A method for automatic extraction of fish phenotypes based on 3D scanning models

Abstract

The present invention relates to the field of three-dimensional scanning technology, and in particular to a method for automatically extracting fish phenotypes based on a three-dimensional scanning model. The method comprises the following steps: performing three-dimensional scanning processing on a fish sample by a three-dimensional scanning device to obtain three-dimensional point cloud data of the fish sample surface and three-dimensional texture data of the fish sample surface; reconstructing a three-dimensional scanning model of the fish sample to obtain a three-dimensional scanning reconstruction model of the fish sample; performing fish region grid segmentation processing and sub-region fish phenotypic feature analysis on the three-dimensional scanning reconstruction model of the fish sample to obtain fish phenotypic feature data sets of different fish segmentation sub-regions; performing fish phenotypic parameter calculation and chart visualization conversion on the fish phenotypic feature data sets of different fish segmentation sub-regions to generate fish feature parameter visualization phenotypes of different fish segmentation sub-regions. The present invention can realize automatic and accurate extraction of fish phenotypes by intelligently processing three-dimensional scanning data.

Description

Technical Field

[0001] The invention relates to the technical field of three-dimensional scanning, and in particular to a method for automatically extracting fish phenotypes based on a three-dimensional scanning model. Background Art

[0002] In modern biological research and aquaculture, the phenotypic characteristics of fish are of great significance for biological research, breeding and environmental monitoring. In recent years, 3D scanning technology, as a high-precision and high-efficiency data collection method, has been widely used in the morphological measurement of various organisms. By adopting advanced 3D scanning technology to accurately collect 3D data of fish, the resolution and scanning parameters of the scanner are optimized to ensure that the acquired data has sufficient details and accuracy. In addition, by introducing automated processing algorithms and using deep learning and computer vision technology to automatically analyze 3D scanning data, the phenotypic characteristics of fish such as body shape, body length and body width are automatically recognized and extracted by training neural network models. At the same time, the data analysis method based on the 3D model accurately extracts various phenotypic characteristics of fish, including the extraction of fish body contours, volume measurement, shape analysis, etc., to fully reflect the phenotypic characteristics of fish. However, traditional fish phenotypic extraction methods often require complex algorithms and a large amount of computing resources, and cannot automatically recognize and process subtle changes in the fish body, resulting in inaccurate extraction results. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a method for automatically extracting fish phenotypes based on a three-dimensional scanning model to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for automatically extracting fish phenotypes based on a three-dimensional scanning model comprises the following steps:

[0005] Step S1: performing three-dimensional scanning processing on the fish sample by a three-dimensional scanning device to obtain three-dimensional point cloud data of the fish sample surface and three-dimensional texture data of the fish sample surface; performing three-dimensional scanning model reconstruction on the fish sample based on the three-dimensional point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstruction model of the fish sample;

[0006] Step S2: performing fish region grid segmentation processing on the fish sample 3D scanning and reconstruction model to obtain different fish sample 3D grid segmentation sub-regions, wherein the different fish sample 3D grid segmentation sub-regions include a fish head 3D grid segmentation sub-region, a fish trunk 3D grid segmentation sub-region, and a fish tail fin 3D grid segmentation sub-region;

[0007] Step S3: performing sub-region fish phenotypic characteristic analysis on different fish sample three-dimensional grid segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions, wherein the fish phenotypic characteristic data sets of different fish segmentation sub-regions include the body length, body width, body height and external contour shape of the fish body part of the different fish segmentation sub-regions;

[0008] Step S4: Calculate fish phenotypic parameters for fish phenotypic feature data sets of different fish segmentation sub-regions to obtain fish phenotypic parameter sets for different fish segmentation sub-regions; perform graphical visualization conversion on the fish phenotypic parameter sets for different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters for different fish segmentation sub-regions.

[0009] Further, step S1 includes the following steps:

[0010] Step S11: optimizing and adjusting the scanning position and posture of the fish sample by a high-precision positioning device and an automatic adjustment mechanism to obtain the fish sample in a standardized scanning position and posture;

[0011] Step S12: performing three-dimensional scanning processing on the fish sample in the standardized scanning position and posture by a three-dimensional scanning device to obtain three-dimensional point cloud data of the surface of the fish sample;

[0012] Step S13: performing noise sparse filtering on the three-dimensional point cloud data of the fish sample surface to obtain three-dimensional denoised sparse point cloud data of the fish sample surface;

[0013] Step S14: performing a three-dimensional texture feature analysis on the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface to obtain three-dimensional texture data of the fish sample surface;

[0014] Step S15: reconstructing a three-dimensional scanning model of the fish sample in a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstructed model of the fish sample.

[0015] Further, step S13 includes the following steps:

[0016] Step S131: performing noise point distribution detection and annotation on the three-dimensional point cloud data of the fish sample surface to obtain the three-dimensional noise point distribution annotated point cloud data on the fish sample surface;

[0017] Step S132: performing two-dimensional section segmentation processing on the three-dimensional noise point distribution annotated point cloud data on the surface of the fish sample to obtain the noise point distribution annotated point cloud data under the two-dimensional segmented section of each fish sample;

[0018] Step S133: measuring and calculating the spatial distance between any two fish sample noise points in the noise point distribution annotated point cloud data under each fish sample two-dimensional segmentation section, and obtaining the spatial position distance value between any two fish sample noise points under each fish sample two-dimensional segmentation section;

[0019] Step S134: comparing and judging the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section according to a preset spatial position spacing threshold value; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is greater than or equal to the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is increased by 1; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is less than the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is reduced by 1, until the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is compared and judged, so as to construct a sparse matrix of the spatial distribution of fish sample noise points under each fish sample two-dimensional segmentation section;

[0020] Step S135: Based on the sparse matrix of the spatial distribution of fish sample noise points under each two-dimensional segmentation section, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to noise point sparse filtering processing to obtain the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section of the fish sample; the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section is subjected to cross-section three-dimensional integration and restoration processing to obtain the three-dimensional denoised sparse point cloud data of the fish sample surface.

