Target volume detection method and system based on binocular vision and storage medium thereof

Through the target volume detection method based on binocular vision, using image preprocessing, highlight removal and edge contour extraction, combined with the shell method and least squares optimization, the problem of high-precision measurement of colloid volume in a small space and high-interference environment is solved, and real-time and accurate detection of colloid volume is achieved.

CN120823187APending Publication Date: 2025-10-21XIAMEN UNIV OF TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202510982258.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-16
Publication Date
2025-10-21

AI Technical Summary

Technical Problem

Traditional methods make it difficult to perform high-precision, real-time three-dimensional measurement of colloid volume in confined spaces and high-interference environments. This is especially true in the field of microelectronics packaging, where the high reflectivity and textureless properties of colloids make binocular vision volume detection difficult.

Method used

A target volume detection method based on binocular vision is adopted, including image preprocessing, highlight removal, edge contour extraction and shell method to construct a mathematical model. Combined with least squares optimization, the colloid volume is calculated through mirror highlight detection and removal algorithm, image segmentation method and shell method.

Benefits of technology

It achieves high-precision, real-time measurement of colloid volume in a small space and high-interference environment, improves the accuracy and consistency of detection, reduces measurement errors, and meets the high-precision measurement needs in the field of microelectronics packaging.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120823187A_ABST
    Figure CN120823187A_ABST
Patent Text Reader

Abstract

The invention provides a target volume detection method and system based on binocular vision and a storage medium, and the method comprises the steps: obtaining a binocular target image of a to-be-detected target point location through a binocular camera, carrying out the preprocessing of the binocular target image, and obtaining a gray binocular target image without highlight interference based on a mirror surface highlight detection and removal algorithm; obtaining a top view target edge contour and a side view target edge contour of the gray binocular target image based on an image segmentation method, and extracting morphological feature information corresponding to the top view target edge contour and the side view target edge contour; according to the extracted morphological feature information, a mathematical model for calculating the volume of a to-be-measured target is constructed through a shell layer method, and the to-be-measured target is a round point-like rotating body. The method aims at accurately measuring the volume of the dotted rotating body, the precision and efficiency of target volume detection are remarkably improved, the application range of the binocular vision technology in the field of volume measurement is greatly expanded, and development and progress of related industries are promoted.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of computer vision and image processing, and in particular relates to a target volume detection method and system based on binocular vision and a storage medium thereof. Background Art

[0002] In industries such as electronics manufacturing, medical devices, and optoelectronics, the demand for high precision and high efficiency continues to rise, making accurate colloid volume measurement crucial in these fields. In microelectronics packaging, in particular, as circuit integration continues to increase, the requirements for colloid volume measurement are becoming increasingly stringent. Dispensing operations often occur in confined spaces with high-density component layouts, requiring precise and minute colloid volume measurements. However, in confined working spaces or environments with high interference, rapid and accurate three-dimensional measurement becomes a technical challenge.

[0003] Traditional drainage methods cannot accurately measure the volume of colloids due to their high viscosity and irregular shape. Weighing methods are offline, inefficient, and lack real-time measurement, impacting production efficiency and increasing labor costs. Traditional binocular vision volume measurement relies on stereo matching technology. The textureless and highly reflective nature of colloids makes it difficult to acquire and match pheromones in binocular images. Measuring colloid volume with high precision and consistency has become a pressing challenge in modern production. Summary of the Invention

[0004] The purpose of this application is to provide a target volume detection method, system and storage medium based on binocular vision to solve the above technical problems.

[0005] According to one aspect of the present application, a method for detecting target volume based on binocular vision is proposed, the method comprising the following steps: S1. Using a binocular camera to obtain a binocular target image of a target point to be measured, the binocular target image includes a top view and a side view of the target; S2. After preprocessing the binocular target image, a gray binocular target image with highlight interference removed is obtained based on a specular highlight detection and removal algorithm; S3, based on the image segmentation method, respectively obtaining the top view target edge contour and the side view target edge contour of the gray binocular target image, and extracting the morphological feature information corresponding to the top view target edge contour and the side view target edge contour; S4. Based on the extracted morphological feature information, a mathematical model for calculating the volume of the target to be measured is constructed by the shell method, wherein the target to be measured is a dot-shaped rotating body.

[0006] In the above technical solution, images are acquired through a binocular camera, and a mathematical model is constructed through preprocessing, highlight removal, edge contour extraction and shell method to realize the method of volume detection of dot-shaped rotating body targets, thereby improving the accuracy and pertinence of target volume detection.

[0007] Furthermore, the method also includes using the least squares method to modify and optimize the mathematical model. Using the least squares method to modify and optimize the mathematical model further improves the accuracy of the target volume detection method and reduces measurement errors.

[0008] Furthermore, step S2 includes the following sub-steps: S21, preprocessing the binocular target image by using distortion correction and epipolar correction; S22. By calculating the 95th percentile ratio of the RGB channel of the binocular target image to the grayscale image, the specular highlight area is extracted and the highlight mask area is generated, where the specular highlight area is represented as , where , , Represents the red, green, and blue channel values ​​of the pixel respectively. Represents the grayscale value of the pixel, Indicates the preset baseline threshold; S23. Calculate the average color of the pixels in the highlight mask area, fill all pixels in the connected area with the average color, and obtain a gray binocular target image.

[0009] In the above technical solution, through preprocessing steps such as image preprocessing, highlight area extraction and filling, the foundation is laid for the subsequent accurate acquisition of target edge contour and morphological feature information, thereby enhancing image quality and data reliability.

[0010] Furthermore, the S3 steps include: S31, converting the gray binocular target image into a grayscale image and performing binarization processing; S32. Obtaining the edge contour of the target in the top view and the edge contour of the target in the side view by using an image segmentation method, where the image segmentation method includes one or more of morphological operations, multi-scale filtering, and contour detection; S33. Calculate the physical radius of the target edge contour in the top view by using the ring fitting and radius extraction method. Based on the target edge contour in the side view, use the edge curve extraction to obtain the edge curve fitting function of the morphological feature information.

