data features meets the positioning termination condition; If so, output the corresponding candidate point as the center coordinate of the circular texture target; Otherwise, the range of the candidate region is increased and steps 5 to 6 are performed again.

2. The circular texture target positioning method based on radial difference guidance according to claim 1 is characterized in that: In step 1, the center position coordinates of the initial positioning area in the original image are calculated as follows: Performing Gaussian filtering on the circular texture target image corresponding to the initial positioning area; According to the circular texture target image after Gaussian filtering, the center position coordinates of the initial positioning area in the original image are calculated. ,for: in, The largest Coordinate values ​​and Coordinate values, is the smallest in the initial positioning area Coordinate values ​​and Coordinate value.

3. The circular texture target positioning method based on radial difference guidance according to claim 1, characterized in that: In step 2, the correction area is wide and high A rectangular area, where , is the diameter of the initial positioning area, is a regional coefficient greater than 1.

4. The circular texture target positioning method based on radial difference guidance according to claim 1, characterized in that: In step 4, the width and height of the candidate region are: in, is the width and height of the candidate region, is the maximum value of the grayscale difference in the search area, 、 、 is the size coefficient obtained by fitting, The base value of the size.

5. The circular texture target positioning method based on radial difference guidance according to any one of claims 1 to 4, characterized in that: In step 5.1, the angle between two adjacent rays is specifically: in, is the area of ​​the candidate region, is the number of pixels in the candidate region.

6. The circular texture target positioning method based on radial difference guidance according to any one of claims 1 to 4, characterized in that: In step 5.2, the data characteristic value is entropy distribution, mode, maximum value, median value or position index.

7. The circular texture target positioning method based on radial difference guidance according to any one of claims 1 to 4, characterized in that: In step 5.3, the calculation process of the radial difference of the data characteristics of the candidate points is specifically as follows: Adding the radial difference of the data characteristic of each ray corresponding to the candidate point, thereby obtaining the radial difference of the data characteristic of the ray; or The sine values ​​of the radial differences of the data characteristics of each ray corresponding to the candidate point are added together to obtain the radial differences of the data characteristics of the ray.

8. The circular texture target positioning method based on radial difference guidance according to any one of claims 1 to 4, characterized in that: In step 6, the positioning termination condition is specifically: Take the corresponding candidate point as the center coordinate of the circular texture target and calculate the grayscale change value of the edge of the circular texture target in the four quadrants respectively. 、 、 、 ; When there is When , it is determined that the positioning termination condition is not met, otherwise it meets the positioning termination condition, where 1 or 2 or 3 or 4, The determination coefficient is greater than 1.

9. The circular texture target positioning method based on radial difference guidance according to any one of claims 1 to 4, characterized in that: In step 6, the scope of increasing the candidate area is specifically as follows: in, is the width and height of the candidate region after the range is enlarged, To increase the width and height of the candidate region before the range is increased, is an increase factor greater than 1.

Images