Animal gait scoring system and method

Through the combination of the depth image acquisition module and the gait scoring module, the volume value and limp ratio value of the cow's back are calculated, which solves the problems of the error and cost of limp detection in the existing technology, and realizes an accurate and economical limp scoring system.

CN114821637BActive Publication Date: 2024-11-15YINCHUAN AOTOSO INFORMATION TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202210361817.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-07
Publication Date
2024-11-15
Estimated Expiration
2042-04-07

AI Technical Summary

Technical Problem

There are errors in existing cow limp detection methods, especially when the ground is slippery or stressed, the accuracy of machine vision, force sensors and acceleration sensors is affected, and the costly force sensor methods are time-consuming and limited in application. .

Method used

The depth image acquisition module is used to obtain the depth image from the back of the cow. By extracting the pixel information and three-dimensional coordinate points of the key and reference parts, a reference bottom surface is constructed, the volume value is calculated and the limp ratio value is calculated, and the standard proportion value is matched to determine the limp score.

Benefits of technology

It realizes accurate judgment of limp when cows do not need to walk, reduces detection errors, and is simple and cost-effective, which is suitable for practical applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114821637B_ABST
    Figure CN114821637B_ABST
Patent Text Reader

Abstract

An animal gait scoring system includes a depth image acquisition module and a gait scoring module; the depth image acquisition module is used to acquire a back depth image reflecting the back contour of the animal from above the animal to be tested, and provide the back depth image to the gait scoring module; the gait scoring module is used to extract image information reflecting the key parts of the animal and image information of the reference parts from the acquired back depth image; the left and right actual volumes in the back depth image of the animal are analyzed according to the height reference feature point information, the plane feature point information and the preset height value to score the animal gait, and the lameness degree of the lame animal is determined by the animal gait score value and the reference value determined through actual measurement. The present invention also provides an animal gait scoring method.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of animal health monitoring, and in particular to an animal gait scoring system and method. Background Art

[0002] Lameness has become the second largest disease affecting dairy production. It not only affects the normal growth of dairy cows, but also leads to a significant decline in dairy cow production performance and an increase in the elimination rate, causing considerable economic losses to the dairy farming industry and also affecting the efficiency of breeding ranches. At present, dairy cow lameness is mainly identified and studied through machine vision technology, force sensor technology and acceleration sensor technology.

[0003] The machine vision-based method mainly uses image processing technology to analyze the gait or behavior of dairy cows to identify lameness of dairy cows. The existing technology has the following technical problems: it is based on the natural walking of dairy cows, but during the walking process, dairy cows often slip due to slippery ground, stress, etc., which causes the cows to slip and fall, resulting in abnormal feature extraction, affecting the normal detection of lameness of dairy cows by computer vision, and causing detection errors.

[0004] The force sensor method measures the reaction force of the cow's hoof on the ground through a one-dimensional or three-dimensional pressure plate, extracts relevant dynamic and kinematic variables, and identifies cow lameness. The existing technology has technical problems such as being expensive, time-consuming, and only being able to measure a few gait cycles, thus limiting its practical application.

[0005] The acceleration sensor measurement method has the characteristics of low cost, low power consumption, and large data volume. The technical problem of this existing technology is that when cows walk, they often slip due to slippery ground, stress, etc., which makes the acceleration sensor obtain inaccurate data and eventually causes detection errors. Summary of the invention

[0006] In view of this, it is necessary to provide a simple animal gait scoring system that can determine animal lameness without relying on the animal walking.

[0007] There is also a need for a method of scoring animal gait that is simple and does not rely on the animal walking to determine lameness.

[0008] An animal gait scoring system comprises a depth image acquisition module and a gait scoring module; the depth image acquisition module is used to acquire a back depth image reflecting the back contour of the animal from above the animal to be tested, and provide the back depth image to the gait scoring module;

[0009] The gait scoring module is used to extract image information reflecting the key parts of the animal and image information of the reference parts from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and the reference parts, and three-dimensional coordinate points corresponding to each pixel information; construct a reference bottom surface according to the height reference feature point information, the plane feature point information and the preset height value; the target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area, and the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, and the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal. left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right right The size of the volume is calculated with the smaller volume value as the numerator and the larger volume value as the denominator to obtain the corresponding actual lameness ratio value, and the actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value.

