Meat rabbit body weight and body size automatic measuring system based on machine vision

Abstract

The invention discloses a meat rabbit body weight and body size automatic measuring system based on machine vision, and relates to the technical field of breeding data analysis. The automatic measurement platform is in communication connection with an image capture module, a meat rabbit feature extraction module, a posture transformation module, a body size measurement module, a weight measurement module, a data calibration verification module and a user interaction module, and all the modules are in electric signal connection; and the image capturing module is used for capturing image data of the meat rabbit from different angles. According to the meat rabbit body weight and body size automatic measurement system based on machine vision, measurement of the body weight and body size of a meat rabbit, monitoring of the growth condition of the meat rabbit and capturing of the change trend of the body weight and body size are realized through an image processing technology, so that managers can find abnormal growth or potential health problems of the meat rabbit in time and take corresponding intervention measures, and the management efficiency is improved. Therefore, the feeding management strategy is optimized, and the growth performance and health condition of the meat rabbits are improved.

Description

Technical Field

[0001] The invention relates to the technical field of breeding data analysis, and in particular to an automatic measurement system for meat rabbit weight and size based on machine vision. Background Art

[0002] With the advancement of science and technology, animal husbandry is gradually developing towards intelligence and automation. Meat rabbit breeding occupies a certain position in animal husbandry. However, due to its high breeding density and complex management, the traditional manual method of measuring weight and body size has many inconveniences. Machine vision technology can realize accurate identification and measurement of individual animals through image acquisition, processing and analysis. In the process of measuring the weight and body size of meat rabbits, the use of machine vision technology can achieve rapid and accurate identification of meat rabbits.

[0003] In the prior art, due to individual differences between meat rabbits, different body shapes and postures of meat rabbits will lead to morphological differences, which in turn interferes with the measurement of the weight and body size of meat rabbits. Therefore, how to weaken the differences in body shape and posture of meat rabbits to improve the accuracy of machine vision recognition and reduce measurement errors is the problem we need to solve. To this end, a meat rabbit weight and body size automatic measurement system based on machine vision is proposed. Summary of the invention

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: a machine vision-based meat rabbit weight and body size automated measurement system, comprising an automated measurement platform, wherein the automated measurement platform is communicatively connected with an image capture module, a meat rabbit feature extraction module, a posture transformation module, a body size measurement module, a weight measurement module, a data calibration verification module, and a user interaction module, wherein the modules are electrically connected with each other;

[0005] The image capture module is used to capture image data of meat rabbits from different angles, and pre-process the image data, including denoising, contrast enhancement, and brightness adjustment, so as to improve image quality and generate a multi-view image set;

[0006] The meat rabbit feature extraction module uses edge detection and texture analysis technology to extract the body features of the meat rabbit from the image data of the multi-view image set, including body contours and joint positions, and matches feature point pairs to form a body feature sequence;

[0007] The posture transformation module transforms different postures of the meat rabbit based on the feature data of the body feature sequence, and adjusts the posture of the meat rabbit to a standard posture through image transformation and geometric correction technology to reduce the measurement error caused by posture differences and facilitate subsequent processing;

[0008] The body size measurement module performs 3D point cloud reconstruction based on the matched feature point pairs in the image data in combination with the 3D reconstruction technology to measure the body size data of the meat rabbit;

[0009] The weight measurement module measures the weight of the rabbit according to its body shape characteristics based on the measured body size data and the corrected posture of the rabbit and a weight prediction model constructed using machine learning technology;

[0010] The data calibration and verification module calculates the weight and body size matching coefficient by comparing with the set standard value, calibrates and verifies the measurement results, and analyzes the degree of deviation between the automated measurement data and the standard value;

[0011] The user interaction module includes a display screen and buttons, which are used to display the captured pictures, measurement results and user-set related human-computer interaction parameters.

[0012] Preferably, in the image capturing module, the process of capturing the image data of the meat rabbit includes:

[0013] Configure cameras with high resolution and wide-angle lenses at multiple angles, initialize image capture parameters of resolution, frame rate, and exposure, and preset capture angles and positions of meat rabbit images. The cameras have high resolution and wide-angle lenses to ensure that clear and complete meat rabbit images are captured;

[0014] The meat rabbit is placed in the capture area and ensured to be within the field of view of the camera. According to the preset capture angle and position, the configured camera is adjusted to capture multiple perspective image data of the meat rabbit, and the captured image covers the whole body of the meat rabbit, including the side, front and back;

[0015] The images captured by different cameras are synchronized in time to ensure the consistency of image data, and the captured image data is preprocessed, including denoising, contrast enhancement and brightness adjustment. Denoising is used to eliminate noise and interference in the image. Denoising methods include Gaussian filtering and median filtering. Contrast enhancement is used to increase the brightness difference between different areas in the image to make the image clearer. Contrast enhancement methods include linear transformation and histogram equalization. Brightness adjustment is used to improve the brightness level of the image to make it more suitable for observation or analysis needs.