[0011] In the above technical solution, grayscale conversion, binarization and various image segmentation methods are used to obtain the target edge contour, and the physical radius and edge curve fitting function are calculated through ring fitting, etc. to improve the target morphological feature extraction and assist in accurate volume measurement.

[0012] Furthermore, step S33 includes: S331, randomly selecting a number of points in the edge contour of the target in the top view, and linearizing the equation using the least squares method to obtain a fitted outer ring; S332, calculating the disparity between the center of the fitted outer circle and a point on the circle, and calculating the depth information of the two points based on the intrinsic parameters of the binocular camera; S333. Calculate the three-dimensional coordinates of the target using the depth information, and calculate the physical radius of the target edge contour in the top view using the two-point distance formula. .

[0013] In the above technical solution, the physical radius of the edge contour of the target in the top view is accurately calculated based on specific steps, which provides key parameters for constructing a volume calculation model and improves the accuracy of volume calculation.

[0014] Furthermore, the edge curve fitting function of the side view target edge contour The calculation formula is as follows:

[0015] Where, is represented as the fitting parameter, , , It is represented as the segmentation point of the OQ curve segment, Represents the endpoint of the second MQ curve.

[0016] In the above technical solution, the edge curve fitting function formula of the side view target edge contour is obtained to accurately describe the target shape characteristics and optimize the volume calculation model.

[0017] Furthermore, step S4 includes: S41. Calculate the pixel volume of the target using the shell method based on the edge curve fitting function of the morphological feature information ,Right now , where Expressed as pixel volume, Expressed as the horizontal coordinate value of the rotation axis in the pixel coordinate system, Expressed as edge curve fitting function in pixel coordinate system; S42, using the physical radius of the morphological feature information and the fitting curve The ratio of the lengths of the integration intervals on the axis is used to calculate the scaling factor ; S43, based on scale conversion factor Perform proportional transformation on the parameters of the two-dimensional coordinates to obtain the physical volume of the target to be measured ,Right now .

[0018] In the above technical solution, the edge curve fitting function is combined with the shell method and the scale conversion factor to calculate the physical volume of the target to be measured, thereby achieving accurate conversion from pixel volume to physical volume and improving the practicality and accuracy of volume measurement.

[0019] In a second aspect, the present application proposes a target volume detection system based on binocular vision, the system comprising: An image acquisition module is configured to acquire a binocular target image of a target point to be measured using a binocular camera, wherein the binocular target image includes a top view and a side view of the target; The gray binocular target image processing module is configured to obtain a gray binocular target image with highlight interference removed based on a specular highlight detection and removal algorithm after preprocessing the binocular target image; An image segmentation and extraction module is configured to obtain the top view target edge contour and the side view target edge contour of the gray binocular target image based on the image segmentation method, and extract morphological feature information corresponding to the top view target edge contour and the side view target edge contour; The model building module constructs a mathematical model for calculating the volume of the target to be measured through the shell method based on the extracted morphological feature information; wherein the target to be measured is a dot-shaped rotating body.

[0020] In the above technical solution, a system including image acquisition, processing, segmentation and extraction, and model building modules is constructed, which work together to achieve efficient detection of the volume of dot-shaped rotating body targets, thereby improving detection efficiency and accuracy.

[0021] In a third aspect, this application proposes a binocular vision-based colloid volume detection method, employing the binocular vision-based target volume detection method described in the first aspect and / or the binocular vision-based target volume detection system described in the second aspect, wherein the dot-shaped quasi-rotating bodies are colloid materials. This method achieves high-precision volume measurement based on the characteristics of the colloid, meeting the requirements of colloid production inspection.

[0022] Fourthly, this application proposes a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements any of the aforementioned binocular vision-based target volume detection methods and / or binocular vision-based colloid volume detection methods. Executing the target volume detection method using the computer program stored in the storage medium facilitates the widespread application of the method, improving its operability and accessibility.

[0023] Compared with the prior art, the present invention has the following advantages: (1) This application is specifically used for volume measurement of dot-shaped colloids on a three-axis platform. It is designed based on the highly reflective and textureless characteristics of colloids, overcomes the limitations of general methods, and accurately adapts to the needs of colloid measurement.

[0024] (2) This application takes into account the properties of colloids and uses specialized technology to address the problems of high reflectivity causing interference with mirror highlights and the lack of texture causing failure of traditional texture measurement methods, thereby ensuring measurement accuracy. (3) This application uses precise image acquisition and processing and an algorithm adapted to dot-shaped rotating bodies to measure the volume of the target with high precision, thereby improving the reliability and consistency of the results and reducing errors.

[0025] (4) In actual production, accurate measurement of colloid volume can improve product quality and production efficiency, and provide powerful tools and intuitive examples for practical teaching of related majors. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated into and constitute a part of this specification. The accompanying drawings illustrate the embodiments and, together with the description, serve to explain the principles of the present invention. Other embodiments and many of the expected advantages of the embodiments will be readily apparent as they become better understood by reference to the following detailed description. The elements of the drawings are not necessarily to scale with respect to each other. Like reference numerals designate corresponding similar parts.