[0010] A method for scoring animal gait comprises the following steps:

[0011] Using a depth camera to obtain a back depth image reflecting the back contour of the animal from above the animal to be tested;

[0012] Extracting image information of key parts and reference parts of the animal from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to each pixel information;

[0013] According to the extracted image information of the key parts, a predetermined number of plane feature point information for constructing the plane is obtained according to the first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information;

[0014] According to the extracted image information of the reference part, a height reference feature point information is obtained according to the second extraction rule, the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information;

[0015] Constructing a reference bottom surface according to the height reference feature point information, the plane feature point information and the preset height value;

[0016] The target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area;

[0017] According to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated; the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal left ;

[0018] According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ;

[0019] Compare the actual volume value V on the left left 、The actual volume value V on the right right The smaller volume value is used as the numerator and the larger volume value is used as the denominator to calculate to obtain the corresponding actual lameness ratio value;

[0020] The actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value.

[0021] In the above-mentioned animal gait scoring system and method, a back depth image reflecting the back contour of the animal is obtained from above the animal to be tested, image information reflecting the key parts of the animal and image information of the reference parts are extracted from the back depth image, a predetermined number of plane feature point information for constructing a plane is obtained according to the extracted image information of the key parts, a height reference feature point information is obtained according to the extracted image information of the reference part, and a reference bottom surface is constructed according to the height reference feature point information, the plane feature point information and the preset height value; the target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area; the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface; the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right right The size of the volume is calculated with the smaller volume value as the numerator and the larger volume value as the denominator to obtain the corresponding actual lameness ratio value; the actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value, and then determine whether the animal is lame through the gait score of the animal. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Schematic diagram of the functional modules of a cattle gait scoring system according to a preferred embodiment.

[0023] Figure 2 for Figure 1 Schematic diagram of the functional modules of the gait scoring module.

[0024] Figure 3 The diagram is a schematic diagram for determining points 1 to 7 of the depth image of the cow's back.

[0025] Figure 4 This is a volumetric diagram on the left side of the key parts of a dairy cow.

[0026] Figure 5 This is a volume diagram of the right side of the key parts of a dairy cow.

[0027] Figure 6 A flowchart of a method for scoring animal gait according to a preferred embodiment.

[0028] In the figure: an animal gait scoring system 10, a depth image acquisition module 20, a gait scoring module 30, an image information extraction unit 301, a first extraction unit 302, a second extraction unit 303, a reference bottom surface construction unit 304, a target image area bisection unit 305, a left actual volume calculation unit 306, a right actual volume calculation unit 307, an actual lameness ratio value calculation unit 308, a lameness score value search unit 309, a search information table generation unit 310, a gait score value output unit 311, an animal identification device 40, and animal gait scoring method steps S300 to S318. DETAILED DESCRIPTION

[0029] The animal gait scoring system provided by the present invention utilizes the left and right asymmetry of the characteristic back part of a lame animal, analyzes and calculates the depth image of the animal's back to obtain the left and right actual volume values ​​of the characteristic back part, calculates the ratio of the left actual volume value to the right actual volume value to obtain the animal gait scoring value, and determines the lameness degree of the lame animal through the animal gait scoring value and a reference value determined through actual measurement.

[0030] Please also see Figure 1 The animal gait scoring system 10 includes a depth image acquisition module 20 and a gait scoring module 30; the depth image acquisition module 20 is used to acquire a back depth image reflecting the back contour of the animal from above the animal to be tested, and provide the back depth image to the gait scoring module 30. The depth image acquisition module 20 can be a depth camera, and the animal is kept in a stationary standing state during shooting. The depth image can accurately know the spatial distance of each pixel in the image, so by adding the (x, y) coordinates of each pixel in the two-dimensional image, the three-dimensional spatial coordinates of each point in the image can be obtained.