[0016] The preprocessed image data are classified and sorted according to different capture angles, and the classified image data are integrated to generate a multi-view image set containing multiple viewpoints of meat rabbits.

[0017] Preferably, in the meat rabbit feature extraction module, the process of acquiring the body shape feature sequence includes:

[0018] Receive image data from a multi-view image set, including full-body images of the side, front, and back of the rabbit, and ensure that the image data is clear and complete without obvious blur, occlusion, or noise interference;

[0019] Apply the edge detection technology of Sobel operator to the image data, calculate the gradient amplitude of the image to highlight the edge, and identify the edge features of the body contour and joint position of the meat rabbit in the image through the result of Sobel operator. Among them, the edge features are manifested as high gradient value areas in the image.

[0020] The texture features in the image are analyzed using the texture analysis technology of the gray-level co-occurrence matrix. The texture feature parameters of contrast, energy, entropy, and correlation are extracted from the gray-level co-occurrence matrix. The skin texture and hair texture features of the meat rabbit are analyzed to reflect the fineness, uniformity, and complexity of the skin texture and hair texture of the meat rabbit.

[0021] Match the feature points extracted from the multi-view images, generate feature descriptors for each feature point, and use the SIFT feature point matching algorithm to match the feature point pairs in the images of different view angles to determine the feature point pairs of the body contour and joint positions of the meat rabbit, where the feature point pairs are the positions with obvious features such as corner points and edge intersections in the image;

[0022] The matched feature point pairs are connected in series to form a body feature sequence, which includes the contour line of the body outline (composed of a series of matched edge feature points) and the key points of the joint positions (composed of matched joint position feature points). The feature sequence reflects the body features and joint position information of the meat rabbit at different viewing angles.

[0023] Preferably, in the posture transformation module, the process of adjusting the posture of the rabbit includes:

[0024] Receiving the body feature sequence output from the meat rabbit body feature extraction module, presetting the standard posture of the meat rabbit, and applying computer vision technology to estimate the posture and position of the meat rabbit using key point information in the body feature sequence, wherein the standard posture of the meat rabbit is a four-limb standing posture;

[0025] Applying image transformation techniques such as affine transformation and perspective transformation to the original image, making preliminary posture adjustment to the image of the meat rabbit, and preliminarily adjusting the posture of the meat rabbit to a state close to the standard posture;

[0026] Based on the image transformation, the geometric correction technology is used to calculate the perspective transformation matrix, and it is applied to the image to further adjust the posture of the rabbit and correct the geometric distortion caused by the difference in viewing angle to obtain the rabbit image in a standard posture.

[0027] Perform posture transformation, use the calculated transformation matrix to adjust the posture of the meat rabbit to the preset standard posture, ensure the accuracy of the body proportions and feature point positions of the meat rabbit during the transformation process, use feature point matching technology to check whether the transformed feature points match the feature points of the standard posture, and verify whether the transformed posture meets the requirements of the standard posture. If not, the posture needs to be re-estimated, and the meat rabbit image adjusted to the standard posture is output for use by the subsequent weight measurement module and body size measurement module.

[0028] Preferably, in the body size measurement module, the process of measuring the body size data of the rabbit includes:

[0029] Extract rabbit images captured from different perspectives, ensure that the images contain enough feature information for matching, apply SIFT algorithm to each image, extract feature points (i.e. key points) and their descriptors in the image, and use the distance metric (Euclidean distance) between descriptors to match feature point pairs in different images;

[0030] The camera's intrinsic parameters (focal length, optical center) and extrinsic parameters (rotation matrix, translation vector) are calculated through the known 3D world coordinate points and the corresponding image coordinate points, and the camera's pose, that is, the camera's position and posture in space, is estimated through the PnP optimization algorithm using the matched feature point pairs and the known camera intrinsic parameters;

[0031] According to the matched feature point pairs and the camera's position and posture, the point coordinates in the three-dimensional space are calculated using the triangulation principle to reconstruct a sparse three-dimensional point cloud. The reconstructed three-dimensional point cloud is filtered to remove noise and redundant points caused by matching errors and illumination changes. The point cloud is segmented into different parts according to its geometric features, and the point cloud part representing the meat rabbit is extracted from the segmented point cloud.