[0021] Further, step S132 includes the following steps:

[0022] The global noise distribution spatial model is established for the annotated point cloud data of the three-dimensional noise distribution on the surface of the fish sample, and the three-dimensional global noise distribution spatial model on the surface of the fish sample is obtained;

[0023] The three-dimensional global noise distribution spatial model on the surface of fish samples is used to identify and analyze the three-dimensional morphology of fish samples and the distribution characteristics of noise points, and the three-dimensional distribution morphology of the fish sample surface and the distribution characteristics of three-dimensional noise points on the surface of fish samples are obtained;

[0024] Based on the three-dimensional distribution morphology of the fish sample surface and the three-dimensional noise distribution characteristics of the fish sample surface, a two-dimensional cross-section cutting design is performed on the three-dimensional global noise distribution space model of the fish sample surface to generate a two-dimensional cross-section cutting design plan for the fish sample surface;

[0025] According to the two-dimensional cross-section cutting design plan of the fish sample surface, the three-dimensional noise distribution annotated point cloud data on the fish sample surface is subjected to two-dimensional cross-section segmentation processing to obtain the noise distribution annotated point cloud data under the two-dimensional segmentation cross-section of each fish sample.

[0026] Further, the step S135 described in which the noise point spatial distribution sparse matrix of the fish sample under each fish sample two-dimensional segmentation section is used to perform noise point sparse filtering on the noise point distribution annotated point cloud data under the corresponding fish sample two-dimensional segmentation section includes the following steps:

[0027] Based on the sparse matrix of the spatial distribution of the noise points of the fish samples under the two-dimensional segmentation section of each fish sample, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to point-by-point zeroing filtering processing to obtain the initial sparse filtering point cloud data of the noise point distribution annotation under the two-dimensional segmentation section of each fish sample;

[0028] The local noise density is verified and estimated on the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample, so as to obtain the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample;

[0029] Based on the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample, the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample is adaptively noise suppressed to obtain the fish sample noise sparse filtered point cloud data under the two-dimensional segmentation section of each fish sample.

[0030] Further, step S14 includes the following steps:

[0031] Step S141: performing a fish sample point cloud surface structure topology analysis on the fish sample at a standardized scanning position and posture based on the fish sample surface three-dimensional denoised sparse point cloud data to obtain a fish sample three-dimensional point cloud surface topology structure model;

[0032] Step S142: performing detail contrast enhancement processing on the three-dimensional point cloud surface topological structure model of the fish sample to obtain a three-dimensional surface structure detail contrast enhancement model of the fish sample;

[0033] Step S143: performing three-dimensional texture feature analysis on the three-dimensional surface structure detail contrast enhancement model of the fish sample to obtain three-dimensional texture data of the fish sample surface.

[0034] Further, step S15 includes the following steps:

[0035] Step S151: Preliminarily establish a three-dimensional scanning model of the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface, and obtain a preliminary three-dimensional scanning mapping model of the fish sample;

[0036] Step S152: performing model surface smoothing processing on the preliminary mapping model of the three-dimensional scanning of the fish sample to obtain a smoothed surface structure model of the three-dimensional scanning of the fish sample;

[0037] Step S153: performing texture feature vector conversion on the three-dimensional texture data of the fish sample surface to obtain the three-dimensional texture feature vector of the fish sample surface;

[0038] Step S154: reconstructing the three-dimensional scanned surface structure smoothing model of the fish sample based on the three-dimensional texture feature vector of the fish sample surface to obtain a three-dimensional scanned reconstructed model of the fish sample.

[0039] Further, step S2 includes the following steps:

[0040] Step S21: performing fish area grid subdivision on the fish sample three-dimensional scanning and reconstruction model to obtain a fish sample three-dimensional scanning area subdivision grid model;

[0041] Step S22: determining the key grid points of regional characteristic attributes for each fish area grid unit in the fish sample three-dimensional scanning area subdivision grid model, so as to obtain the fish sample head area grid starting point, the fish sample head area grid end point, the fish sample trunk area grid starting point, the fish sample trunk area grid end point, the fish sample tail fin area grid starting point and the fish sample tail fin area grid end point;

[0042] Step S23: performing mesh segmentation of the fish sample head region on the fish sample three-dimensional scanning region subdivision mesh model based on the fish sample head region mesh starting point and the fish sample head region mesh end point to obtain the fish sample head three-dimensional mesh segmentation sub-region;

[0043] Step S24: performing mesh segmentation of the fish sample trunk region on the fish sample 3D scanning region subdivision mesh model based on the fish sample trunk region mesh starting point and the fish sample trunk region mesh ending point to obtain the fish trunk 3D mesh segmentation sub-region;

[0044] Step S25: performing fish sample tail fin region grid segmentation on the fish sample three-dimensional scanning region subdivision grid model based on the fish sample tail fin region grid starting point and the fish sample tail fin region grid end point to obtain the fish tail fin three-dimensional grid segmentation sub-region.

[0045] Further, step S3 includes the following steps:

[0046] Step S31: measuring the body length of the fish body in the different three-dimensional network segmentation sub-regions of the fish samples to obtain the body length of the fish body in the different fish segmentation sub-regions;

[0047] Step S32: measuring the body width of the fish body in the different three-dimensional grid segmentation sub-regions of the fish sample to obtain the body width of the fish body in the different fish segmentation sub-regions;

[0048] Step S33: measuring the body height of fish in different fish sample three-dimensional grid segmentation sub-regions to obtain the body height of fish in different fish segmentation sub-regions;

[0049] Step S34: performing recognition and analysis of the external contour shape of the fish body part of the sub-regions of the three-dimensional grid segmentation sub-regions of different fish samples to obtain the external contour shape of the fish body part of the different fish segmentation sub-regions;

[0050] Step S35: Integrate the fish phenotypic characteristics of the fish body length, fish body width, fish body height and fish body external contour shape of different fish segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions.

[0051] Further, step S4 includes the following steps:

[0052] Step S41: Calculating the volume of the fish body according to the length, width and height of the fish body in different fish segmentation sub-regions, and obtaining volume parameters of the fish body in different fish segmentation sub-regions;

[0053] Step S42: Calculating the fish body surface area according to the length and height of the fish body parts in different fish segmentation sub-regions, and obtaining the fish body surface area parameters of the different fish segmentation sub-regions;

[0054] Step S43: calculating the fish body shape factors of the external contours of the fish body parts of the different fish segmentation sub-regions based on the fish body part volume parameters and the fish body part surface area parameters of the different fish segmentation sub-regions, to obtain the fish body part shape factors of the different fish segmentation sub-regions;

[0055] Step S44: merging the fish body volume parameters, fish body surface area parameters and fish body shape factors of different fish segmentation sub-regions into fish phenotypic parameters to obtain fish phenotypic parameter sets of different fish segmentation sub-regions;

[0056] Step S45: performing a graphical visualization conversion on the fish phenotypic parameter sets of different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters of different fish segmentation sub-regions.