[0031] The gait scoring module 30 is used to extract image information of key parts of the animal and image information of reference parts from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to each pixel information; according to the extracted image information of the key parts, a predetermined number of plane feature point information for constructing a plane is obtained according to a first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information; according to the extracted image information of the reference part, a height reference feature point information is obtained according to a second extraction rule, and the height reference feature point information includes pixel information corresponding to the height reference feature point and three-dimensional coordinate points corresponding to the pixel information; according to the height reference feature point information, the plane feature point information and a preset height value, a reference bottom surface is constructed; for example, if the animal is a cow, please also check Figures 3 to 5 The key parts are the hip angle, tail root and hip bone of the animal. In the hip angle image information, the two hip angle edges are marked as point 4 and point 6, and the endpoint pixel information of point 4 and point 6 are obtained from the hip angle image information according to the marked points 4 and 6; in the tail root image information, the tail root edge is marked as point 5, and the endpoint pixel information of point 5 is obtained from the tail root image information according to the marked point 5; point 1 and point 3 are marked in the two opposite edges in the hip bone image information, the distance between point 1 and point 3 is the longest, and the endpoint pixel information of point 1 and point 3 are obtained from the hip bone image information according to the marked points 1 and 3, and the endpoint pixel information of point 1 and point 3 are determined in the back depth image according to the length between the endpoint pixel information of point 1 and the endpoint pixel information of point 3 according to the requirements of an equilateral triangle, and the endpoint pixel information of point 2 is far away from the endpoint pixel information of point 5; the above endpoint pixel information of points 1 to 6 is used as the plane feature point information. The reference part is the hip tubercle, point 7 is marked in the hip tubercle image information, and the endpoint pixel information of point 7 is obtained from the hip tubercle image information based on the marked point 7, wherein point 7 is the point highest from the ground in the back depth image, and the endpoint pixel information of point 7 is used as the height reference feature point information; the height reference feature point information is used as the height reference point, and the reference bottom surface (plane ABCDFE) is determined according to the plane feature point information and a preset height value (for example, the height value is 20 cm), and the reference bottom surface has a lower mapping feature point corresponding to the plane feature point.

[0032] The gait scoring module 30 evenly divides the target image area surrounded by the plane feature points according to the head-tail direction of the animal to obtain a left half target image area and a right half target image area, calculates the relative height value of each pixel in the left half target image area to the reference bottom surface according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculates the volume value corresponding to each pixel point in the left half target image area according to the relative height value, and integrates the volume of each pixel point in the target image area to obtain the actual volume value V reflecting the left side of the animal. left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right right The size of is calculated by taking the smaller volume value as the numerator and the larger volume value as the denominator to obtain the corresponding actual lameness ratio value, and matching the actual lameness ratio value with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value. For example, the actual lameness ratio value is δ, and the lameness status of dairy cows is divided into 5 levels (level 1, level 2, level 3, level 4, and level 5) by continuously monitoring the health status of a large number of dairy cows. The higher the level, the more serious the lameness status of the dairy cow. The average value range of all δ values ​​of dairy cows at level 1 is defined as K1{X 1begin , X 1end The average value of all standard lameness ratio values ​​of dairy cows at level 2 is defined as K2{X 2begin , X 2end The average value range of all standard lameness ratio values ​​of grade 3 dairy cows is defined as K3{X 3begin , X 3end}; The average value range of all standard lameness ratio values ​​of dairy cows at level 4 is defined as K4{X 4begin , X 4end The average value of all standard lameness ratio values ​​of dairy cows of level 5 is defined as K5{X 5begin , X 5end}, that is, the range of the standard lameness ratio is {X ibegin , X iend}, i = 1 or 2 or 3 or 4 or 5, according to V left 、V rightObtain the δ value, compare the δ value with the standard lameness ratio value, and then find the corresponding lameness score value (grade).

[0033] Furthermore, the animal gait scoring system 10 also includes an animal identification device 40,

[0034] The animal identification device 40 is used to obtain the identification information carried by the animal and provide the identification information to the gait scoring module; the gait scoring module 30 is also used to establish a corresponding relationship based on the identification information and the back depth image to form a search information table; the gait scoring module 30 is also used to display the obtained animal lameness score value and the corresponding identification information based on the search information table.

[0035] The gait scoring module 30 may be a single chip microcomputer or a microcomputer running a set of computer applications, wherein the computer applications are used to complete the gait scoring function of the animal. After the gait scoring module runs the computer application, the following functional modules are generated. Please also refer to Figure 2 :

[0036] The gait scoring module 30 includes an image information extraction unit 301, a first extraction unit 302, a second extraction unit 303, a reference bottom surface construction unit 304, a target image area bisection unit 305, a left actual volume calculation unit 306, a right actual volume calculation unit 307, an actual lameness ratio value calculation unit 308, a lameness score value search unit 309, a search information table generation unit 310 and a gait score value output unit 311.