[0032] In the extracted meat rabbit point cloud data, the geometric features of body length, body height and body width are identified. Body length refers to the straight-line distance from the head to the tail, body height refers to the distance from the ground to the highest point, and body width refers to the lateral distance at the widest point. Based on the identified geometric features, the corresponding dimensions are measured in the point cloud data and the body size data of the meat rabbit is calculated.

[0033] Preferably, the calculation formula of the body length is:

[0034]

[0035]

[0036] Where L is the body length, N is the number of feature points in the point cloud used to calculate the body length, and p i is the coordinate of the i-th feature point in the point cloud, q i For iThe corresponding points represent the corresponding points from the head to the tail, d(p i ,q i ) is point p i and q i The Euclidean distance between i The coordinates of (x ip ,y ip , z ip ), q i The coordinates of (x iq ,y iq , z iq );

[0037] The calculation formula of the body height is:

[0038] H=max(z i )-min(z i );

[0039] Where H is the height, z i is the z coordinate value of all points in the point cloud, that is, the height vertical to the ground, max(z i ) is the maximum value of the z coordinate of all points in the point cloud, min(z i ) is the minimum value of the z coordinate of all points in the point cloud;

[0040] The calculation formula of the body width is:

[0041]

[0042] Where W is the body width, M is the number of feature points in the point cloud used to calculate the body width, and w j is the vertical distance from the jth feature point in the point cloud to the body length line.

[0043] Preferably, in the weight measurement module, the process of measuring the weight of the rabbit includes:

[0044] Receive the body size data of the meat rabbit output from the body size measurement module and the posture image processed by the posture correction module, and extract body shape features related to weight from the corrected posture image, including body size, proportion, and shape, wherein the body size includes body length, body width, and body height, which are obtained by measuring the outline of the meat rabbit in the image or the distance between key points, the proportion includes the head-to-body ratio, the proportion of the limbs to the body, which is obtained by calculating the ratio of related sizes, and the shape includes the outline shape of the body and the bending degree of the limbs, which are obtained by edge detection and contour extraction image processing technology;

[0045] Pre-train the weight prediction model and deploy it in the production environment. Input the extracted body shape features into the pre-trained weight prediction model and output the predicted weight range.

[0046] According to the predicted weight range and the body characteristics of meat rabbits, the weight measurement index is calculated, and the weight measurement standard of meat rabbits is preset and compared with the calculated weight measurement index to evaluate the growth and development status and health level of meat rabbits and obtain the weight measurement evaluation result.

[0047] Preferably, the calculation formula of the body weight index is:

[0048]

[0049] Where I is the body mass index, W is the body width, and W base W is the reference body width of meat rabbits based on age and body size. std is the standard deviation of body width, used to adjust the sensitivity of the exponential function, L is body length, L base is the standard body length of meat rabbits based on age and body shape, H is the height, and H base The base height of meat rabbits based on age and size.

[0050] Preferably, in the data calibration verification module, the analysis process of automatically determining the degree of deviation between the data and the standard value includes:

[0051] According to the breeding standards, feeding management requirements and industry standards of meat rabbits, standard values ​​of weight and body size (body length, body height and body width) are set, and the weight and body size data output by the automated measurement system are collected to ensure a unified data format, including date, time, individual identification, weight, body length, body height and body width fields, to facilitate subsequent processing and analysis;

[0052] Compare the automated measurement data with the set standard value, check the difference between each measurement value (weight, length, height, width) and the standard value one by one, and calculate the deviation of each measurement value, that is, the difference between the measurement value and the standard value;

[0053] According to the deviation, calculate the weight and size matching coefficient, analyze the closeness of the automated measurement data to the standard value, analyze the deviation of the automated measurement data from the standard value, and determine whether the data is within the acceptable error range. If the deviation is large, further investigation is required, including measurement equipment failure, data input error or feeding management problems;

[0054] Based on the analysis results of the degree of deviation and the weight and body size matching coefficient, the automated measurement system is calibrated and adjusted, and the calibration and verification results are output to generate a measurement report, which includes detailed data comparison, weight and body size matching coefficient, and calibrated data. The calibration verification results and measurement report are then fed back to relevant personnel.

[0055] Preferably, the calculation formula of the weight and body size matching coefficient is:

[0056]

[0057] Among them, P is the weight and size matching coefficient, X k is the kth data point (weight, length, height, width) measured automatically, Y k is the kth standard value corresponding to the automated determination, n is the number of determination values, |X k -Y k | is the absolute difference between the kth measured value and its standard value, and the value of P ranges from 0 to 1.