[0057] Beneficial effects of the present invention:

[0058] The method for automatically extracting fish phenotypes based on a three-dimensional scanning model proposed in the present invention has the beneficial effect of accurately capturing the surface details of fish samples and generating highly dense point cloud data by using a three-dimensional scanning device to perform three-dimensional scanning processing on fish samples, and generating highly dense point cloud data. The key to this process is to provide detailed surface information of fish samples, which provides basic data for subsequent analysis and modeling. Through high-precision three-dimensional scanning, the true morphology and structure of the samples can be obtained, thereby improving the accuracy and reliability of the data. By using the three-dimensional point cloud data of the fish sample surface after denoising, the three-dimensional texture feature analysis of the fish sample is performed. The key to this step is to combine the surface geometric information with the texture information, so as to more comprehensively describe the surface features of the sample. The three-dimensional texture data not only provides the surface details of the sample, but also can reveal the changes in texture, roughness and other visual features, which is of great significance for more detailed biological research, sample classification and feature extraction. The texture feature data can make the subsequent model reconstruction more realistic and delicate, thereby improving the expressiveness and practicality of the model. At the same time, the 3D scanning model of the fish sample is reconstructed based on the 3D point cloud data of the fish sample surface and the 3D texture data of the fish sample surface. The key to this step is to generate a complete, real and detailed 3D model, which integrates geometric information and texture features. The reconstructed model not only accurately reflects the geometric shape of the fish sample, but also can show its surface texture, making the model more realistic and practical. The final 3D scanning reconstruction model can provide a comprehensive data basis for further analysis and research, and improve the application value and actual effect of the model. Secondly, by performing grid segmentation of the fish sample head area on the 3D scanning reconstruction model of the fish sample, the fish head can be extracted from the overall 3D scanning model to form an independent 3D grid segmentation sub-area of ​​the head. This segmentation can clarify the boundary of the head area, and then conduct a more detailed analysis of it. Head segmentation also helps to deeply study the various components of the fish head, such as the mouth structure, eye position, gill distribution, etc., thereby providing high-precision model data support for various subsequent applications and research. By performing mesh segmentation on the trunk area of ​​the fish sample 3D scan reconstruction model, an independent trunk 3D mesh segmentation sub-area can be obtained, which provides an accurate model for further structural analysis. The trunk area of ​​fish usually includes important organs and structures, such as the spine, internal organs and muscle tissue, which play a key role in the movement and physiological functions of fish. Through segmentation, the detailed structure of the trunk area can be studied independently. The mesh segmentation of the trunk area not only improves the level of detail of the model, but also provides a more accurate data basis for related research and applications.In addition, by performing grid segmentation on the caudal fin region of the fish sample 3D scanning reconstruction model, an independent caudal fin 3D grid segmentation sub-region can be extracted from the overall 3D scanning model. The caudal fin is an important locomotion organ of fish, responsible for propulsion and directional control. Understanding the structural characteristics of the caudal fin is crucial to studying the locomotion ability and swimming efficiency of fish. Through independent segmentation of the caudal fin region, the morphology, structure and function of the caudal fin can be analyzed in detail, including the shape, size, arrangement of fin rays and movement mechanism of the caudal fin. This fine segmentation can also be used for functional simulation and optimization design of the caudal fin, and can provide basic data guarantee for subsequent fish phenotypic parameter data extraction, so as to better identify and analyze subtle changes in various parts of the fish body. Then, by performing sub-region fish phenotypic feature analysis on different 3D grid segmentation sub-regions of fish samples, specific phenotypic feature data sets of each sub-region can be obtained, such as body length, body width, body height and external contour shape. These feature data sets are crucial for in-depth understanding of the biological characteristics, population variation and ecological adaptability of fish. Through the detailed analysis of these phenotypic characteristics, a comprehensive fish phenotypic characteristic data set can be formed. This analysis and processing process can fully reflect the comprehensive morphological characteristics of different fish samples, thereby providing basic data guarantee for the subsequent fish phenotypic parameter calculation process. Finally, by calculating the fish phenotypic parameters of the fish phenotypic characteristic data set of different fish segmentation sub-regions and converting them through chart visualization, the complex phenotypic data can be converted into a visualization form that is easy to understand and analyze. This visualization method enables researchers to intuitively observe the phenotypic parameters of different fish segmentation sub-regions, such as volume, surface area, shape factor and other phenotypic parameters, thereby revealing potential morphological laws and trends. Visualization charts not only help simplify the data analysis process, but also effectively display research results, facilitating sharing and discussion with others. This step ultimately enhances the interpretability and dissemination of the data, thereby improving the transparency and accuracy of fish phenotypic data extraction, and providing a powerful tool for fish farming decision support. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0060] Figure 1 It is a schematic diagram of the steps of the method for automatically extracting fish phenotypes based on a three-dimensional scanning model of the present invention;

[0061] Figure 2 for Figure 1 Detailed step flow diagram of step S1;

[0062] Figure 3 for Figure 2 Detailed step flow chart of step S13 in FIG. DETAILED DESCRIPTION

[0063] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0064] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0065] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0066] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for automatically extracting fish phenotypes based on a three-dimensional scanning model, the method comprising the following steps:

[0067] Step S1: performing three-dimensional scanning processing on the fish sample by a three-dimensional scanning device to obtain three-dimensional point cloud data of the fish sample surface and three-dimensional texture data of the fish sample surface; performing three-dimensional scanning model reconstruction on the fish sample based on the three-dimensional point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstruction model of the fish sample;

[0068] Step S2: performing fish region grid segmentation processing on the fish sample 3D scanning and reconstruction model to obtain different fish sample 3D grid segmentation sub-regions, wherein the different fish sample 3D grid segmentation sub-regions include a fish head 3D grid segmentation sub-region, a fish trunk 3D grid segmentation sub-region, and a fish tail fin 3D grid segmentation sub-region;

[0069] Step S3: performing sub-region fish phenotypic characteristic analysis on different fish sample three-dimensional grid segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions, wherein the fish phenotypic characteristic data sets of different fish segmentation sub-regions include the body length, body width, body height and external contour shape of the fish body part of the different fish segmentation sub-regions;

[0070] Step S4: Calculate fish phenotypic parameters for fish phenotypic feature data sets of different fish segmentation sub-regions to obtain fish phenotypic parameter sets for different fish segmentation sub-regions; perform graphical visualization conversion on the fish phenotypic parameter sets for different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters for different fish segmentation sub-regions.