[0037] The image information extraction unit 301 is used to extract image information of key parts and reference parts of the animal from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to the pixel information;

[0038] The first extraction unit 302 is used to obtain a predetermined number of plane feature point information for constructing a plane according to the first extraction rule based on the image information of the extracted key parts, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information;

[0039] The second extraction unit 303 is used to obtain a height reference feature point information according to the extracted image information of the reference part according to the second extraction rule, and the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information;

[0040] A reference bottom surface construction unit 304 is used to construct a reference bottom surface according to the height reference feature point information, the plane feature point information and a preset height value;

[0041] The target image region bisection unit 305 is used to equally divide the target image region surrounded by the plane feature points according to the head and tail direction of the animal to obtain a left half target image region and a right half target image region;

[0042] The left side actual volume calculation unit 306 is used to calculate the relative height value from each pixel in the left half target image area to the reference bottom surface according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculate the volume value corresponding to each pixel point in the left half target image area according to the relative height value, and integrate the volume of each pixel point in the target image area to obtain the left side actual volume value V reflecting the animal. left ;

[0043] The right side actual volume calculation unit 307 is used to calculate the relative height value from each pixel in the right half target image area to the reference bottom surface according to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculate the volume value corresponding to each pixel point in the right half target image area according to the relative height value, and integrate the volume of each pixel point in the right half target image area to obtain the right side actual volume value V reflecting the animal. right ;

[0044] The actual limp ratio value calculation unit 308 is used to compare the actual volume value V left 、The actual volume value V on the right right The smaller volume value is used as the numerator and the larger volume value is used as the denominator to calculate to obtain the corresponding actual lameness ratio value;

[0045] The lameness score value searching unit 309 is used to match the actual lameness ratio value with the pre-stored standard lameness ratio value, so as to search for the corresponding lameness score value according to the matched standard lameness ratio value.

[0046] A lookup information table generating unit 310 is used to establish a corresponding relationship according to the identification information and the back depth image to form a lookup information table;

[0047] The gait score output unit 311 is used to display the obtained lameness score of the animal and the corresponding identification information according to the search information table or provide it to a third-party application software for display using the third-party application software. For example, the gait score output unit 311 can provide the lameness score of the animal and the corresponding identification information to a cloud platform or a mobile phone APP.

[0048] In other implementations, the above functional modules generated by the gait scoring module may also be implemented by hardware logic circuits and embedded codes.

[0049] Furthermore, the present invention also provides an animal gait scoring method, see Figure 6 , the animal gait scoring method comprises the following steps:

[0050] Step S300: using a depth camera to obtain a back depth image reflecting the back contour of the animal from above the animal to be tested.

[0051] Step S302, extracting image information of key parts and reference parts of the animal from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to each pixel information;

[0052] Step S304, according to the extracted image information of the key parts, a predetermined number of plane feature point information for constructing the plane is obtained according to the first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information. The first extraction rule is: determine the endpoint pixel information of each key part according to the edge of each key part, and record each endpoint pixel information as plane feature point information.

[0053] Step S306, based on the extracted image information of the reference part, obtain a height reference feature point information according to the second extraction rule, the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information; wherein the second extraction rule is: select the highest pixel information from the pixel information of the reference part from the ground, and use the selected highest pixel information as the height reference feature point information.

[0054] Step S308, constructing a reference bottom surface according to the height reference feature point information, the plane feature point information and the preset height value;

[0055] Step S310, dividing the target image area surrounded by the plane feature points evenly according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area;

[0056] Step S312, based on the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculate the relative height value from each pixel in the left half target image area to the reference bottom surface; calculate the volume value corresponding to each pixel point in the left half target image area based on the relative height value, and integrate the volume of each pixel point in the target image area to obtain the actual volume value V reflecting the left side of the animalleft ;

[0057] Step S314, based on the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated based on the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ;

[0058] Step S316, compare the actual volume value V on the left side left 、The actual volume value V on the right right The smaller volume value is used as the numerator and the larger volume value is used as the denominator to calculate to obtain the corresponding actual lameness ratio value;

[0059] Step S318, matching the actual lameness ratio value with the pre-stored standard lameness ratio value, so as to search for a corresponding lameness score value according to the matched standard lameness ratio value.