[0027] Figure 1 is a flow chart of a method for detecting target volume based on binocular vision according to one embodiment of the present application; Figure 2 is a flow chart of a colloid volume detection method based on binocular vision according to one embodiment of the present application; Figure 3 1 is a schematic diagram of dual-view image acquisition of a colloid volume detection method based on binocular vision according to an embodiment of the present application; Figure 4 A two-dimensional parallel structure of a binocular vision system for a colloid volume detection method based on binocular vision according to an embodiment of the present application; Figure 5 1 is a schematic diagram of side view target edge contour image processing of a colloid volume detection method based on binocular vision according to an embodiment of the present application; Figure 6 This is a framework diagram of a target volume detection system based on binocular vision according to an embodiment of the present application; Figure 7 It is a structural diagram of a computer system suitable for implementing the electronic device of the embodiment of the present application. DETAILED DESCRIPTION

[0028] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0029] refer to Figure 1 , Figure 1 The flowchart of the target volume detection method based on binocular vision of the present application is shown. As shown in the figure, the method includes the following steps: S101. Acquire a binocular target image of a target point to be measured using a binocular camera, wherein the binocular target image includes a top view and a side view of the target, wherein the target to be measured is a dot-shaped rotating body.

[0030] In some specific embodiments, before the binocular camera is operated for the first time, the Zhang Zhengyou calibration method is used to calibrate the binocular camera to determine the mapping relationship between the world coordinate system and the image coordinate system of the binocular camera. The core of the calibration process is to calculate the intrinsic and extrinsic parameter matrices of the camera and establish a mapping formula from the world coordinate system to the image coordinate system. This relationship is described by the following formula: Internal parameter matrix , used to describe the internal parameters of the camera, and its form is:

[0031] Where, , Indicates that the camera is and focal length in direction; , The main point coordinates.

[0032] External parameter matrix , which is used to describe the relative position relationship between the world coordinate system and the camera coordinate system. It consists of the rotation matrix R and the translation vector T, and its form is:

[0033]

[0034]

[0035] Where, Represents the rotation from the world coordinate system to the camera coordinate system, Represents the translation vector, which defines the displacement relationship between the two coordinate systems.

[0036] The mapping from the world coordinate system to the pixel coordinate system can be done by combining the intrinsic matrix and the extrinsic matrix to convert the point in the world coordinate system ( , , ) is mapped to the point in the image coordinate system ( , ), the specific mapping formula is:

[0037] Where, Expressed as a scaling factor, ( , ) is expressed as pixel coordinates on the image plane, Represented as the camera’s intrinsic parameter matrix, Represented as the camera's extrinsic parameter matrix.

[0038] S102 , after pre-processing the binocular target image, a gray binocular target image with highlight interference removed is obtained based on a specular highlight detection and removal algorithm.

[0039] In some specific embodiments, the binocular target image is preprocessed including distortion correction and epipolar correction to obtain a corrected binocular image.

[0040] Specifically, distortion correction includes radial distortion correction and tangential distortion correction. The radial distortion correction formula is:

[0041]

[0042] The tangential distortion correction formula is:

[0043]

[0044] Where, 、 、 Expressed as the radial distortion coefficient of the camera, 、 Expressed as the tangential distortion coefficient of the camera.

[0045] The camera external parameters reflect the rotation between the camera coordinate system and the world coordinate system and pan Relationship. If the intrinsic parameters of the two cameras and the rotation matrices between them and the world coordinate system are known , and translation matrices , , the relative rotation between the two cameras can be calculated and relative translation , that is, to realize the coordinate transformation from one camera coordinate system to another camera coordinate system, the calculation formulas for relative rotation and relative translation are respectively 、 The above formula can be used to determine the spatial position and direction relationship between the two cameras, providing a basis for geometric calculations in stereo vision.

[0046] Specifically, the purpose of epipolar correction is to adjust the epipolar lines of the two images to horizontal lines so that the corresponding points of the same name in the two images are located on the same horizontal line, thereby simplifying the disparity calculation and improving the matching efficiency.

[0047] Basic Matrix : In binocular system, is the basic matrix; is the point in the left image; are the corresponding points in the right image. The corresponding points of the two images satisfy the following constraints:

[0048]

[0049]

[0050] Homography matrix , : Epipolar correction via homography matrix and Transform the image coordinates so that the epipolar lines become horizontal lines, where and are the coordinates of the corrected points, then , .

[0051] Pole Constraints: Pole and For the intersection of the polar lines, the poles are aligned to the horizontal infinity position through correction:

[0052]

[0053] Corrected projection formula: After correction, the points of the left and right images are and Satisfies on the same line: ,Through the above formula, epipolar correction is achieved to align the left and right binocular images.

[0054] In some specific embodiments, the specular highlight detection and removal algorithm includes the following sub-steps: S1021. Calculate the 95th percentile ratio of the RGB channels of the binocular target image to the grayscale image to extract the specular highlight area and generate a highlight mask area. The specular highlight area is expressed as: , where , , Represents the red, green, and blue channel values ​​of the pixel respectively. Represents the grayscale value of the pixel, Indicates the preset baseline threshold.

[0055] Specifically, due to the uniformity of reflection of the gray target in the red, green, and blue channels, it is difficult for a single channel to accurately detect the highlight area, and a fixed threshold is prone to false detection or missed detection. In order to balance the brightness difference of the channels, the specular reflection detection (SR detection) is performed by calculating the red, green, and blue channels ( , , ) and grayscale image ), extract the specular highlight area and generate a highlight mask to achieve adaptive color balance detection. The specific calculation formula is as follows:

[0056]

[0057]

[0058]

[0059] Where P95(c) represents the 95th percentile of channel c. The value represents the lower limit of the brightest 5% pixels in that channel. Using the 95th percentile brightness threshold can effectively compensate for color balance issues, making detection more robust under different lighting conditions and color balance settings.

[0060] Based on the above ratios, an initial highlight area mask can be generated. For each pixel, if its red, green, blue or gray value exceeds the corresponding adjustment threshold, it is marked as a possible target specular highlight:

[0061] in, It is a preset baseline threshold. By multiplying it with the adjusted channel ratio, it can more effectively take into account the target highlight areas that may exist in each channel, making highlight detection more comprehensive and accurate.

[0062] S1022. Calculate the average color of pixels in the highlight mask area, fill all pixels in the connected area with the average color, and obtain a gray binocular target image.