[0071] In the embodiment of the present invention, please refer to Figure 1 As shown, it is a schematic diagram of the steps of the method for automatically extracting fish phenotypes based on a three-dimensional scanning model of the present invention. In this example, the method for automatically extracting fish phenotypes based on a three-dimensional scanning model includes the following steps:

[0072] Step S1: performing three-dimensional scanning processing on the fish sample by a three-dimensional scanning device to obtain three-dimensional point cloud data of the fish sample surface and three-dimensional texture data of the fish sample surface; performing three-dimensional scanning model reconstruction on the fish sample based on the three-dimensional point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstruction model of the fish sample;

[0073] In an embodiment of the present invention, a fish sample is scanned and processed by using a high-resolution three-dimensional scanning device (such as a structured light scanner or a laser scanner). By emitting a light beam and receiving a reflected light signal, the device captures subtle shape changes on the sample surface. The collected data is converted into three-dimensional point cloud data through a computer system to represent the surface features of the fish sample, thereby obtaining three-dimensional point cloud data of the fish sample surface. At the same time, the obtained three-dimensional point cloud data of the fish sample surface is filtered by a sparse matrix of noise points, so as to input the point cloud data into point cloud processing software (such as MeshLab or CloudCompare), and use a sparse filtering algorithm (such as statistical outlier removal or Gaussian filtering) to identify and remove noise points in the point cloud. During the processing, the algorithm removes outliers according to the local density and spatial distribution characteristics of the point cloud, so as to obtain clearer and more accurate three-dimensional point cloud data of the fish sample surface. By combining the three-dimensional point cloud data of the fish sample surface obtained after denoising, a texture analysis tool (such as Texture Mapping algorithm) is used to perform statistical analysis of the three-dimensional texture features of the corresponding fish samples, so as to map the point cloud data to the three-dimensional model of the fish sample, and then extract surface texture features, such as color, brightness and surface texture details. These data include the texture distribution and detail features of the fish sample, so as to obtain the three-dimensional texture data of the fish sample surface. Then, by combining the 3D point cloud data of the fish sample surface obtained after denoising and the 3D texture data of the fish sample surface obtained by texture feature analysis, a 3D scanning model of the fish sample corresponding to the standardized scanning position and posture is reconstructed using 3D modeling software (such as Blender or Meshlab). By combining the denoised sparse point cloud data and texture data, a 3D scanning reconstruction model of the fish sample is generated through a reconstruction algorithm in the modeling software (such as a mesh reconstruction algorithm or a surface fitting algorithm). During the reconstruction process, the software converts the point cloud data into a 3D mesh model and applies the texture data to the mesh, thereby generating a 3D scanning reconstruction model containing detailed surface features, and finally obtaining a 3D scanning reconstruction model of the fish sample.

[0074] Step S2: performing fish region grid segmentation processing on the fish sample 3D scanning and reconstruction model to obtain different fish sample 3D grid segmentation sub-regions, wherein the different fish sample 3D grid segmentation sub-regions include a fish head 3D grid segmentation sub-region, a fish trunk 3D grid segmentation sub-region, and a fish tail fin 3D grid segmentation sub-region;

[0075] In an embodiment of the present invention, a computer-aided design (CAD) software is used to subdivide the regional grid of the previously reconstructed three-dimensional scanning reconstruction model of the fish sample. Specifically, by setting the parameters of the grid subdivision, it is ensured that the surface of the fish body is uniformly subdivided so as to improve the accuracy of the model, and the original coarse grid is decomposed into more fine grid units. In addition, a grid analysis tool is applied to each fish area grid unit in the three-dimensional scanning reconstruction model of the fish sample after the grid subdivision, and specific feature attributes (including feature attributes such as fish head, trunk and tail fin) are set to filter conditions to identify the key grid points of the corresponding fish body area. These points represent the boundaries of the fish head, trunk and tail fin in the grid, that is, the corresponding The starting and ending points of the fish body are identified by using an automated tool to identify the points, and the starting and ending points of the fish head area mesh, the trunk area mesh, and the tail fin area mesh are identified therefrom. The head area of ​​the fish sample three-dimensional scanning reconstruction model is segmented by using a mesh segmentation tool in combination with the previously determined starting and ending points of the head area mesh. In the specific operation, a polygonal mesh is automatically generated to cover the head area by inputting the starting and ending point coordinates in the software. The segmentation algorithm such as Voronoi segmentation or boundary-based segmentation separates the head area from the overall model to form an independent three-dimensional head mesh sub-region, thereby obtaining the fish head three-dimensional mesh segmentation sub-region. Similarly, the trunk area of ​​the fish sample three-dimensional scanning reconstruction model is segmented by using a mesh segmentation tool in combination with the previously determined trunk area mesh starting and ending points. In the specific operation, a polygonal mesh is automatically generated to cover the trunk area by inputting the starting and ending point coordinates in the software. The segmentation algorithm such as Voronoi segmentation or boundary-based segmentation separates the trunk area from the overall model to form an independent three-dimensional trunk mesh sub-region, thereby obtaining the fish trunk three-dimensional mesh segmentation sub-region. Then, similarly, the mesh segmentation tool is used to segment the caudal fin area of ​​the fish sample 3D scan area subdivision mesh model by combining the previously determined caudal fin area mesh start and end points. In the specific operation, a polygonal mesh is automatically generated to cover the caudal fin area by entering the start and end point coordinates in the software. Segmentation algorithms such as Voronoi segmentation or boundary-based segmentation separate the caudal fin area from the overall model to form an independent caudal fin 3D mesh sub-area, and finally the fish caudal fin 3D mesh segmentation sub-area is obtained.

[0076] Step S3: performing sub-region fish phenotypic characteristic analysis on different fish sample three-dimensional grid segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions, wherein the fish phenotypic characteristic data sets of different fish segmentation sub-regions include the body length, body width, body height and external contour shape of the fish body part of the different fish segmentation sub-regions;

[0077] In an embodiment of the present invention, the body length of the fish body is measured by using a measuring tool on different previously divided three-dimensional network segmentation sub-regions of fish samples (including sub-regions such as fish head, trunk and tail fin), so as to measure and record the corresponding body length data, that is, to measure the body length of the corresponding head, trunk and tail fin. Similarly, the body width of the fish body is measured by using a measuring tool on different previously divided three-dimensional grid segmentation sub-regions of fish samples (including sub-regions such as fish head, trunk and tail fin), so as to measure and record the corresponding body width data, that is, to measure the maximum width of the corresponding head, trunk and tail fin. Similarly, the body height of the fish body is measured by using a measuring tool on different previously divided three-dimensional grid segmentation sub-regions of fish samples (including sub-regions such as fish head, trunk and tail fin), so as to measure and record the corresponding body height data, that is, to measure the maximum height of the corresponding head, trunk and tail fin, and through Image processing software (such as MATLAB's image processing toolbox or OpenCV) is used to process the three-dimensional grid segmentation sub-regions corresponding to different fish samples. By applying shape recognition algorithms (such as edge detection and contour extraction), the contour shape of the fish body is extracted. Then, the previously measured data of body length, body width, body height and external contour shape are integrated into the corresponding fish phenotypic characteristic data set. By inputting the measurement results in a data processing platform (such as Excel or Python's Pandas library), the detailed data of body length, body width, body height and external contour shape of each segmentation sub-region of each fish sample are organized into a table format. Through data integration operations, all data are integrated into a comprehensive data set for analysis, which includes all body shape characteristics of different fish segmentation sub-regions, and finally the fish phenotypic characteristic data sets of different fish segmentation sub-regions are obtained.