[0060] Among them, step S304 is specifically as follows: the key parts are the hip angle, tail root, and hip bone of the animal. In the hip angle image information, the two hip angle edges are marked as point 4 and point 6, and the endpoint pixel information of point 4 and point 6 are obtained from the hip angle image information according to the marked points 4 and 6; in the tail root image information, the tail root edge is marked as point 5, and the endpoint pixel information of point 5 is obtained from the tail root image information according to the marked point 5; point 1 and point 3 are marked in the two opposite edges in the hip bone image information, the distance between point 1 and point 3 is the longest, and the endpoint pixel information of point 1 and point 3 are obtained from the hip bone image information according to the marked points 1 and 3, and the endpoint pixel information of point 2 and point 3 are determined in the back depth image according to the length between the endpoint pixel information 1 and the endpoint pixel information 3 according to the requirements of an equilateral triangle, and the endpoint pixel information of point 2 is far away from the endpoint pixel information of point 5; the above-mentioned endpoint pixel information of points 1 to 6 is used as the plane feature point information.

[0061] Among them, step S306 is specifically as follows: the reference part is the hip nodule, point 7 is marked in the hip nodule image information, and the endpoint pixel information of point 7 is obtained from the hip nodule image information based on the marked point 7, wherein point 7 is the highest point from the ground in the back depth image, and the endpoint pixel information of point 7 is used as the height reference feature point information.

[0062] Specifically, step S308 includes: using the height reference feature point information as the height reference point, determining a reference bottom surface according to the plane feature point information and a preset height value, wherein the reference bottom surface has mapping feature points corresponding to the plane feature points.

[0063] Furthermore, the above-mentioned animal gait scoring method further comprises the following steps: obtaining identification information carried by the animal;

[0064] Establishing a correspondence between the identification information and the back depth image;

[0065] A corresponding relationship is established based on the identification information and the back depth image, and the obtained lameness score value of the animal and the corresponding identification information are displayed.

[0066] In the above-mentioned animal gait scoring system 10 and method, a back depth image reflecting the back contour of the animal is obtained from above the animal to be tested, image information reflecting the key parts of the animal and image information of the reference parts are extracted from the back depth image, a predetermined number of plane feature point information for constructing a plane is obtained according to the extracted image information of the key parts, a height reference feature point information is obtained according to the extracted image information of the reference part, and a reference bottom surface is constructed according to the height reference feature point information, the plane feature point information and the preset height value; the target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area; the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface; the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right rightThe size of the volume is calculated by taking the smaller volume value as the numerator and the larger volume value as the denominator to obtain the corresponding actual lameness ratio value; the actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value, and then the animal's gait score is used to determine whether the animal is lame. Animal gait scoring can reveal whether the dairy cows on the ranch are lame at an early stage. Preemptive measures for lame dairy cows on the farm are the key to reducing the expensive disease of lameness. Good records of lameness and hoof diseases of dairy cows are crucial. Early detection and treatment of common hoof diseases and normalization of hoof care are required. Combined with regular gait scoring, it is the standard for farm management, which can track and improve the herd for the farm. Recording the incidence of lameness can also find the cause of abnormal cow gait and develop a plan to prevent cow lameness. Make each work detailed, process-oriented and data-oriented, truly reduce the occurrence of hoof diseases and improve the economic benefits of dairy farming.

Claims

1. An animal gait scoring system, characterized in that: Including depth image acquisition module and gait scoring module; The depth image acquisition module is used to acquire a back depth image reflecting the back contour of the animal from above the animal to be tested, and provide the back depth image to the gait scoring module; The gait scoring module is used to extract image information of key parts of the animal and image information of reference parts from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and the reference parts, and three-dimensional coordinate points corresponding to each pixel information; according to the extracted image information of the key parts, a predetermined number of plane feature point information for constructing a plane is obtained according to a first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information; according to the extracted image information of the reference part, a height reference feature point information is obtained according to a second extraction rule, and the height reference feature point information includes pixel information corresponding to the height reference feature point and three-dimensional coordinate points corresponding to the pixel information The three-dimensional coordinate points of the reference bottom surface are constructed according to the height reference feature point information, the plane feature point information and the preset height value; the target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area, and the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, and the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal. left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right right The size of the volume is calculated with the smaller volume value as the numerator and the larger volume value as the denominator to obtain the corresponding actual lameness ratio value, and the actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value.