[0058] The present invention provides a machine vision-based automatic measurement system for rabbit weight and size. It has the following beneficial effects:

[0059] 1. This machine vision-based automated measurement system for rabbit weight and size uses image processing technology to measure the weight and size of rabbits, monitor their growth, and capture changing trends in weight and size, so that managers can promptly detect abnormal growth or potential health problems in rabbits and take appropriate intervention measures, thereby optimizing feeding and management strategies and improving the growth performance and health of rabbits.

[0060] 2. This machine vision-based automated measurement system for meat rabbit weight and size avoids direct contact and interference with meat rabbits through an automated non-contact measurement method, reduces the stress response of meat rabbits, and calculates the weight and size matching coefficient by comparing the automated measurement data with the standard value, and analyzes the degree of proximity between the automated measurement data and the standard value, providing a scientific basis for the breeding and feeding management of meat rabbits. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a module diagram of a machine vision-based automatic measurement system for meat rabbit weight and size of the present invention;

[0062] Figure 2 It is a flow chart of measuring the body size data of meat rabbits of the present invention;

[0063] Figure 3 The present invention is a flow chart for measuring the weight of meat rabbits. DETAILED DESCRIPTION

[0064] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0065] The first embodiment, as Figure 1 , Figure 2 As shown, the present invention provides a technical solution: a machine vision-based meat rabbit weight and body size automatic measurement system, including an automatic measurement platform, the automatic measurement platform is communicatively connected with an image capture module, a meat rabbit feature extraction module, a posture transformation module, a body size measurement module, a weight measurement module, a data calibration verification module and a user interaction module, wherein the modules are connected by electrical signals;

[0066] The image capture module is used to capture the image data of the meat rabbit from different angles, and pre-process the image data, including denoising, contrast enhancement, and brightness adjustment to improve image quality, generate a multi-view image set, configure a camera with a high-resolution and wide-angle lens at multiple angles, initialize the image capture parameters of resolution, frame rate, and exposure, and preset the capture angle and position of the meat rabbit image. The camera has a high-resolution and wide-angle lens to ensure that a clear and complete meat rabbit image is captured. The meat rabbit is within the capture area and is ensured to be within the field of view of the camera. According to the preset capture angle and position, the configured camera is adjusted to capture the image data of the meat rabbit from multiple perspectives. The captured image covers the whole body of the meat rabbit, including the side, front, and back. , time synchronization is performed on the images captured by different cameras to ensure the consistency of image data, and the captured image data is preprocessed, including denoising, contrast enhancement and brightness adjustment, wherein denoising is used to eliminate noise and interference in the image, denoising methods include Gaussian filtering and median filtering, contrast enhancement is used to increase the brightness difference between different areas in the image to make the image clearer, contrast enhancement methods include linear transformation and histogram equalization, brightness adjustment is used to improve the brightness level of the image to make it more suitable for observation or analysis needs, the preprocessed image data is classified and sorted according to different capture angles, and the classified image data is integrated to generate a multi-view image set containing multiple viewpoints of meat rabbits;

[0067] The meat rabbit feature extraction module uses edge detection and texture analysis technology to extract the body features of the meat rabbit from the image data of the multi-view image set, including the body contour and joint position, and matches the feature point pairs to form a body feature sequence. It receives the image data in the multi-view image set, including the full-body images of the meat rabbit from the side, front and back, to ensure that the image data is clear and complete without obvious blur, occlusion or noise interference. The edge detection technology of the Sobel operator is applied to the image data, and the gradient amplitude of the image is calculated to highlight the edge. The edge features of the body contour and joint position of the meat rabbit in the image are identified through the results of the Sobel operator. Among them, the edge features are manifested as high gradient value areas in the image. The texture analysis technology of the grayscale co-occurrence matrix is ​​used to analyze the texture features in the image, and the contrast, energy, and entropy are extracted from the grayscale co-occurrence matrix. , and correlated texture feature parameters, analyze the skin texture and hair texture features of the meat rabbit to reflect the fineness, uniformity, and complexity of the skin texture and hair texture of the meat rabbit, match the feature points extracted from the multi-view images, generate a feature descriptor for each feature point, and use the SIFT feature point matching algorithm to match the feature point pairs in the images of different view angles to determine the feature point pairs of the body contour and joint positions of the meat rabbit, where the feature point pairs are positions with obvious features such as corner points and edge intersections in the image, connect the matched feature point pairs in series to form a body feature sequence, which includes the contour line of the body contour (composed of a series of matched edge feature points) and the key points of the joint positions (composed of matched joint position feature points), and the feature sequence reflects the body features and joint position information of the meat rabbit at different view angles;

