Method and system for estimating volume ratio of high-value part of beef cattle based on point cloud

Abstract

The invention discloses a method and system for estimating the volume ratio of a high-value part of beef cattle based on point cloud, and relates to the technical field of livestock informatization. The method comprises the following steps: firstly, acquiring three-dimensional point cloud data of beef cattle, and preprocessing the three-dimensional point cloud data of the beef cattle; then, based on the preprocessed three-dimensional point cloud data of the beef cattle, extracting the volume of the beef cattle and the volume of a target part; and finally, determining whether the value of the beef cattle exceeds a preset average value according to the volume of the target part. According to the method and the system provided by the invention, the volume ratio of the target part in the beef cattle can be effectively determined, so that whether the value of one beef cattle exceeds the average value or not is judged, manual errors are effectively reduced, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of animal husbandry informatization, and in particular to a method and system for estimating the volume proportion of high-value parts of beef cattle based on point clouds. Background Art

[0002] In the wholesale and slaughter of beef cattle, it is necessary to select individual beef cattle and choose the more valuable individuals based on their growth conditions.

[0003] Currently, the selection of individual beef cattle is still done by manual visual inspection, which has certain safety risks, is time-consuming and labor-intensive, and may even lead to misjudgment.

[0004] With the development of artificial intelligence technology, the method of combining point cloud data captured by depth cameras and deep learning methods can realize automatic selection of beef cattle individuals according to actual needs. The selection method using artificial intelligence can better avoid safety hazards and misjudgments.

[0005] However, the existing methods using artificial intelligence cannot effectively calculate the proportion of the volume of the target part from the beef cattle itself, and thus cannot judge whether the value of a beef cattle exceeds the average based on the volume proportion of the target part, resulting in the need for more manual intervention and cooperation, which in turn leads to problems of manual errors and low work efficiency. Summary of the invention

[0006] In order to overcome the problem in existing animal husbandry information technology that it is impossible to judge whether the value of a beef cattle exceeds the average value based on the volume proportion of the target part, resulting in human errors and low work efficiency, the present invention proposes a method and system for estimating the volume proportion of high-value parts of beef cattle based on point clouds, which can effectively confirm the volume proportion of the target part in the beef cattle to judge whether the value of a beef cattle exceeds the average value, effectively reducing human errors and improving work efficiency.

[0007] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0008] A method for estimating the volume proportion of high-value parts of beef cattle based on point clouds, the method comprising:

[0009] Acquiring three-dimensional point cloud data of beef cattle, and preprocessing the three-dimensional point cloud data of beef cattle;

[0010] Based on the preprocessed three-dimensional point cloud data of beef cattle, the volume of beef cattle and the volume of target parts are extracted;

[0011] Determine if the value of the cattle exceeds a preset average based on target part volumes.

[0012] In the above technical scheme, by preprocessing the acquired three-dimensional point cloud data of beef cattle, the reliability and stability of the data can be improved, thereby improving work efficiency; extracting the volume of beef cattle and the volume of target parts, and extracting the volume of beef cattle and the volume of target parts can effectively confirm the volume proportion of the target part in the beef cattle, so as to judge whether the value of a beef cattle exceeds the average value, effectively reducing human errors and improving work efficiency.

[0013] Furthermore, the process of preprocessing the three-dimensional point cloud data of beef cattle includes:

[0014] The acquired three-dimensional point cloud data of beef cattle is subjected to noise reduction and point cloud alignment.

[0015] In the above technical solution, the noise reduction and point cloud alignment processing of the acquired beef cattle three-dimensional point cloud data can effectively reduce the interference of external factors, while improving the accuracy and completeness of the point cloud data, thereby improving work efficiency.

[0016] Further, the process of extracting the volume of beef cattle and the volume of target parts includes:

[0017] Input the three-dimensional point cloud data of beef cattle into the preset neural network to segment the target parts;

[0018] Based on the segmented target parts, the volume of beef cattle and the volume of the target parts are calculated;

[0019] Calculate the proportion of the target part volume based on the calculated beef cattle volume and the target part volume;

[0020] Determine whether the current value of beef cattle exceeds the average based on the proportion of the target part volume.

[0021] Furthermore, the process of inputting the three-dimensional point cloud data of beef cattle into the preset neural network model for target part segmentation includes:

[0022] The neural network model performs beef cattle point cloud segmentation on the three-dimensional point cloud data of beef cattle to obtain several different parts, including the front, middle and back of the cattle;

[0023] Calculate the midpoint of the point cloud in the up-down direction, traverse each point, and divide the front of the cow into two parts according to its positional relationship with the midpoint of the up-down direction.