[0063] Specifically, highlight area filling is used to process large-area highlight masks and eliminate interference from non-target highlight areas. Based on connectivity, small highlight areas are not filled due to insufficient connectivity, and the target highlight area is retained to facilitate subsequent accurate extraction of the target highlight mask. For each highlight mask area, the average color of all pixels in the area is calculated. The specific calculation method is as follows:

[0064] Where, Indicates the The average color of the connected regions, Indicates the number of pixels in the area, Indicates the area Pixel index.

[0065] Fill all pixels within each connected region with the average color of the region: , Preferably, step S1022 is followed by multi-scale processing, i.e., multi-scale processing is introduced during the restoration process to improve the detection and restoration of highlight areas by analyzing images and highlight masks at different scales. Compared to single-scale processing, multi-scale processing is more adaptable to the diversity of highlight areas and improves the accuracy and robustness of fill fusion.

[0066] S103 , obtaining the top view target edge contour and the side view target edge contour of the gray binocular target image based on an image segmentation method, and extracting morphological feature information corresponding to the top view target edge contour and the side view target edge contour.

[0067] In some specific embodiments, the step of segmenting and extracting the edge contour of the top view object and the edge contour of the side view object includes: S1031, converting the gray binocular target image into a grayscale image and performing binarization processing; Specifically, the target to be tested RGB The image is converted into a grayscale image and binarized using the adaptive threshold method to convert the target area into white and the background area into black to achieve separation of foreground and background.

[0068] S1032. Obtaining the edge contour of the target in the top view and the edge contour of the target in the side view using an image segmentation method, where the image segmentation method includes one or more of morphological operations, multi-scale filtering, and contour detection; Specifically, morphological operations remove noise points through opening operations and connect broken edges through closing operations, thereby enhancing the continuity and integrity of the target object's edges; multi-scale filtering uses Gaussian smoothing methods to perform multi-scale processing on images, reduce noise and smooth edges, and fuse results at different scales to improve the robustness of segmentation; contour detection is used to detect all contours in the image, sort them according to area, and select the contour with the largest area as the target contour to ensure accurate segmentation of the target object contour in top and side views.

[0069] S1033. Calculate the physical radius of the target edge contour in the top view using the ring fitting and radius extraction method. Based on the target edge contour in the side view, extract the edge curve to obtain an edge curve fitting function of the morphological feature information.

[0070] Specifically, the ring fitting is specifically for the target image contour of the top view of the target to be measured. Since its shape is approximately disk-shaped, the outer ring is fitted through the edge contour points. The fitting process is based on the least squares method. By minimizing the sum of square errors between the point set and the fitting circle, the center coordinates and radius that best fit the target contour are obtained. Several points are randomly selected from the edge contour of the top view target to construct the data point set ( , ). The center of the fitting circle is ( , ), the radius is , the standard equation is:

[0071] Linearize the equation using the least squares method: , Where, , so as to quickly obtain the fitted outer ring.

[0072] Specifically, the parallax between the center of the circle and a point on the circle is calculated when fitting a circle in the left and right eye views. The depth of the two points is calculated using the depth calculation formula, and the radius is finally obtained using the distance formula between the two points. After binocular epipolar correction, the same-name points in the left and right image pairs are aligned to the same horizontal epipolar line. The images are adjusted to a parallel structure. In the left and right views, the parallax of the corresponding points on the horizontal line is calculated. , combined with the camera internal parameters (focal length , baseline length ), the depth of the target object can be obtained.

[0073] The parallax formula is:

[0074] According to parallax and camera parameters, depth It can be calculated by the following formula:

[0075] Using the depth information, we can further calculate the coordinates of the target point in three-dimensional space. Substitute the pixels of the left and right views into the triangular relationship to obtain the three-dimensional coordinates of the target object. Combined with the three-dimensional coordinates, the actual physical distance between the boundary point of the object and the center of the circle is further calculated. Assume that the coordinates of the two points are A( , , ) and B( , , ), the distance formula between two points is:

[0076] Where, Indicates the distance between the center of the object and the boundary point, that is, the physical radius length of the object.

[0077] Specifically, edge curve fitting is used to process the side view image of the target to be measured, and a segmented fitting model is constructed by extracting edge points of the upper half of the contour of the target to be measured.

[0078] Edge extraction: In the target image, the edge points of the side view target edge contour are extracted, and the upper half of the edge is selected as the target area for the fitting curve. This target area usually presents a certain arc or convex shape. The edge curve fitting segmentation process is to divide the edge of the target area into two segments, namely the OM segment curve and the MQ segment curve. The OM segment curve is described by a logarithmic function, and its expression is:

[0079] Where, and is the fitting parameter.

[0080] The MQ segment curve is described by a quadratic polynomial function, which keeps the curve smooth and symmetrical. Its expression is:

[0081] Where, 、 and is the fitting parameter.

[0082] Combining the above two curves, we get the edge curve fitting function , as shown below:

[0083] Where, is represented as the fitting parameter, , , It is represented as the OQ segment point of the curve segment, Represents the endpoint of the second MQ curve.

[0084] S104 . Constructing a mathematical model for calculating the volume of the target to be measured by a shell method based on the extracted morphological feature information, wherein the target to be measured is a dot-shaped rotating body.

[0085] In some specific embodiments, the construction of the mathematical model of the volume of the target to be measured is based on the fact that the target to be measured is a quasi-rotational body in the shape of a dot as a whole. Assuming that the target to be measured is an axisymmetric object rotating about a vertical direction, its volume can be calculated using the shell method, and the object volume is solved using the scale conversion factor, and finally a volume mathematical model is established. The mathematical model of the volume of the target to be measured includes the shell method, the scale conversion factor, the mathematical model construction, and the model correction. It specifically includes the following steps: S1041, based on the edge curve fitting function of the morphological feature information, the pixel volume of the target is calculated using the shell method ,Right now , where Expressed as pixel volume, Expressed as the horizontal coordinate value of the rotation axis in the pixel coordinate system, Expressed as edge curve fitting function in pixel coordinate system; Specifically, the shell method formula can be expressed as:

[0086] Where, is the pixel volume of the rotation body, (0, ) is the fitting curve in The integration interval on the axis, is the horizontal coordinate value of the rotation axis, is the edge curve fitting function, is the radius of rotation.