[0068] The posture transformation module transforms different postures of the meat rabbit based on the feature data of the body feature sequence, and adjusts the posture of the meat rabbit to a standard posture through image transformation and geometric correction technology to reduce the measurement error caused by posture differences and facilitate subsequent processing. The body feature sequence output from the meat rabbit body feature extraction module is received, the standard posture of the meat rabbit is preset, and computer vision technology is applied to estimate the posture and position of the meat rabbit using the key point information in the body feature sequence. Among them, the standard posture of the meat rabbit is a standing posture with four limbs. The image transformation technology of affine transformation and perspective transformation is applied to the original image to make a preliminary posture adjustment to the image of the meat rabbit, and the posture of the meat rabbit is preliminarily adjusted to a state close to the standard posture. On the basis of the transformation, the perspective transformation matrix is ​​calculated by applying geometric correction technology and applied to the image to further adjust the posture of the meat rabbit and correct the geometric distortion caused by the difference in viewing angle to obtain the meat rabbit image in the standard posture, perform posture transformation, and use the calculated transformation matrix to adjust the posture of the meat rabbit to the preset standard posture to ensure that the accuracy of the body proportions and feature point positions of the meat rabbit are maintained during the transformation process. The feature point matching technology is used to check whether the transformed feature points match the feature points of the standard posture and verify whether the transformed posture meets the requirements of the standard posture. If not, the posture needs to be re-estimated and the meat rabbit image adjusted to the standard posture is output for use by the subsequent weight measurement module and body size measurement module;

[0069] The body size measurement module reconstructs the 3D point cloud based on the matched feature point pairs in the image data in combination with the 3D reconstruction technology, measures the body size data of the meat rabbit, extracts the meat rabbit images captured from different perspectives, ensures that the images contain sufficient feature information for matching, applies the SIFT algorithm to each image, extracts the feature points (i.e., key points) in the image and their descriptors, uses the distance metric (Euclidean distance) between the descriptors to match the feature point pairs in different images, calculates the camera's intrinsic parameters (focal length, optical center) and extrinsic parameters (rotation matrix, translation vector) through the known 3D world coordinate points and the corresponding image coordinate points, and uses the matched feature point pairs and the known camera intrinsic parameters to estimate the camera's pose, i.e., the camera's position in space, through the PnP optimization algorithm. and posture, according to the matched feature point pairs and the camera's posture, the point coordinates in the three-dimensional space are calculated using the triangulation principle, so as to reconstruct a sparse three-dimensional point cloud, and the reconstructed three-dimensional point cloud is filtered to remove noise and redundant points caused by matching errors and illumination changes, and the point cloud is segmented into different parts according to the geometric features of the point cloud, and the point cloud part representing the meat rabbit is extracted from the segmented point cloud. In the extracted meat rabbit point cloud data, the geometric features of body length, body height, and body width are identified, where body length refers to the straight-line distance from the head to the tail, body height refers to the distance from the ground to the highest point, and body width refers to the lateral distance at the widest point. According to the identified geometric features, the corresponding dimensions are measured in the point cloud data, and the body size data of the meat rabbit is calculated;

[0070] Furthermore, the calculation formula of body length is:

[0071]

[0072] Where L is the body length, N is the number of feature points in the point cloud used to calculate the body length, and p i is the coordinate of the i-th feature point in the point cloud, q i For i The corresponding points represent the corresponding points from the head to the tail, d(p i ,q i ) is point p i and q i The Euclidean distance between i The coordinates of (x ip ,y ip , z ip ), q i The coordinates of (x iq ,y iq , z iq );

[0073] The calculation formula for body height is:

[0074] H=max(z i )-min(z i );

[0075] Where H is the height, z i is the z coordinate value of all points in the point cloud, that is, the height vertical to the ground, max(z i ) is the maximum value of the z coordinate of all points in the point cloud, min(z i ) is the minimum value of the z coordinate of all points in the point cloud;

[0076] The calculation formula for body width is:

[0077]

[0078] Where W is the body width, M is the number of feature points in the point cloud used to calculate the body width, and w j is the vertical distance from the jth feature point in the point cloud to the body length line. As the meat rabbit grows, the values ​​of L, H, and W usually increase, reflecting the growth and development of the meat rabbit;

[0079] The weight measurement module measures the weight of the rabbit based on its body shape characteristics, based on the measured body size data and corrected posture, combined with a weight prediction model built using machine learning technology;

[0080] The data calibration and verification module calculates the weight and body size matching coefficient by comparing with the set standard value, calibrates and verifies the measurement results, and analyzes the degree of deviation between the automated measurement data and the standard value;

[0081] The user interaction module includes a display screen and buttons, which are used to display the captured pictures, measurement results and user-set related human-computer interaction parameters.