[0024] According to the point cloud data of the cow body, the point cloud data is projected onto a two-dimensional plane, and key points are marked equidistantly along the abdominal curve according to the two-dimensional plane, and the marked key points are restored to a three-dimensional point cloud, thereby dividing the point cloud of the cow body into an upper and lower part;

[0025] Calculate the horizontal boundary of the cow's middle body point cloud and the horizontal boundary of each part, and traverse each point to divide the upper half of the cow's middle body into three equal parts.

[0026] Further, the process of calculating the volume of beef cattle and the volume of target parts includes:

[0027] Read the point cloud data that needs to be volumetrically calculated, and slice the beef cattle point cloud into slices according to the preset thickness;

[0028] Establish a spatial coordinate system and project the sliced ​​beef cattle point cloud onto the coordinate axis to generate three-dimensional scattered points;

[0029] Generate a cross-sectional contour boundary according to the three-dimensional scattered points, and calculate the cross-sectional area;

[0030] Based on the calculated cross-sectional area, the point cloud volume is calculated.

[0031] In the above technical solution, the three-dimensional point cloud data of beef cattle is input into a preset neural network for target part segmentation, which can accurately process the data, reduce errors and the number of repeated verifications in the data calculation process, so as to improve the calculation efficiency; the volume of beef cattle and the volume of target parts are calculated, and the proportion of the volume of target parts is calculated based on the calculated volume of beef cattle and the volume of target parts, which can more quickly determine whether the current value of beef cattle exceeds the average value, thereby reducing manual errors and improving work efficiency.

[0032] Furthermore, the process of slicing the beef cattle point cloud according to a preset thickness includes:

[0033] Assume that the coordinates of each point in the point cloud are (x i ,y i ,z i ), calculate the minimum and maximum values ​​of the bounding box of the point cloud, the expression is:

[0034] x max =min(x i ), x min =max(x i );

[0035] y max =min(y i ), y min =max(y i );

[0036] z max =min(z i ), z min =max(z i );

[0037] Calculate the height of the point cloud in each coordinate axis direction. The expression is:

[0038]

[0039] Calculate the point cloud density: count the number of points N in the point cloud and calculate the total spatial volume of the point cloud. The expression is:

[0040] V=H x ×H y ×H z ;

[0041] Then the point cloud density ρ is:

[0042]

[0043] The slice thickness is dynamically adjusted according to the point cloud density. Thinner slices are used in high-density areas, and thicker slices are used in low-density areas. Let the baseline slice thickness be Δh0, corresponding to the baseline density ρ0; for the area with density ρ0, the slice thickness Δh0 is:

[0044]

[0045] Calculate the slice position, and determine the position z of each slice in the Z-axis direction according to the calculated slice thickness Δh k , the expression is:

[0046]

[0047] Slice the beef cattle point cloud according to the position of each slice;

[0048] in, Indicates rounding down.

[0049] Furthermore, the expression for calculating the cross-sectional area is:

[0050]

[0051] The contour cross-section polygon has m points (x0, y0), (x0, y0),…, (x m-1 ,y m-1 ) surrounded by the profile section V i Area Calculated by Shoelace Formula, a represents the profile section V i The vertex number of .

[0052] Furthermore, the expression for calculating the point cloud volume is:

[0053]

[0054] Among them, h lastIndicates the slice thickness of the last beef cattle point cloud.

[0055] Furthermore, the neural network model is a Pointnet++ segmentation network model.

[0056] In the above technical solution, the beef cattle point cloud is sliced ​​according to a preset thickness, which can reduce the error and the number of repeated verifications in the data calculation process to improve the calculation efficiency; the Pointnet++ segmentation network model used can better segment the beef cattle point cloud and reduce errors.

[0057] A system for estimating the volume proportion of high-value parts of beef cattle based on point clouds, the system comprising:

[0058] An acquisition module is used to acquire three-dimensional point cloud data of beef cattle;

[0059] A first processing module, used for preprocessing the three-dimensional point cloud data of the beef cattle;

[0060] The second processing module is used to extract the volume of the beef cattle and the volume of the target part based on the preprocessed three-dimensional point cloud data of the beef cattle;