[0078] Step S4: Calculate fish phenotypic parameters for fish phenotypic feature data sets of different fish segmentation sub-regions to obtain fish phenotypic parameter sets for different fish segmentation sub-regions; perform graphical visualization conversion on the fish phenotypic parameter sets for different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters for different fish segmentation sub-regions.

[0079] In an embodiment of the present invention, the body length, body width and body height of the fish body parts of different fish segmentation sub-regions are obtained by using the fish phenotypic characteristic data set obtained by previous analysis to quantitatively calculate the volume of each fish body part, and the volume calculation formula is V=L*W*H, wherein L is the body length, W is the body width, and H is the body height. The calculated volume parameters can be used to evaluate the size and space occupancy of the fish body. Similarly, the body length and body height of the fish body parts of different fish segmentation sub-regions obtained by previous analysis are used to quantitatively calculate the surface area of ​​each fish body part, and the surface area calculation formula is A=2*L*H to calculate the surface area of ​​the corresponding fish body part. The shape factor of the corresponding fish body part external contour shape is quantitatively calculated by combining the fish body part volume parameter and the fish body part surface area parameter obtained by previous quantitative calculation, wherein the shape factor is generally calculated using the formula , is the volume, The calculated shape factor reflects the shape characteristics and surface smoothness of the fish body. At the same time, the previously calculated volume, surface area and shape factor data of different fish segmentation sub-regions are integrated into a data table by using data processing tools (such as Excel or Pandas library in Python). Each row represents a segmentation sub-region, and each column contains the corresponding volume, surface area and shape factor values, thereby obtaining the fish phenotypic parameter sets of different fish segmentation sub-regions. Then, various charts, including bar charts, scatter plots and three-dimensional surface charts, are generated by using data visualization tools (such as Tableau, Matplotlib or Excel) to generate various charts of parameters such as volume, surface area and shape factor. Through these charts, the morphological characteristics of different fish segmentation sub-regions can be intuitively displayed, which is convenient for further analysis and comparison, thereby converting the fish characteristic parameter data into easy-to-understand visual information, and finally drawing and generating the fish characteristic parameter visualization phenotype of different fish segmentation sub-regions.

[0080] Further, step S1 includes the following steps:

[0081] Step S11: optimizing and adjusting the scanning position and posture of the fish sample by a high-precision positioning device and an automatic adjustment mechanism to obtain the fish sample in a standardized scanning position and posture;

[0082] Step S12: performing three-dimensional scanning processing on the fish sample in the standardized scanning position and posture by a three-dimensional scanning device to obtain three-dimensional point cloud data of the surface of the fish sample;

[0083] Step S13: performing noise sparse filtering on the three-dimensional point cloud data of the fish sample surface to obtain three-dimensional denoised sparse point cloud data of the fish sample surface;

[0084] Step S14: performing a three-dimensional texture feature analysis on the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface to obtain three-dimensional texture data of the fish sample surface;

[0085] Step S15: reconstructing a three-dimensional scanning model of the fish sample in a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstructed model of the fish sample.

[0086] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in FIG. 1 , in this embodiment, step S1 includes the following steps:

[0087] Step S11: optimizing and adjusting the scanning position and posture of the fish sample by a high-precision positioning device and an automatic adjustment mechanism to obtain the fish sample in a standardized scanning position and posture;

[0088] In an embodiment of the present invention, a high-precision positioning device (such as a laser positioning system) and an automatic adjustment mechanism (such as a robotic arm) are used to optimize the scanning position and posture of the corresponding fish sample, so that the position of the fish sample is determined by the high-precision positioning device through laser ranging technology, and the data is transmitted to the control system. The control system controls the automatic adjustment mechanism (such as a servo motor) to adjust the position and posture of the fish sample according to the set standardized scanning position and posture (such as a predefined angle and distance) to ensure that the sample is at a standardized angle and distance during the scanning process, and finally the fish sample is obtained in a standardized scanning position and posture.

[0089] Step S12: performing three-dimensional scanning processing on the fish sample in the standardized scanning position and posture by a three-dimensional scanning device to obtain three-dimensional point cloud data of the surface of the fish sample;

[0090] In an embodiment of the present invention, a fish sample in a standardized scanning position and posture is scanned and processed by using a high-resolution three-dimensional scanning device (such as a structured light scanner or a laser scanner). By emitting a light beam and receiving a reflected light signal, the device captures subtle shape changes on the sample surface, and converts the collected data into three-dimensional point cloud data through a computer system to represent the surface features of the fish sample, thereby finally obtaining three-dimensional point cloud data of the fish sample surface.

[0091] Step S13: performing noise sparse filtering on the three-dimensional point cloud data of the fish sample surface to obtain three-dimensional denoised sparse point cloud data of the fish sample surface;

[0092] In an embodiment of the present invention, the acquired three-dimensional point cloud data of the surface of the fish sample is subjected to a noise sparse matrix filtering process, so as to input the point cloud data into a point cloud processing software (such as MeshLab or CloudCompare), and utilize a sparse filtering algorithm (such as statistical outlier removal or Gaussian filtering) to identify and remove noise points in the point cloud. During the processing, the algorithm removes outliers according to the local density and spatial distribution characteristics of the point cloud, thereby obtaining clearer and more accurate three-dimensional point cloud data of the surface of the fish sample, and finally obtaining three-dimensional denoised sparse point cloud data of the surface of the fish sample.

[0093] Step S14: performing a three-dimensional texture feature analysis on the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface to obtain three-dimensional texture data of the fish sample surface;

[0094] In an embodiment of the present invention, a texture analysis tool (such as a Texture Mapping algorithm) is used to perform statistical analysis of the three-dimensional texture features of the fish samples corresponding to the standardized scanning position and posture by combining the three-dimensional denoised sparse point cloud data of the fish sample surface obtained after previous denoising, so as to map the point cloud data to the three-dimensional model of the fish sample, and then extract surface texture features such as color, brightness and surface texture details. These data include the texture distribution and detail features of the fish sample, and finally obtain the three-dimensional texture data of the fish sample surface.