[0082] The second embodiment is based on the first embodiment. Figure 3 As shown, in the weight measurement module, the process of measuring the weight of the meat rabbit includes:

[0083] Receive the body size data of the meat rabbit output from the body size measurement module and the posture image processed by the posture correction module, and extract the body shape features related to weight from the corrected posture image, including body size, proportion, and shape, wherein the body size includes body length, body width, and body height, which are obtained by measuring the outline of the meat rabbit in the image or the distance between key points, the proportion includes the head-to-body ratio, the proportion of the limbs to the body, which is obtained by calculating the ratio of related sizes, and the shape includes the outline shape of the body and the bending degree of the limbs, which are obtained by edge detection and contour extraction image processing technology, pre-train the weight prediction model, and deploy the weight prediction model in the production environment, and input the extracted body shape features into the pre-trained In the weight prediction model, the predicted weight range is output. A large amount of body feature data and corresponding weight data from experimental measurements, medical records, and health monitoring are obtained, covering meat rabbits of different body shapes, ages, and gender factors. The data is cleaned, duplicates, missing or outliers are removed, and normalized or standardized. The data is divided into a training set and a validation set. The training set data is used in combination with the neural network model to train the weight prediction model. The parameters of the model are adjusted through the optimization algorithm. During the training process, the loss function value and accuracy index of the model are monitored to evaluate the training effect of the model. The trained model is evaluated using the validation set data to calculate the prediction error, mean square error, root mean square error, R 2 Score index: Based on the evaluation results, determine whether the prediction performance of the model meets the requirements. Based on the evaluation results, optimize the model, calculate the weight measurement index based on the predicted weight range and the body characteristics of the meat rabbits, and preset the weight measurement standard of the meat rabbits. Compare it with the calculated weight measurement index, evaluate the growth and development status and health level of the meat rabbits, and obtain the weight measurement evaluation results;

[0084] Furthermore, the calculation formula of the body mass index is:

[0085]

[0086] Where I is the body mass index, W is the body width, and W baseW is the reference body width of meat rabbits based on age and body size. std is the standard deviation of body width, used to adjust the sensitivity of the exponential function, L is body length, L base is the standard body length of meat rabbits based on age and body shape, H is the height, and H base is the benchmark height of the meat rabbit based on its age and body shape. When the actual body width, length and height of the meat rabbit are close to or equal to the benchmark value, I is close to 1, indicating that the weight of the meat rabbit is within the healthy range. When the actual body width of the meat rabbit is lower than the benchmark body width, and the body length or height is lower than the benchmark value, the I value will decrease, indicating that the meat rabbit may be thin.

[0087] In the data calibration and verification module, the analysis process of automatically determining the degree of deviation between the data and the standard value includes:

[0088] According to the breeding standards, feeding management requirements and industry standards of meat rabbits, standard values ​​of weight and body size (body length, body height and body width) are set, and the weight and body size data output by the automated measurement system are collected to ensure a unified data format, including date, time, individual identification, weight, body length, body height and body width fields, to facilitate subsequent processing and analysis. The automated measurement data are compared with the set standard values, and the difference between each measurement value (weight, body length, body height and body width) and the standard value is checked one by one. The deviation of each measurement value, that is, the difference between the measurement value and the standard value, is calculated. According to the size of the deviation, the weight and body size matching system is calculated. Analyze the degree of proximity between the automated measurement data and the standard value, analyze the degree of deviation between the automated measurement data and the standard value, and determine whether the data is within the acceptable error range. If the deviation is large, further investigation is required, including measurement equipment failure, data input error, or feeding management problems. According to the analysis results of the deviation degree and weight and body size matching coefficient, calibrate and adjust the automated measurement system, output the calibration and verification results, and generate a measurement report. The report contains detailed data comparison, weight and body size matching coefficient, and calibrated data, and then feedback the calibration verification results and measurement report to relevant personnel;

[0089] Furthermore, the calculation formula of the weight and body size matching coefficient is:

[0090]

[0091] Among them, P is the weight and size matching coefficient, X k is the kth data point (weight, length, height, width) measured automatically, Y k is the kth standard value corresponding to the automated determination, n is the number of determination values, |X k -Y k| is the absolute difference between the kth measured value and its standard value. The value of P ranges from 0 to 1. The closer the value is to 1, the closer the automated measurement data is to the standard value and the higher the matching degree is. The closer the value of P is to 0, the greater the deviation between the automated measurement data and the standard value is and the lower the matching degree is.