[0061] An evaluation module determines whether the value of the beef cattle exceeds a preset average value based on the target part volume.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The present invention proposes a method and system for estimating the volume proportion of high-value parts of beef cattle based on point clouds. Firstly, by preprocessing the acquired three-dimensional point cloud data of beef cattle, the reliability and stability of the data can be improved, thereby improving work efficiency; then, the volume of the beef cattle and the volume of the target part are extracted, and the volume proportion of the target part in the beef cattle can be effectively confirmed to judge whether the value of a beef cattle exceeds the average value, thereby effectively reducing manual errors and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A flow chart of a method for estimating the volume proportion of high-value parts of beef cattle based on point clouds provided in an embodiment of the present application;

[0065] Figure 2 A three-dimensional point cloud data map of beef cattle provided in an embodiment of the present application;

[0066] Figure 3 A schematic diagram of target part segmentation of beef cattle provided in an embodiment of the present application;

[0067] Figure 4 A schematic diagram of the structure of a system for estimating the volume percentage of high-value parts of beef cattle based on point clouds provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Preferred embodiments of the present invention are provided in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0070] Embodiment 1:

[0071] This embodiment provides a method for estimating the volume proportion of high-value parts of beef cattle based on point clouds. Figure 1 , the method comprising:

[0072] Step S1: acquiring three-dimensional point cloud data of beef cattle, and preprocessing the three-dimensional point cloud data of beef cattle;

[0073] Step S2: extracting the volume of the beef cattle and the volume of the target part based on the preprocessed three-dimensional point cloud data of the beef cattle;

[0074] Step S3: Determine whether the value of the beef cattle exceeds a preset average value based on the volume of the target part.

[0075] Specifically, in step S1, the process of preprocessing the three-dimensional point cloud data of beef cattle includes:

[0076] The acquired three-dimensional point cloud data of beef cattle is subjected to noise reduction and point cloud alignment.

[0077] It can be understood that denoising and point cloud alignment processing of the acquired beef cattle three-dimensional point cloud data can effectively reduce the interference of external factors, while improving the accuracy and completeness of the point cloud data, thereby improving work efficiency.

[0078] As a preferred embodiment, in step S2, the process of extracting the volume of the beef cattle and the volume of the target part includes:

[0079] Step S21: inputting the three-dimensional point cloud data of beef cattle into a preset neural network to perform target part segmentation;

[0080] Step S22: calculating the volume of the beef cattle and the volume of the target part based on the segmented target part;

[0081] Step S23: Calculating the proportion of the target part volume according to the calculated beef cattle volume and the target part volume;

[0082] Step S24: Determine whether the current value of the beef cattle exceeds the average value based on the proportion of the target part volume.

[0083] Specifically, in step S21, see Figure 2 ,The process of inputting the beef cattle 3D point cloud data into the preset neural network model for target part segmentation includes:

[0084] The neural network model (i.e., Pointnet++ segmentation network model) performs beef cattle point cloud segmentation on the three-dimensional point cloud data of beef cattle, and obtains several different parts, which are divided into 14 parts in total, namely, the head, the front of the body, the middle of the body, the back of the body, the tail, the limbs, and the ears. The limbs include the upper left front leg, the lower left front leg, the upper right front leg, the lower right front leg, the upper left hind leg, the lower left hind leg, the upper right hind leg, and the lower right hind leg.

[0085] Specifically, the midpoint of the point cloud in the up-down direction is calculated, and each point is traversed. According to its positional relationship with the midpoint in the up-down direction, the front of the cow is divided into two parts according to the midpoint.

[0086] Specifically, the point cloud data of the cow's body is read, and according to the point cloud data of the cow's body, the point cloud data is projected onto a two-dimensional plane (left view and right view), key points are equidistantly marked along the abdomen curve in the two-dimensional left view and right view, the marked key points are restored to a three-dimensional point cloud, a curve is fitted through the key points of the left view and the right view, the key points in the left and right views are connected, and the point cloud of the cow's body is divided into upper and lower parts according to the surface formed by the two curves;

[0087] Calculate the horizontal boundary of the cow's middle body point cloud and the horizontal boundary of each part, and traverse each point to divide the upper half of the cow's middle body into three equal parts.

[0088] As a preferred embodiment, in step S22, the process of calculating the volume of the beef cattle and the volume of the target part includes:

[0089] Step S221: reading point cloud data for which volume is required, and slicing the beef cattle point cloud according to a preset thickness;

[0090] Step S222: establishing a spatial coordinate system, projecting the sliced ​​beef cattle point cloud onto the coordinate axis to generate three-dimensional scattered points;

[0091] Step S223: generating a cross-sectional contour boundary according to the three-dimensional scattered points, and calculating the cross-sectional area;

[0092] Step S224: Calculate the point cloud volume based on the calculated cross-sectional area.