[0095] Step S15: reconstructing a three-dimensional scanning model of the fish sample in a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstructed model of the fish sample.

[0096] In an embodiment of the present invention, a three-dimensional scanning model of the fish sample corresponding to the standardized scanning position and posture is reconstructed using a three-dimensional modeling software (such as Blender or Meshlab) by combining the three-dimensional denoised sparse point cloud data of the fish sample surface obtained after previous denoising and the three-dimensional texture data of the fish sample surface obtained by texture feature analysis, so as to generate a three-dimensional scanning reconstruction model of the fish sample by combining the denoised sparse point cloud data and the texture data and using the reconstruction algorithm in the modeling software (such as a mesh reconstruction algorithm or a surface fitting algorithm). During the reconstruction process, the software converts the point cloud data into a three-dimensional mesh model and applies the texture data to the mesh, thereby generating a three-dimensional scanning reconstruction model containing detailed surface features, and finally obtaining a three-dimensional scanning reconstruction model of the fish sample.

[0097] Further, step S13 includes the following steps:

[0098] Step S131: performing noise point distribution detection and annotation on the three-dimensional point cloud data of the fish sample surface to obtain the three-dimensional noise point distribution annotated point cloud data on the fish sample surface;

[0099] Step S132: performing two-dimensional section segmentation processing on the three-dimensional noise point distribution annotated point cloud data on the surface of the fish sample to obtain the noise point distribution annotated point cloud data under the two-dimensional segmented section of each fish sample;

[0100] Step S133: measuring and calculating the spatial distance between any two fish sample noise points in the noise point distribution annotated point cloud data under each fish sample two-dimensional segmentation section, and obtaining the spatial position distance value between any two fish sample noise points under each fish sample two-dimensional segmentation section;

[0101] Step S134: comparing and judging the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section according to a preset spatial position spacing threshold value; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is greater than or equal to the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is increased by 1; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is less than the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is reduced by 1, until the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is compared and judged, so as to construct a sparse matrix of the spatial distribution of fish sample noise points under each fish sample two-dimensional segmentation section;

[0102] Step S135: Based on the sparse matrix of the spatial distribution of fish sample noise points under each two-dimensional segmentation section, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to noise point sparse filtering processing to obtain the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section of the fish sample; the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section is subjected to cross-section three-dimensional integration and restoration processing to obtain the three-dimensional denoised sparse point cloud data of the fish sample surface.

[0103] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 2 FIG. 1 is a schematic diagram of a detailed step flow chart of step S13 in FIG. 1 . In this embodiment, step S13 includes the following steps:

[0104] Step S131: performing noise point distribution detection and annotation on the three-dimensional point cloud data of the fish sample surface to obtain the three-dimensional noise point distribution annotated point cloud data on the fish sample surface;

[0105] In an embodiment of the present invention, the three-dimensional point cloud data of the fish sample surface previously obtained by three-dimensional scanning is imported into point cloud processing software (such as CloudCompare), and by applying a noise point detection algorithm (such as statistical outlier detection), each point cloud in the three-dimensional point cloud data of the fish sample surface is analyzed to identify the corresponding noise points, and these noise points are marked using a point cloud annotation tool to show the distribution of noise points on the fish sample surface, and finally the three-dimensional noise point distribution annotated point cloud data of the fish sample surface is obtained.

[0106] Step S132: performing two-dimensional section segmentation processing on the three-dimensional noise point distribution annotated point cloud data on the surface of the fish sample to obtain the noise point distribution annotated point cloud data under the two-dimensional segmented section of each fish sample;

[0107] In an embodiment of the present invention, two-dimensional cross-section segmentation is performed on the three-dimensional noise distribution annotated point cloud data on the surface of the fish sample obtained after the noise points are previously annotated, so as to import the annotated point cloud data into a three-dimensional point cloud processing platform, and set the two-dimensional cross-section cutting plane according to the cutting position and direction of each two-dimensional cross-section. The cutting is usually performed based on the set cutting plane parameters (such as height or angle). Through the point cloud cutting function, the three-dimensional point cloud data is segmented into point cloud data sets under multiple two-dimensional cross-sections, and the noise distribution annotated point cloud data under each cross-section is extracted using a data processing tool (such as MeshLab or CloudCompare), so as to finally obtain the noise distribution annotated point cloud data under each two-dimensional segmented cross-section of each fish sample.

[0108] Step S133: measuring and calculating the spatial distance between any two fish sample noise points in the noise point distribution annotated point cloud data under each fish sample two-dimensional segmentation section, and obtaining the spatial position distance value between any two fish sample noise points under each fish sample two-dimensional segmentation section;

[0109] In an embodiment of the present invention, the spatial distance between any two noise points is measured and calculated by annotating the point cloud data of the noise point distribution under each two-dimensional segmentation section, so as to utilize the distance measurement tool in the point cloud processing software to calculate the spatial position distance between each pair of noise points, and use the Euclidean distance formula to calculate the distance values ​​between all noise points one by one, and record each distance value, and finally quantify and calculate the spatial position distance value between any two noise points of the fish samples under the two-dimensional segmentation section of each fish sample.

[0110] Step S134: comparing and judging the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section according to a preset spatial position spacing threshold value; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is greater than or equal to the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is increased by 1; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is less than the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is reduced by 1, until the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is compared and judged, so as to construct a sparse matrix of the spatial distribution of fish sample noise points under each fish sample two-dimensional segmentation section;

[0111] In an embodiment of the present invention, a preset spatial position spacing threshold is used to compare the spatial position spacing value between any two fish sample noise points under each two-dimensional segmentation section of each fish sample obtained by previous quantitative calculation, so as to set a spatial position spacing threshold, and compare it with the actual spacing value of each pair of noise points. When the spacing value between any two noise points is greater than or equal to the threshold, the sparsity at the spatial position of the corresponding points is increased by 1; when the spacing value is less than the threshold, the sparsity is reduced by 1. This operation traverses all noise point pairs, thereby constructing a sparse matrix of noise point distribution under each two-dimensional section, and finally obtaining a sparse matrix of fish sample noise point spatial distribution under each two-dimensional segmentation section of the fish sample.