[0092] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A machine vision-based automated measurement system for rabbit weight and size, comprising an automated measurement platform, characterized in that: The automated measurement platform is communicatively connected with an image capture module, a rabbit feature extraction module, a posture transformation module, a body measurement module, a weight measurement module, a data calibration verification module, and a user interaction module, wherein electrical signals are connected between the modules; The image capture module is used to capture image data of the meat rabbit from different angles and pre-process the image data; The meat rabbit feature extraction module uses edge detection and texture analysis technology to extract the body features of the meat rabbit from the image data of the multi-view image set, and matches the feature point pairs to form a body feature sequence; The posture transformation module transforms different postures of the meat rabbit based on the feature data of the body feature sequence, and adjusts the posture of the meat rabbit to a standard posture; The body size measurement module performs 3D point cloud reconstruction based on the matched feature point pairs in the image data in combination with the 3D reconstruction technology to measure the body size data of the meat rabbit; The weight measurement module measures the weight of the rabbit according to its body shape characteristics based on the measured body size data and the corrected posture of the rabbit and a weight prediction model constructed using machine learning technology; The data calibration and verification module calculates the weight and body size matching coefficient by comparing with the set standard value, and analyzes the degree of deviation between the automated measurement data and the standard value; The user interaction module includes a display screen and buttons, which are used to display the captured pictures, measurement results and user-set related human-computer interaction parameters.

2. The machine vision-based automated measurement system for rabbit weight and size according to claim 1, characterized in that: In the image capture module, the process of capturing the image data of the meat rabbit includes: Configure cameras with high-resolution and wide-angle lenses at multiple angles, initialize image capture parameters such as resolution, frame rate, and exposure, and preset capture angles and positions of meat rabbit images; The meat rabbit is placed in a capture area, and the configured camera is adjusted to capture multiple perspective image data of the meat rabbit according to a preset capture angle and position, and the captured image covers the whole body of the meat rabbit, including the side, front and back; Time synchronization of images captured by different cameras and preprocessing of captured image data, including denoising, contrast enhancement and brightness adjustment; The preprocessed image data are classified and sorted according to different capture angles, and the classified image data are integrated to generate a multi-view image set containing multiple viewpoints of meat rabbits.

3. The machine vision-based automated measurement system for rabbit weight and size according to claim 2, characterized in that: In the meat rabbit feature extraction module, the process of acquiring the body feature sequence includes: Receiving image data from a multi-view image set, including full-body images of the side, front, and back of a meat rabbit; Apply the edge detection technology of Sobel operator to the image data, calculate the gradient amplitude of the image to highlight the edge, and identify the edge features of the body contour and joint position of the meat rabbit in the image through the result of Sobel operator; The texture features in the image are analyzed using the texture analysis technology of gray-level co-occurrence matrix, and the texture feature parameters of contrast, energy, entropy, and correlation are extracted from the gray-level co-occurrence matrix to analyze the skin texture and hair texture features of meat rabbits. Match the feature points extracted from the multi-view images, generate feature descriptors for each feature point, and use the SIFT feature point matching algorithm to match the feature point pairs in the images of different view angles to determine the feature point pairs of the body contour and joint positions of the meat rabbit; The matched feature point pairs are connected in series to form a body feature sequence, which includes the contour line of the body contour and the key points of the joint positions.

4. The machine vision-based automated measurement system for rabbit weight and size according to claim 3, characterized in that: In the posture transformation module, the process of adjusting the posture of the rabbit includes: Receiving the body feature sequence output from the meat rabbit body feature extraction module, presetting the standard posture of the meat rabbit, and applying computer vision technology to estimate the posture and position of the meat rabbit using key point information in the body feature sequence, wherein the standard posture of the meat rabbit is a four-limb standing posture; Applying image transformation techniques such as affine transformation and perspective transformation to the original image, making preliminary posture adjustment to the image of the meat rabbit, and preliminarily adjusting the posture of the meat rabbit to a state close to the standard posture; Based on the image transformation, the geometric correction technology is used to calculate the perspective transformation matrix, and it is applied to the image to further adjust the posture of the rabbit and correct the geometric distortion caused by the difference in viewing angle to obtain the rabbit image in a standard posture. Perform posture transformation, use the calculated transformation matrix to adjust the posture of the meat rabbit to the preset standard posture, use feature point matching technology to check whether the transformed feature points match the feature points of the standard posture, and verify whether the transformed posture meets the requirements of the standard posture. If not, the posture needs to be re-estimated and the meat rabbit image adjusted to the standard posture is output.