[0093] It can be understood that inputting the three-dimensional point cloud data of beef cattle into a preset neural network for target part segmentation can accurately process the data, reduce errors and the number of repeated verifications in the data calculation process, so as to improve calculation efficiency; calculating the volume of beef cattle and the volume of target parts, and calculating the proportion of the target part volume based on the calculated volume of beef cattle and target parts, can more quickly determine whether the current value of beef cattle exceeds the average value, thereby reducing manual errors and improving work efficiency.

[0094] Specifically, in step S3, the average value is the average value of the volume percentage of 100 beef cattle measured in advance according to a method for estimating the volume percentage of high-value parts of beef cattle based on point clouds.

[0095] In an embodiment, by preprocessing the acquired three-dimensional point cloud data of beef cattle, the reliability and stability of the data can be improved, thereby improving work efficiency; extracting the volume of beef cattle and the volume of target parts, and extracting the volume of beef cattle and the volume of target parts can effectively confirm the volume proportion of the target part in the beef cattle, so as to judge whether the value of a beef cattle exceeds the average value, effectively reducing human errors and improving work efficiency.

[0096] Embodiment 2:

[0097] This embodiment further describes steps S221 to S224 in the first embodiment, as follows:

[0098] Specifically, in step S221, see Figure 3 ,The process of slicing the beef cattle point cloud according to the preset thickness includes:

[0099] Assume that the coordinates of each point in the point cloud are (x i ,y i ,z i ), calculate the minimum and maximum values ​​of the bounding box of the point cloud, the expression is:

[0100] x max =min(x i ), x min =max(x i );

[0101] y max =min(y i ), y min =max(y i );

[0102] z max =min(z i ), z min=max(z i );

[0103] Calculate the height of the point cloud in each coordinate axis direction. The expression is:

[0104]

[0105] Calculate the point cloud density: count the number of points N in the point cloud and calculate the total spatial volume of the point cloud. The expression is:

[0106] V=H x ×H y ×H z ;

[0107] Then the point cloud density ρ is:

[0108]

[0109] The slice thickness is dynamically adjusted according to the point cloud density. Thinner slices are used in high-density areas, and thicker slices are used in low-density areas. Let the baseline slice thickness be Δh0, corresponding to the baseline density ρ0; for the area with density ρ0, the slice thickness Δh0 is:

[0110]

[0111] If the density is above a threshold ρ0, the slice thickness is reduced:

[0112]

[0113] If the density is below the threshold ρ0, the slice thickness increases, but does not exceed a maximum value Δh max :

[0114]

[0115] Calculate the slice position, and determine the position z of each slice in the Z-axis direction according to the calculated slice thickness Δh k , the expression is:

[0116]

[0117] The number of slices is:

[0118] n=Ceiling[H / Δh];

[0119] Ceiling means rounding up, and the thickness of the last slice is:

[0120] h last =H-(n-1)*Δh;

[0121] Slice the beef cattle point cloud according to the position and thickness of each slice;

[0122] in, Indicates rounding down.

[0123] Specifically, in step S223, the expression for calculating the cross-sectional area is:

[0124]

[0125] Among them, the contour section polygon has m points (*0,y0), (x0,y0),…, (s m-1 ,y m-1 ) surrounded by the profile section V i Area Calculated by Shoelace Formula, a represents the profile section V i The vertex number of .

[0126] Specifically, in step S224, the expression for calculating the point cloud volume is:

[0127]

[0128] Among them, h last Indicates the slice thickness of the last beef cattle point cloud.

[0129] In this embodiment, the beef cattle point cloud is sliced ​​according to a preset thickness, which can reduce the error and the number of repeated verifications in the data calculation process to improve the calculation efficiency; the Pointnet++ segmentation network model used can better segment the beef cattle point cloud and reduce errors.

[0130] Embodiment three:

[0131] A system for estimating the volume proportion of high-value parts of beef cattle based on point clouds, see Figure 4 , the system comprising:

[0132] An acquisition module is used to acquire three-dimensional point cloud data of beef cattle;

[0133] A first processing module, used for preprocessing the three-dimensional point cloud data of the beef cattle;

[0134] The second processing module is used to extract the volume of the beef cattle and the volume of the target part based on the preprocessed three-dimensional point cloud data of the beef cattle;

[0135] An evaluation module determines whether the value of the beef cattle exceeds a preset average value based on the target part volume.