[0112] Step S135: Based on the sparse matrix of the spatial distribution of fish sample noise points under each two-dimensional segmentation section, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to noise point sparse filtering processing to obtain the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section of the fish sample; the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section is subjected to cross-section three-dimensional integration and restoration processing to obtain the three-dimensional denoised sparse point cloud data of the fish sample surface.

[0113] In an embodiment of the present invention, a noise sparse filtering process is performed on the noise distribution annotated point cloud data under the corresponding two-dimensional segmentation section of the fish sample by combining the previously constructed sparse matrix of the spatial distribution of noise points of the fish sample under each two-dimensional segmentation section of the fish sample, so that the sparse matrix is ​​applied to the point cloud data under each two-dimensional section, and the size of the specific numerical value of the sparsity of the noise point at the specific distribution position is determined according to the corresponding sparse matrix to determine the distribution of the surrounding noise points, that is, whether the distribution of the surrounding noise points is closer or farther, and filtering processing is performed point by point around it by adopting a corresponding filtering algorithm (such as a Gaussian filter) to eliminate the influence of the surrounding noise points on the overall point cloud data, and the sparse value of the noise point is decreased or increased to zero point by point, and the corresponding noise point is removed from the point cloud data set, so as to obtain the sparse filtered point cloud data of the fish sample noise under each two-dimensional segmentation section of the fish sample. At the same time, the processed two-dimensional cross-sectional point cloud data is restored through three-dimensional integration, so that each cross-sectional data is synthesized into a complete three-dimensional point cloud by using a point cloud integration algorithm (such as point cloud stitching). The obtained three-dimensional denoised sparse point cloud data shows the surface point cloud data of the fish sample after noise processing, and finally the three-dimensional denoised sparse point cloud data of the fish sample surface is reintegrated.

[0114] Further, step S132 includes the following steps:

[0115] A global noise distribution spatial model is established for the annotated point cloud data of the three-dimensional noise distribution on the surface of the fish sample, and a three-dimensional global noise distribution spatial model is obtained on the surface of the fish sample;

[0116] In an embodiment of the present invention, a global noise distribution spatial model is established by using point cloud data with three-dimensional noise distribution on the surface of fish samples. First, the point cloud data is imported into a point cloud processing platform (such as CloudCompare), and the point cloud data is subjected to spatial distribution statistical analysis by adopting global noise distribution modeling technology to construct a corresponding global noise distribution model. The model is based on the noise distribution information in the point cloud data, thereby describing the distribution of noise on the entire fish sample surface, and finally obtaining a three-dimensional global noise distribution spatial model on the fish sample surface.

[0117] Preferably, the three-dimensional morphology of the fish sample and the noise distribution characteristics are identified and analyzed on the three-dimensional global noise distribution spatial model on the surface of the fish sample to obtain the three-dimensional distribution morphology of the fish sample surface and the three-dimensional noise distribution characteristics on the surface of the fish sample;

[0118] In an embodiment of the present invention, a morphological analysis tool (such as Matplotlib or Paraview) is used to identify and analyze the morphology and noise distribution characteristics of the corresponding three-dimensional global noise distribution spatial model of the fish sample surface, so as to extract the three-dimensional morphological features through a surface mesh generation algorithm, and use a noise distribution analysis algorithm (such as heat map analysis) to identify the spatial distribution characteristics of the noise, thereby obtaining a detailed description of the three-dimensional distribution morphology of the fish sample surface and the noise distribution characteristics, including the concentrated area and distribution density of the noise, and finally obtaining the three-dimensional distribution morphology of the fish sample surface and the three-dimensional noise distribution characteristics of the fish sample surface.

[0119] Preferably, based on the three-dimensional distribution morphology of the fish sample surface and the three-dimensional noise distribution characteristics of the fish sample surface, a two-dimensional cross-section cutting design is performed on the three-dimensional global noise distribution space model of the fish sample surface to generate a two-dimensional cross-section cutting design plan for the fish sample surface;

[0120] In an embodiment of the present invention, a two-dimensional section cutting design is performed on the corresponding three-dimensional global noise distribution space model of the fish sample surface by combining the three-dimensional distribution morphology of the fish sample surface obtained by previous analysis and the three-dimensional noise distribution characteristics of the fish sample surface. By analyzing the three-dimensional morphology of the model, key section positions are selected for cutting design, and a two-dimensional cutting plan is generated at a specific section position using a cutting tool (such as the section function of MeshLab). The plan describes in detail how to perform two-dimensional section cutting in a three-dimensional model, and also clarifies the cutting position and direction of each two-dimensional section. Finally, a two-dimensional section cutting design plan for the fish sample surface is designed and generated.

[0121] Preferably, according to the two-dimensional section cutting design plan of the fish sample surface, the three-dimensional noise distribution annotated point cloud data on the fish sample surface is subjected to two-dimensional section segmentation processing to obtain the noise distribution annotated point cloud data under each two-dimensional segmented section of the fish sample.

[0122] In an embodiment of the present invention, a two-dimensional cross-section cutting design plan for the surface of a fish sample generated previously is used to segment the corresponding three-dimensional noise distribution annotated point cloud data on the surface of the fish sample into two-dimensional sections, so that the cutting design plan is applied to the three-dimensional noise distribution data, and a cutting algorithm (such as a customized two-dimensional section cutting program) is used for processing. This process cuts the three-dimensional data into multiple two-dimensional sections including noise point distribution according to the design plan, and extracts the noise point distribution annotated data under each section, thereby displaying the noise point distribution in different two-dimensional sections, and finally obtaining the noise point distribution annotated point cloud data under each two-dimensional segmented section of the fish sample.

[0123] Further, the step S135 described in which the noise point spatial distribution sparse matrix of the fish sample under each fish sample two-dimensional segmentation section is used to perform noise point sparse filtering on the noise point distribution annotated point cloud data under the corresponding fish sample two-dimensional segmentation section includes the following steps:

[0124] Based on the sparse matrix of the spatial distribution of the noise points of the fish samples under the two-dimensional segmentation section of each fish sample, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to point-by-point zeroing filtering processing to obtain the initial sparse filtering point cloud data of the noise point distribution annotation under the two-dimensional segmentation section of each fish sample;

[0125] In an embodiment of the present invention, a sparse matrix of the spatial distribution of noise points of fish samples under the two-dimensional segmentation section of each fish sample obtained by previous analysis is combined to perform point-by-point zeroing filtering processing on the noise points at specific distribution positions in the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample, so as to determine the specific numerical value of the sparsity of the noise point position at the specific distribution position according to the corresponding sparse matrix to determine the distribution of the surrounding noise points, that is, whether the distribution of the surrounding noise points is closer or farther, and a corresponding filtering algorithm (such as a Gaussian filter) is used to perform point-by-point filtering processing around it to eliminate the influence of the surrounding noise points on the overall point cloud data, so as to reduce or increase the sparse value of the noise point point to zero point by point, thereby ensuring that the distribution of the noise point on the two-dimensional section is effectively reduced, and finally obtaining the initial sparse filtered point cloud data of the noise point distribution annotation under the two-dimensional segmentation section of each fish sample.