[0087] In the pixel coordinate system, the pixel volume formula of the target to be measured can be expressed as:

[0088] in, is the pixel volume, is the horizontal coordinate value of the rotation axis in pixel coordinates, Expressed as the edge curve fitting function under pixel coordinates, is the edge curve fitting function Substitute the physical quantities of the pixel coordinate system into the edge curve fitting function.

[0089] S1042, using the physical radius of the morphological feature information and the fitting curve The ratio of the lengths of the integration intervals on the axis is used to calculate the scaling factor ; Specifically, according to the physical radius of the target edge contour of the top view in step S1033 , and the fitting curve is Length of the on-axis integration interval , calculate the scale conversion factor between physical size and pixel size :

[0090] S1043, based on scale conversion factor Perform proportional transformation on the parameters of the two-dimensional coordinates to obtain the physical volume of the target to be measured ,Right now .

[0091] Specifically, the scale conversion factor is used to perform proportional transformation on the parameters in the two-dimensional real-world coordinate system, and the following relationship can be established:

[0092] Among them, the above physical quantities are the mapping values ​​of the pixel coordinate system in the two-dimensional real-world coordinate system. According to the above relationship, the physical volume integral formula of the target to be measured can be obtained:

[0093]

[0094]

[0095] in, V phy is the physical volume of the target to be measured, | x phy -x phy0 | is the physical rotation radius. By calculating the above formula, we can get the mathematical model of the volume of the dot-shaped target to be measured:

[0096] From the above relationship, it can be obtained that when the diameter is proportional When the double transformation is performed, The volume of the rotating body formed by the rotation is converted to the original volume times, that is, the physical volume of the target to be measured is the pixel volume times.

[0097] Preferably, the model correction adopts the least square method of first-order linear fitting to correct the original mathematical model of the target volume to improve the accuracy of volume detection. The formula is as follows:

[0098] in, is the fitting coefficient, is a constant offset, is the total number of observations, and are the original volume measurement and the actual volume value, and respectively and After applying the least squares method, the modified model form is:

[0099] in, is the approximate volume (i.e. the corrected volume output), is the volume value output by the original mathematical model, that is .

[0100] Continue to refer Figure 2 , Figure 2 The figure shows a flow chart of the colloid volume detection method based on binocular vision according to the present application, which includes the following steps: Step 201: Image preprocessing. The binocular camera captures different morphological features of the target colloid from both top and side views. Image preprocessing includes processing the rectified left and right views, which serve as input data.

[0101] Specific, combined Figure 3 , Figure 3 FIG. 4 shows a schematic diagram of dual-view image acquisition of a colloid volume detection method based on binocular vision according to an embodiment of the present invention, as shown in FIG. Figure 3 As shown in the figure, the binocular camera captures images of the target colloid from both top and side views and extracts morphological features from each perspective. To ensure accurate capture of the target colloid image, the binocular camera is used to shoot on a white background. The white background effectively avoids the overexposure problem that may be caused by a black background, thereby improving image quality. The collected colloid image is then subjected to image preprocessing, which includes epipolar correction and distortion correction for radial and tangential distortion correction: In distortion correction, 、 、 Expressed as the radial distortion coefficient of the camera, 、 Expressed as the tangential distortion coefficient of the camera, Represents the horizontal coordinate of the pixel in the original image, Represents the vertical coordinate of the pixel in the original image, Represents image coordinates The distance to the center of the image, and represents the coordinates after radial distortion correction, and represents the coordinates after tangential distortion correction, and the radial distortion correction formula is as follows:

[0102]

[0103] The tangential distortion correction formula is as follows:

[0104]

[0105] The camera external parameters reflect the rotation between the camera coordinate system and the world coordinate system and pan Relationship. If the intrinsic parameters of the two cameras and the rotation matrices between them and the world coordinate system are known , and translation matrices , , the relative rotation between the two cameras can be calculated and relative translation , that is, to achieve the coordinate transformation from one camera coordinate system to another camera coordinate system. The following formula can be used to determine the spatial position and direction relationship between the two cameras, providing a basis for geometric calculations in stereo vision. The calculation formulas for relative rotation and relative translation are:

[0106]

[0107] The purpose of epipolar correction is to adjust the epipolar lines of the two images to horizontal lines so that the corresponding points of the same name in the two images are on the same horizontal line, thereby simplifying the disparity calculation and improving the matching efficiency. Based on the basic matrix F, in the binocular system, the corresponding points of the two images meet the following constraints:

[0108]

[0109]

[0110] in, is the basic matrix, is a point in the left image, are the corresponding points in the right image.

[0111] Homography matrix , : Epipolar correction via homography matrix and Transform the image coordinates so that the epipolar lines become horizontal lines, where and They are the corrected point coordinates:

[0112]

[0113] Pole Constraints: Pole and For the intersection of the polar lines, the poles are aligned to the horizontal infinity position through correction:

[0114]

[0115] Corrected projection formula: After correction, the points of the left and right images are and Satisfies on the same line: ,Through the above formula, epipolar correction is achieved to align the left and right binocular images.

[0116] Step 202: Specular highlight detection and removal. After image preprocessing, specular highlight detection and removal are performed. The processed left and right images are de-highlighted, yielding the de-highlighted left and right images, respectively. Highlight removal facilitates more accurate colloid contour extraction.