[0136] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for estimating the volume proportion of high-value parts of beef cattle based on point cloud, characterized in that: The method comprises: Acquiring three-dimensional point cloud data of beef cattle, and preprocessing the three-dimensional point cloud data of beef cattle; Based on the preprocessed three-dimensional point cloud data of beef cattle, the volume of beef cattle and the volume of target parts are extracted; Determine if the value of the cattle exceeds a preset average based on target part volumes.

2. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 1, characterized in that: The process of preprocessing the three-dimensional point cloud data of beef cattle includes: The acquired three-dimensional point cloud data of beef cattle is subjected to noise reduction and point cloud alignment.

3. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 1, characterized in that: The process of extracting beef cattle volume and target part volume includes: Input the three-dimensional point cloud data of beef cattle into the preset neural network to segment the target parts; Based on the segmented target parts, the volume of beef cattle and the volume of the target parts are calculated; Calculate the proportion of the target part volume based on the calculated beef cattle volume and the target part volume; Determine whether the current value of beef cattle exceeds the average based on the proportion of the target part volume.

4. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 3 is characterized in that: The process of inputting the 3D point cloud data of beef cattle into the preset neural network model for target part segmentation includes: The neural network model performs beef cattle point cloud segmentation on the three-dimensional point cloud data of beef cattle to obtain several different parts, including the front, middle and back of the cattle; Calculate the midpoint of the point cloud in the up-down direction, traverse each point, and divide the front of the cow into two parts according to its positional relationship with the midpoint of the up-down direction. According to the point cloud data of the cow body, the point cloud data is projected onto a two-dimensional plane, and key points are marked equidistantly along the abdominal curve according to the two-dimensional plane, and the marked key points are restored to a three-dimensional point cloud, thereby dividing the point cloud of the cow body into an upper and lower part; Calculate the horizontal boundary of the cow's middle body point cloud and the horizontal boundary of each part, and traverse each point to divide the upper half of the cow's middle body into three equal parts.

5. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 3, characterized in that: The process for calculating beef cattle volume and target part volumes includes: Read the point cloud data that needs to be volumetrically calculated, and slice the beef cattle point cloud into slices according to the preset thickness; Establish a spatial coordinate system and project the sliced ​​beef cattle point cloud onto the coordinate axis to generate three-dimensional scattered points; Generate a cross-sectional contour boundary according to the three-dimensional scattered points, and calculate the cross-sectional area; Based on the calculated cross-sectional area, the point cloud volume is calculated.

6. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 5, characterized in that: The process of slicing the beef cattle point cloud to a preset thickness includes: Assume that the coordinates of each point in the point cloud are (x i ,y i ,z i ), calculate the minimum and maximum values ​​of the bounding box of the point cloud, the expression is: x max =min(x i ),x min =max(x i ); and max =min(and i ),and min =max(y i ); With max =min(of i ),With min =max(from i ); Calculate the height of the point cloud in each coordinate axis direction. The expression is: Calculate the point cloud density: count the number of points N in the point cloud and calculate the total spatial volume of the point cloud. The expression is: V=H x ×H y ×H z ; Then the point cloud density ρ is: The slice thickness is dynamically adjusted according to the point cloud density. Thinner slices are used in high-density areas, and thicker slices are used in low-density areas. Let the baseline slice thickness be Δh0, corresponding to the baseline density ρ0; for the area with density ρ0, the slice thickness Δh0 is: Calculate the slice position, and determine the position z of each slice in the Z-axis direction according to the calculated slice thickness Δh k , the expression is: Slice the beef cattle point cloud according to the position of each slice; in, Indicates rounding down.

7. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 5, characterized in that: The expression for calculating the cross-sectional area is: The contour cross-section polygon has m points (x0, y0), (x0, y0),…, (x m-1 ,y m-1 ) surrounded by the profile section V i Area A Vi Calculated by Shoelace Formula, a represents the profile section V i The vertex number of .

8. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 5, characterized in that: The expression for calculating the point cloud volume is: Among them, h last Indicates the slice thickness of the last beef cattle point cloud.

9. The method for estimating the volume proportion of high-value parts of beef cattle based on point cloud according to claim 4, characterized in that: The neural network model is a Pointnet++ segmentation network model.

10. A system for estimating the volume ratio of high-value parts of beef cattle based on point clouds, characterized in that: The system comprises: An acquisition module is used to acquire three-dimensional point cloud data of beef cattle; A first processing module, used for preprocessing the three-dimensional point cloud data of the beef cattle; The second processing module is used to extract the volume of the beef cattle and the volume of the target part based on the preprocessed three-dimensional point cloud data of the beef cattle; An evaluation module determines whether the value of the beef cattle exceeds a preset average value based on the target part volume.

Images