5. The machine vision-based automated measurement system for rabbit weight and size according to claim 4, characterized in that: In the body size measurement module, the measurement process of the rabbit body size data includes: Extract rabbit images captured from different perspectives, apply SIFT algorithm to each image, extract feature points and their descriptors in the image, and use the distance metric between descriptors to match feature point pairs in different images; The camera's intrinsic and extrinsic parameters are calculated through the known 3D world coordinate points and the corresponding image coordinate points, and the camera's pose, that is, the camera's position and posture in space, is estimated through the PnP optimization algorithm using the matched feature point pairs and the known camera intrinsic parameters. According to the matched feature point pairs and the camera's position and posture, the point coordinates in the three-dimensional space are calculated using the triangulation principle to reconstruct a sparse three-dimensional point cloud. The reconstructed three-dimensional point cloud is filtered and segmented into different parts according to the geometric features of the point cloud. The point cloud part representing the meat rabbit is extracted from the segmented point cloud. In the extracted meat rabbit point cloud data, the geometric features of body length, body height and body width are identified. Body length refers to the straight-line distance from the head to the tail, body height refers to the distance from the ground to the highest point, and body width refers to the lateral distance at the widest point. Based on the identified geometric features, the corresponding dimensions are measured in the point cloud data and the body size data of the meat rabbit is calculated.

6. The machine vision-based automated measurement system for rabbit weight and size according to claim 5, characterized in that: The calculation formula of the body length is: Where L is the body length, N is the number of feature points in the point cloud used to calculate the body length, and p i is the coordinate of the i-th feature point in the point cloud, q i For i The corresponding points represent the corresponding points from the head to the tail, d ( p i ,q i) For point p i and q i The Euclidean distance between i The coordinates of ( x ip ,y ip , z ip ) ,q i The coordinates of ( x iq ,y iq , z iq ) ; The calculation formula of the body height is: H=max ( With i ) -min ( With i ) ; Where H is the height, z i is the z coordinate value of all points in the point cloud, that is, the height vertical to the ground, max ( z i) is the maximum value of the z coordinates of all points in the point cloud, min ( z i) is the minimum value of the z coordinate of all points in the point cloud; The calculation formula of the body width is: Where W is the body width, M is the number of feature points in the point cloud used to calculate the body width, and w j is the vertical distance from the jth feature point in the point cloud to the body length line.

7. The machine vision-based automated measurement system for rabbit weight and size according to claim 6, characterized in that: In the weight measurement module, the process of measuring the weight of the rabbit includes: Receiving the body size data of the meat rabbit output from the body size measurement module and the posture image processed by the posture correction module, and extracting body shape features related to weight from the corrected posture image, including body size, proportion, and shape, wherein the body size includes body length, body width, and body height; Pre-train the weight prediction model and deploy it in the production environment. Input the extracted body shape features into the pre-trained weight prediction model and output the predicted weight range. According to the predicted weight range and the body characteristics of meat rabbits, the weight measurement index is calculated, and the weight measurement standard of meat rabbits is preset and compared with the calculated weight measurement index to evaluate the growth and development status and health level of meat rabbits and obtain the weight measurement evaluation result.

8. The machine vision-based automated measurement system for rabbit weight and size according to claim 7, characterized in that: The calculation formula of the body mass index is: Where I is the body mass index, W is the body width, and W base W is the reference body width of meat rabbits based on age and body size. std is the standard deviation of body width, L is body length, L base is the standard body length of meat rabbits based on age and body shape, H is the height, and H base The base height of meat rabbits based on age and size.

9. The machine vision-based automated measurement system for rabbit weight and size according to claim 8, characterized in that: In the data calibration verification module, the analysis process of automatically determining the degree of deviation between the data and the standard value includes: According to the breeding standards, feeding management requirements and industry standards of meat rabbits, standard values ​​for weight and body size are set, and weight and body size data output by the automated measurement system are collected, including date, time, individual identification, weight, body length, body height and body width fields; Compare the automated measurement data with the set standard value, check the difference between each measured value and the standard value one by one, and calculate the deviation of each measured value, that is, the difference between the measured value and the standard value; According to the deviation, calculate the weight and body size matching coefficient, analyze the closeness of the automated measurement data to the standard value, analyze the deviation of the automated measurement data from the standard value, and determine whether the data is within the acceptable error range; Based on the analysis results of the deviation degree and weight and body size matching coefficient, the automated measurement system is calibrated and adjusted, and the calibration and verification results are output, and a measurement report is generated. The calibration verification results and measurement report are then fed back to relevant personnel.

10. The machine vision-based automated measurement system for rabbit weight and size according to claim 9, characterized in that: The calculation formula of the weight and body size matching coefficient is: Among them, P is the weight and size matching coefficient, X k is the kth data point measured automatically, Y k is the kth standard value corresponding to the automated determination, n is the number of determination values, | X k -Y k| is the absolute difference between the kth measured value and its standard value.

Images