[0117] Specifically, the purpose of specular highlight detection and removal is to effectively deal with the highlight interference generated by the gray colloid surface. By accurately detecting and removing the highlight area, the negative impact of the highlight on the image binarization process is significantly reduced, and clear and accurate colloid edge information can be quickly provided. It consists of three parts: color balance adaptation, connected area-based filling method, and multi-scale processing; Color balance adaptation: Due to the reflection uniformity of gray colloid in red, green and blue channels, it is difficult for a single channel to accurately detect the highlight area, and a fixed threshold is prone to false detection or missed detection. In order to balance the brightness difference of the channels, SR detection is performed by calculating the red, green and blue channels ( , , ) and grayscale image ) is used to extract the specular highlight area and generate a highlight mask to achieve adaptive color balance detection. P95(c) represents the 95th percentile of channel c, and the value represents the lower limit of the brightest 5% pixels in the channel. The specific calculation formula is as follows:

[0118]

[0119]

[0120]

[0121] Using the 95th percentile brightness threshold effectively compensates for color balance issues, making the detection more robust under different lighting conditions and color balance settings. Based on the above ratios, an initial highlight area mask can be generated. For each pixel, if its red, green, blue or grayscale value exceeds the corresponding adjustment threshold, it is marked as a possible target specular highlight, where It is a preset reference threshold. By multiplying it with the adjusted channel ratio, it can more effectively take into account the target highlight areas that may exist in each channel, making the highlight detection more comprehensive and accurate. The target specular highlight is expressed as:

[0122] For each pixel, if its red channel value Greater than , or the green channel value Greater than , or the blue channel value Greater than , or grayscale value Greater than , then the pixel is marked as a possible target specular highlight area.

[0123] Connected Region Filling Method: In SR detection, highlight region filling is used to process large-area highlight masks and eliminate interference from non-target highlight regions. Based on connectivity, small highlight regions are not filled due to insufficient connectivity, retaining the target highlight region and facilitating subsequent accurate extraction of the target highlight mask.

[0124] For each highlight mask area, calculate the average color of all pixels in the area. The specific calculation method is , where Indicates the The average color of the connected regions, Indicates the number of pixels in the area, Indicates the area Pixel indices, fill all pixels in each connected region with the average color of the region:

[0125] This method ensures color consistency and improves highlight detection accuracy by filling the connected regions with the average color. Compared to pixel-by-pixel processing, average color filling only requires a single calculation, significantly increasing computation speed and reducing complexity. It is particularly suitable for large highlight areas at low thresholds.

[0126] Multi-scale processing: Multi-scale processing is introduced in the restoration process. By analyzing images and highlight masks at different scales, it improves the detection and restoration of highlight areas. Compared with single-scale processing, multi-scale processing is more adaptable to the diversity of highlight areas, improving the accuracy and robustness of fill fusion.

[0127] Step 203: Image segmentation. The left and right images, after highlight removal, enter the image segmentation step. The edge contours of the colloid's top view and side view are extracted, respectively.

[0128] Specifically, the edge contour of the target colloid image is extracted through a dual-view image segmentation method. Image segmentation includes image preprocessing, morphological operations, multi-scale filtering, and contour detection.

[0129] Image preprocessing: convert the RGB image of the target colloid into a grayscale image, and use the adaptive threshold method to perform binarization processing, converting the target area to white and the background area to black to achieve the separation of foreground and background.

[0130] Morphological operations: Remove noise points through opening operations and connect broken edges through closing operations to enhance the continuity and integrity of the edges of the target objects.

[0131] Multi-scale filtering: Gaussian smoothing method is used to perform multi-scale processing on the image to reduce noise and smooth edges, and the results of different scales are fused to improve the robustness of segmentation.

[0132] Contour detection: Detect all contours in the image, sort them by area, and select the contour with the largest area as the target contour to ensure accurate segmentation of the target object contour in top and side views.

[0133] Step 204: Morphological feature extraction: The physical radius of the colloid top view is extracted from the edge contour of the colloid top view, and the fitting curve of the colloid side view is extracted from the edge contour of the colloid side view.

[0134] Specifically, morphological feature extraction includes ring fitting, radius extraction, and edge curve fitting; The ring fitting is specifically for the target image contour of the colloid top view, which is approximately disk-shaped, and the outer ring is fitted through the edge contour points. The fitting process is based on the least squares method, which minimizes the sum of squared errors between the point set and the fitted circle to obtain the center coordinates and radius that best fit the target contour. Several points are randomly selected from the target object contour to construct a data point set. The center of the fitting circle is , the radius is , the standard equation is: , linearize the equation by the least squares method ,in, , so as to quickly obtain the fitted outer ring.

[0135] Radius extraction is done by fitting a circle in the left and right eye views, calculating the parallax between the center of the circle and a point on the circle, and then using the depth calculation formula to get the depth of the two points. Finally, the radius length is obtained according to the distance formula between the two points. After binocular epipolar correction, the same-name points in the left and right image pairs are aligned to the same horizontal epipolar line, and the images are adjusted to a parallel structure. Figure 4 FIG2 shows a two-dimensional parallel structure of a binocular vision system for a colloid volume detection method based on binocular vision according to an embodiment of the present application, as shown in FIG2. Figure 4 As shown, in the left and right views, the disparity of corresponding points on the horizontal line is calculated , combined with the camera internal parameters (focal length , baseline length ), the depth of the target object can be obtained . Then the parallax formula is: , and Represents the horizontal coordinates (pixel coordinates) of the same point in space in the left camera and right camera images, disparity It is the absolute value of the difference between these two coordinate values. and camera parameters, depth It can be calculated by the formula: Using the depth information, we can further calculate the coordinates of the target point in three-dimensional space. Substitute the pixels of the left and right views into the triangular relationship to obtain the three-dimensional coordinates of the target object. Combined with the three-dimensional coordinates, the actual physical distance between the boundary point of the object and the center of the circle is further calculated. Assume that the coordinates of the two points are A( , , ) and B( , , ), Represents the distance between the center of the object and the boundary point, that is, the physical radius of the object. The distance formula between the two points is: .