2. The animal gait scoring system according to claim 1, characterized in that: The gait scoring module includes an image information extraction unit, a first extraction unit, a second extraction unit, a reference bottom surface construction unit, a target image area bisection unit, a left actual volume calculation unit, a right actual volume calculation unit, an actual lameness ratio value calculation unit and a lameness score value search unit; An image information extraction unit, used to extract image information reflecting key parts of the animal and image information of reference parts from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to the pixel information; A first extraction unit is used to obtain a predetermined number of plane feature point information for constructing a plane according to the extracted image information of the key parts and a first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and a three-dimensional coordinate point corresponding to the pixel information; A second extraction unit is used to obtain a height reference feature point information according to the extracted image information of the reference part and a second extraction rule, wherein the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information; A reference bottom surface construction unit is used to construct a reference bottom surface according to the height reference feature point information, the plane feature point information and a preset height value; The target image region bisection unit is used to equally divide the target image region surrounded by the plane feature points according to the head and tail direction of the animal to obtain a left half target image region and a right half target image region; The left side actual volume calculation unit is used to calculate the relative height value from each pixel in the left half target image area to the reference bottom surface according to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculate the volume value corresponding to each pixel point in the left half target image area according to the relative height value, and integrate the volume of each pixel point in the target image area to obtain the left side actual volume value V reflecting the animal left ; The right side actual volume calculation unit is used to calculate the relative height value from each pixel in the right half target image area to the reference bottom surface according to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, calculate the volume value corresponding to each pixel point in the right half target image area according to the relative height value, and integrate the volume of each pixel point in the right half target image area to obtain the right side actual volume value V reflecting the animal right ; The actual lameness ratio value calculation unit is used to compare the actual volume value V on the left side left 、The actual volume value V on the right right The smaller volume value is used as the numerator and the larger volume value is used as the denominator to calculate to obtain the corresponding actual lameness ratio value; The lameness score value search unit is used to match the actual lameness ratio value with the pre-stored standard lameness ratio value, so as to search for the corresponding lameness score value according to the matched standard lameness ratio value.

3. The animal gait scoring system according to claim 1 or 2, characterized in that: The animal gait scoring system also includes an animal identification device, The animal identification device is used to obtain identification information carried by the animal and provide the identification information to the gait scoring module; The gait scoring module is also used to establish a corresponding relationship according to the identification information and the back depth image to form a search information table; the gait scoring module is also used to display the obtained animal lameness score value and the corresponding identification information according to the search information table.

4. The animal gait scoring system according to claim 3, characterized in that: The first extraction rule is: determine the endpoint pixel information of each key part according to the edge of each key part, and record each endpoint pixel information as plane feature point information; the second extraction rule is: select the highest pixel information from the ground from the pixel information of the reference part, and use the selected highest pixel information as the height reference feature point information.

5. A method for scoring animal gait, comprising the following steps: Using a depth camera to obtain a back depth image reflecting the back contour of the animal from above the animal to be tested; Extracting image information of key parts and reference parts of the animal from the acquired back depth image; wherein the image information includes pixel information reflecting the key parts and reference parts, and three-dimensional coordinate points corresponding to each pixel information; According to the extracted image information of the key parts, a predetermined number of plane feature point information for constructing the plane is obtained according to the first extraction rule, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information; According to the extracted image information of the reference part, a height reference feature point information is obtained according to the second extraction rule, the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information; Constructing a reference bottom surface according to the height reference feature point information, the plane feature point information and the preset height value; The target image area surrounded by the plane feature points is evenly divided according to the head and tail direction of the animal to obtain a left half target image area and a right half target image area; According to the pixel information in the left half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value of each pixel in the left half target image area to the reference bottom surface is calculated; the volume value corresponding to each pixel point in the left half target image area is calculated according to the relative height value, and the volume of each pixel point in the target image area is integrated to obtain the actual volume value V reflecting the left side of the animal left ; According to the pixel information in the right half target image area and the three-dimensional coordinate points corresponding to the pixel information and the three-dimensional coordinate points of the reference bottom surface, the relative height value from each pixel in the right half target image area to the reference bottom surface is calculated, and the volume value corresponding to each pixel point in the right half target image area is calculated according to the relative height value, and the volume of each pixel point in the right half target image area is integrated to obtain the actual volume value V reflecting the right side of the animal right ; Compare the actual volume value V on the left left 、The actual volume value V on the right right The smaller volume value is used as the numerator and the larger volume value is used as the denominator to calculate to obtain the corresponding actual lameness ratio value; The actual lameness ratio value is matched with the pre-stored standard lameness ratio value to find the corresponding lameness score value according to the matched standard lameness ratio value.