[0136] Edge curve fitting is used to process the colloid side view image. The segmented fitting model is constructed by extracting the edge points of the upper half of the colloid contour. Figure 5 FIG. 1 shows a schematic diagram of side view target edge contour image processing of a colloid volume detection method based on binocular vision according to an embodiment of the present application. Figure 5 As shown in the figure, in the target colloid image, the edge points of its contour are extracted, and the upper half of the edge is selected as the target area of ​​the fitting curve. The target area usually presents a certain arc or convex shape. The edge curve fitting segmentation process is to divide the edge of the target area into two segments. The first segment is the OM segment curve. and is the fitting parameter, is the segmentation point of the OQ curve segment. The first segment of the curve is described by a logarithmic function, and the expression is: , the second segment is the MQ segment curve, Expressed as the endpoint value of the MQ curve, the second segment of the curve is described by a quadratic polynomial function, keeping the curve smooth and symmetrical. The expression is: Combining the above two curves, we get the fitting curve , as shown below:

[0137] Step 205: Constructing a mathematical model of the colloid volume.

[0138] Specifically, the colloid volume mathematical model is constructed based on the assumption that the colloid is a quasi-rotational solid, assuming it is an axisymmetric object rotating about a vertical axis. Its volume can be calculated using the shell method, and the object volume is solved using a scale conversion factor, ultimately establishing a volume mathematical model. The colloid volume mathematical model includes the shell method, scale conversion factor, mathematical model construction, and model correction. is the pixel volume of the rotation body, (0, ) is the fitting curve in The integration interval on the axis, is the horizontal coordinate value of the rotation axis, is the edge curve fitting function, is the radius of rotation, the shell method formula can be expressed as: , in pixel coordinate system, is the pixel volume, is the horizontal coordinate value of the rotation axis in pixel coordinates, It is expressed as the edge curve fitting function under pixel coordinates, that is, Substituting the physical quantities of the pixel coordinate system into the pixel coordinate lower edge curve fitting function, the pixel volume formula of the colloid can be expressed as: , the physical radius length extracted from the top view With the fitting curve Length of the on-axis integration interval , calculate the scale conversion factor between physical size and pixel size , , using the scale conversion factor to perform proportional transformation on the parameters in the two-dimensional real-world coordinate system, the following relationship can be established: , where the above physical quantities are all mapping values ​​of the pixel coordinate system in the two-dimensional real-world coordinate system. is the physical volume, is the physical rotation radius. According to the above relationship, the colloid physical volume integral formula can be obtained:

[0139] Right now , when the diameter is proportional When the double transformation is performed, The volume of the rotating body formed by the rotation is converted to the original volume times, that is, the physical volume of the colloid is times the pixel volume times.

[0140] Furthermore, the model is modified by using the least square method of first-order linear fitting to modify the original colloid volume mathematical model to improve the accuracy of volume detection. is the fitting coefficient, is a constant offset, is the total number of observations, and are the original volume measurement and the actual volume value, and respectively and The average value of is as follows:

[0141] After applying the least squares method, the revised model form is:

[0142] in, is the approximate volume (i.e. the corrected volume output), is the volume value output by the original mathematical model, that is .

[0143] This embodiment uses a motion control card to send control instructions to drive the servo motor to discharge glue through the dispensing valve. The dispensing valve is connected to a three-axis motion platform to achieve three-dimensional space dispensing. Using binocular vision technology, high-precision, real-time, non-contact measurement can be achieved. It can quickly obtain colloid volume data in complex three-dimensional space, ensuring high consistency of measurement results in a changing production environment. This application includes five parts: image preprocessing, mirror highlight detection and removal, image segmentation, morphological feature extraction, and colloid volume mathematical model construction. It uses visual technology and optimization control technology to achieve closed-loop feedback control of the dispensing valve system, which can improve production quality and efficiency and reduce costs in the production process. The application of this technology can greatly improve the accuracy and efficiency of dispensing operations, reduce costs, and promote the development of dispensing technology in the direction of miniaturization, three-dimensionalization, and high precision.

[0144] Further references Figure 6 As an implementation of the above method, in the second aspect, the present application provides an embodiment of a target volume detection system 600 based on binocular vision. Figure 1 The system 600 includes an image acquisition module 601, a gray binocular target image processing module 602, an image segmentation and extraction module 603, and a model building module 604, which are interconnected and communicate with each other. The image acquisition module 601 is configured to obtain a binocular target image of a target point to be measured using a binocular camera, wherein the binocular target image includes a top view and a side view of the target; The gray binocular target image processing module 602 is configured to obtain a gray binocular target image with highlight interference removed based on a specular highlight detection and removal algorithm after pre-processing the binocular target image; The image segmentation and extraction module 603 is configured to respectively obtain the top view target edge contour and the side view target edge contour of the gray binocular target image based on the image segmentation method, and extract the morphological feature information corresponding to the top view target edge contour and the side view target edge contour; The model building module 604 builds a mathematical model for calculating the volume of the target to be measured by the shell method according to the extracted morphological feature information, wherein the target to be measured is a dot-shaped rotating body.

[0145] The present application proposes a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned binocular vision-based target volume detection methods is implemented.

[0146] Reference below Figure 7 , which shows a structural diagram of a computer system 700 suitable for implementing a terminal device or server in an embodiment of the present application. Figure 7 The terminal device or server shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0147] like Figure 7 As shown, computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage unit 708 into a random access memory (RAM) 703. Various programs and data required for the operation of computer system 700 are also stored in RAM 703. CPU 701, ROM 702, and RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to bus 704.

[0148] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, and the like; an output section 707 including a liquid crystal display (LCD), speakers, and the like; a storage section 708 including a hard disk and the like; and a communication section 709 including a network interface card such as a LAN card or a modem. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. Removable media 711, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 710 as needed, so that computer programs read therefrom can be installed in the storage section 708 as needed.