6. The animal gait scoring method according to claim 5, characterized in that: The first extraction rule is: determine the endpoint pixel information of each key part according to the edge of each key part, and record each endpoint pixel information as plane feature point information; the second extraction rule is: select the highest pixel information from the ground from the pixel information of the reference part, and use the selected highest pixel information as the height reference feature point information.

7. The animal gait scoring method according to claim 6, characterized in that: The step of "acquiring a predetermined number of plane feature point information for constructing a plane according to the first extraction rule based on the extracted image information of the key parts, wherein the plane feature point information includes pixel information corresponding to the plane feature point and three-dimensional coordinate points corresponding to the pixel information" is specifically: The key parts are the hip angle, tail root and hip bone of the animal. In the hip angle image information, the two hip angle edges are marked as point 4 and point 6, and the endpoint pixel information of point 4 and the endpoint pixel information of point 6 are obtained from the hip angle image information according to the marked point 4 and point 6; In the tail root image information, the edge of the tail root is marked as point 5, and the endpoint pixel information of point 5 is obtained from the tail root image information according to the marked point 5; Point 1 and point 3 are marked on two opposite edges in the hip bone image information, and the distance between point 1 and point 3 is the longest. The endpoint pixel information of point 1 and point 3 are obtained from the hip bone image information based on the marked points 1 and 3. The endpoint pixel information of point 1 and point 3 are determined in the back depth image based on the length between the endpoint pixel information of point 1 and the endpoint pixel information of point 3 and in accordance with the requirement of an equilateral triangle. The endpoint pixel information of point 2 is far away from the endpoint pixel information of point 5. The above endpoint pixel information of points 1 to 6 is used as plane feature point information.

8. The animal gait scoring method according to claim 6 or 5, characterized in that: The specific step of "according to the extracted image information of the reference part, a height reference feature point information is obtained according to the second extraction rule, the height reference feature point information includes pixel information corresponding to the height reference feature point and a three-dimensional coordinate point corresponding to the pixel information" is as follows: the reference part is the hip tubercle, point 7 is marked in the hip tubercle image information, and the endpoint pixel information of point 7 is obtained from the hip tubercle image information based on the marked point 7, wherein point 7 is the highest point from the ground in the back depth image, and the endpoint pixel information of point 7 is used as the height reference feature point information.

9. The animal gait scoring method according to claim 8, characterized in that: The specific step of "constructing a reference bottom surface based on height reference feature point information, plane feature point information and a preset height value" is as follows: using the height reference feature point information as the height reference point, determining the reference bottom surface based on the plane feature point information and the preset height value, and the reference bottom surface has mapping feature points corresponding to the plane feature points.

10. The animal gait scoring method according to claim 9, characterized in that: The steps are also included: obtaining identification information carried by the animal; Establishing a correspondence between the identification information and the back depth image; A corresponding relationship is established based on the identification information and the back depth image, and the obtained lameness score value of the animal and the corresponding identification information are displayed.

Citation Information

Patent Citations

  • Dairy cow early limp recognition method and device

    CN112293295A

  • Cow limp motion characteristic detection method and device based on thermal infrared

    CN112528772A

  • Arrangement and method for determining a body condition score of an animal

    US20110279650A1

  • System and method for detection of lameness in sport horses and other quadrupeds

    WO2016200564A1