[0149] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for executing the method illustrated in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via the communication portion 709 and / or installed from removable media 711. When the computer program is executed by the central processing unit (CPU) 701, the functions defined in the methods of the present application are performed. It should be noted that the computer-readable medium of the present application can be a computer-readable signal medium or a computer-readable medium, or any combination thereof. The computer-readable medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples of computer-readable media include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable medium may be any tangible medium that contains or stores a program for use by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, embodying computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including, but not limited to, wireless, wire, optical cable, RF, or any suitable combination thereof.

[0150] Computer program code for performing the operations of the present application can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0151] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0152] Although the principles of the present invention have been described in detail above in conjunction with the preferred embodiments of the present invention, those skilled in the art should understand that the above embodiments are merely illustrative of the present invention and are not intended to limit the scope of the present invention. The details in the embodiments do not constitute a limitation on the scope of the present invention. Without departing from the spirit and scope of the present invention, any obvious changes such as equivalent transformations and simple substitutions based on the technical solution of the present invention fall within the scope of protection of the present invention.

Claims

1. A target volume detection method based on binocular vision, characterized in that: The method comprises: S1. Acquire a binocular target image of a target point to be measured using a binocular camera, wherein the binocular target image includes a top view and a side view of the target; S2. After pre-processing the binocular target image, a gray binocular target image with highlight interference removed is obtained based on a specular highlight detection and removal algorithm; S3, respectively obtaining a top view target edge contour and a side view target edge contour of the gray binocular target image based on an image segmentation method, and extracting morphological feature information corresponding to the top view target edge contour and the side view target edge contour; S4. Constructing a mathematical model for calculating the volume of the target to be measured by a shell method based on the extracted morphological feature information, wherein the target to be measured is a dot-shaped rotating body.

2. The target volume detection method based on binocular vision according to claim 1, characterized in that: The method further comprises using the least square method to modify and optimize the mathematical model.

3. The target volume detection method based on binocular vision according to claim 1, characterized in that: The S2 step includes the following sub-steps: S21, preprocessing the binocular target image by using distortion correction and epipolar correction; S22, by calculating the 95th percentile ratio of the RGB channel of the binocular target image to the grayscale image, extracting the specular highlight area and generating a highlight mask area, wherein the specular highlight area is represented as , where , , Represents the red, green, and blue channel values ​​of the pixel respectively. Represents the grayscale value of the pixel, Indicates the preset baseline threshold; S23, calculating the average color of the pixels in the highlight mask area, filling all pixels in the connected area with the average color, and obtaining the gray binocular target image.

4. The target volume detection method based on binocular vision according to claim 1, characterized in that: The S3 step includes: S31, converting the gray binocular target image into a grayscale image and performing binarization processing; S32. Obtaining the top view target edge contour and the side view target edge contour by using an image segmentation method, wherein the image segmentation method includes one or more of morphological operations, multi-scale filtering, and contour detection; S33. Calculate the physical radius of the top view target edge contour by using a ring fitting and radius extraction method, and obtain an edge curve fitting function of the morphological feature information by using edge curve extraction based on the side view target edge contour.

5. The target volume detection method based on binocular vision according to claim 4, characterized in that: The S33 step includes: S331, randomly selecting a number of points in the edge contour of the target in the top view, and linearizing the equation using the least squares method to obtain a fitting outer ring; S332, calculating the disparity between the center of the fitted outer circle and a point on the circle, and calculating the depth information of the two points based on the intrinsic parameters of the binocular camera; S333: Calculate the three-dimensional coordinates of the target using the depth information, and calculate the physical radius of the edge contour of the target in the top view using a two-point distance formula.

6. The target volume detection method based on binocular vision according to claim 4, characterized in that: The edge curve fitting function of the side view target edge contour The calculation formula is as follows: Where, is represented as the fitting parameter, , , It is represented as the segmentation point of the OQ curve segment, Represents the endpoint of the second MQ curve.

7. The target volume detection method based on binocular vision according to claim 1, characterized in that: The S4 step includes: S41, based on the edge curve fitting function of the morphological feature information, using the shell method to calculate the pixel volume of the target ,Right now , where Expressed as pixel volume, Expressed as the horizontal coordinate value of the rotation axis in the pixel coordinate system, Expressed as edge curve fitting function in pixel coordinate system; S42, using the physical radius of the morphological feature information and the fitting curve The ratio of the lengths of the integration intervals on the axis is used to calculate the scaling factor ; S43, based on the scale conversion factor Perform proportional transformation on the parameters of the two-dimensional coordinates to obtain the physical volume of the target to be measured ,Right now .

8. A target volume detection system based on binocular vision, characterized in that: The system comprises: An image acquisition module is configured to acquire a binocular target image of a target point to be measured using a binocular camera, wherein the binocular target image includes a top view and a side view of the target; A gray binocular target image processing module is configured to obtain a gray binocular target image with highlight interference removed based on a specular highlight detection and removal algorithm after preprocessing the binocular target image; an image segmentation and extraction module configured to respectively obtain a top view target edge contour and a side view target edge contour of the gray binocular target image based on an image segmentation method, and extract morphological feature information corresponding to the top view target edge contour and the side view target edge contour; The model building module constructs a mathematical model for calculating the volume of the target to be measured by the shell method based on the extracted morphological feature information; wherein the target to be measured is a dot-shaped rotating body.

9. A colloid volume detection method based on binocular vision, characterized in that: The binocular vision-based target volume detection method according to any one of claims 1 to 7 and / or the binocular vision-based target volume detection system according to claim 8 is adopted, wherein the dot-shaped quasi-rotating body is a colloidal material.

10. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for detecting target volume based on binocular vision according to any one of claims 1 to 7 or the method for detecting colloid volume based on binocular vision according to claim 9 is implemented.

Citation Information

Cited By

  • Device and method for detecting roundness of cauliflower

    CN121048532A