[0126] Preferably, the local noise density is verified and estimated for the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample, so as to obtain the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample;

[0127] In an embodiment of the present invention, a density estimation algorithm (such as density estimation based on k-nearest neighbors) is used to verify and estimate the local noise density of the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample previously obtained after filtering, so as to divide the processed point cloud data into multiple local areas, and verify whether there are corresponding noise points in the point cloud in each local area. If so, the corresponding noise density is quantitatively calculated, and if not, no processing is performed. By comparing the point density in the local area, the distribution pattern and density distribution of the corresponding noise points are determined, thereby generating local noise density distribution estimation data for analyzing the distribution characteristics and density distribution of the noise points that still exist in each area, and finally obtaining the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample.

[0128] Preferably, based on the local noise density distribution estimation data under each two-dimensional segmentation section of each fish sample, the initial sparse filtered point cloud data of the noise distribution annotation under each two-dimensional segmentation section of the fish sample is adaptively noise suppressed to obtain the fish sample noise sparse filtered point cloud data under each two-dimensional segmentation section of the fish sample.

[0129] In an embodiment of the present invention, an adaptive noise suppression algorithm (such as a local density threshold method) is applied to the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample obtained by previous verification and estimation to suppress the noise points still existing in the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of the corresponding fish sample, so as to adjust the weight of each point according to the local density estimation data, and gradually remove the noise points with a density lower than a predetermined threshold. Through this method, it is ensured that the noise points are effectively suppressed and the point cloud data is more accurate, and finally the noise sparse filtered point cloud data of the fish samples under the two-dimensional segmentation section of each fish sample is obtained.

[0130] Further, step S14 includes the following steps:

[0131] Step S141: performing a fish sample point cloud surface structure topology analysis on the fish sample at a standardized scanning position and posture based on the fish sample surface three-dimensional denoised sparse point cloud data to obtain a fish sample three-dimensional point cloud surface topology structure model;

[0132] In an embodiment of the present invention, a topological analysis of the surface structure of a fish sample in a standardized scanning position and posture is performed by combining the three-dimensional denoised sparse point cloud data of the fish sample surface that has been previously denoised, so that a topological analysis of the point cloud data is performed by using point cloud processing software (such as CloudCompare), and surface features in the point cloud are identified by algorithms (such as Voronoi segmentation or K-means clustering), and the corresponding fish sample is converted into a three-dimensional topological structure model based on the point cloud data by utilizing Delaunay triangulation technology. The model shows the surface geometry and structural relationship of the fish sample, accurately reflects the topological features of the fish body surface, and finally obtains a three-dimensional point cloud surface topological structure model of the fish sample.

[0133] Step S142: performing detail contrast enhancement processing on the three-dimensional point cloud surface topological structure model of the fish sample to obtain a three-dimensional surface structure detail contrast enhancement model of the fish sample;

[0134] In an embodiment of the present invention, detail contrast enhancement processing is performed on the three-dimensional point cloud surface topological structure model of the fish sample previously obtained through topological analysis, and the model is imported into a detail enhancement tool (such as MeshLab or 3D-Coat), and a detail enhancement algorithm (such as a high-pass filter or an adaptive detail enhancement filter that enhances local details) is applied to improve the contrast of the model surface details, making the tiny features of the fish surface more obvious. Through the detail enhancement processing, the generated detail contrast enhanced model shows a clearer and finer surface structure, and finally a three-dimensional surface structure detail contrast enhanced model of the fish sample is obtained.

[0135] Step S143: performing three-dimensional texture feature analysis on the three-dimensional surface structure detail contrast enhancement model of the fish sample to obtain three-dimensional texture data of the fish sample surface.

[0136] In an embodiment of the present invention, a statistical analysis of the three-dimensional texture features is performed on a three-dimensional surface structure detail contrast enhancement model of a fish sample obtained after detail enhancement, and the model is imported into a texture analysis tool (such as Blender or Maya) for texture feature extraction, and the surface texture of the model is mapped to a two-dimensional plane using a texture mapping technique (such as UV unfolding), and then a texture feature analysis algorithm (such as principal component analysis or texture histogram analysis based on texture features) is applied to statistically analyze the texture features of the model surface, and finally the three-dimensional texture data of the fish sample surface is obtained.

[0137] Further, step S15 includes the following steps:

[0138] Step S151: Preliminarily establish a three-dimensional scanning model of the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface, and obtain a preliminary three-dimensional scanning mapping model of the fish sample;

[0139] In an embodiment of the present invention, a preliminary three-dimensional scanning model of the fish sample corresponding to the standardized scanning position and posture is initially established by combining the three-dimensional denoised sparse point cloud data of the fish sample surface that has been previously denoised using point cloud processing software (such as CloudCompare), so as to generate a preliminary mapping model through a triangulated meshing algorithm (such as Delaunay triangulated mesh generation). This model represents the overall shape of the fish sample, but the surface details are not processed, and finally a preliminary mapping model of the three-dimensional scan of the fish sample is obtained.

[0140] Step S152: performing model surface smoothing processing on the preliminary mapping model of the three-dimensional scanning of the fish sample to obtain a smoothed surface structure model of the three-dimensional scanning of the fish sample;

[0141] In an embodiment of the present invention, a previously established preliminary mapping model of a three-dimensional scan of a fish sample is imported into a surface smoothing tool, such as MeshLab or Blender, and a smoothing algorithm (such as Laplacian smoothing or Taubin smoothing) is used to smooth the surface of the model to remove noise and irregularities in the mesh. During the smoothing process, each mesh point on the surface of the model is adjusted to reduce surface roughness, thereby ensuring that the generated model is smoother and closer to the actual surface shape of the fish body, and finally a smoothed model of the surface structure of the three-dimensional scan of the fish sample is obtained.

[0142] Step S153: performing texture feature vector conversion on the three-dimensional texture data of the fish sample surface to obtain the three-dimensional texture feature vector of the fish sample surface;

[0143] In an embodiment of the present invention, the three-dimensional texture data of the surface of a fish sample previously obtained through texture analysis is converted into a feature vector, and the texture data is converted into a feature vector by using a feature extraction algorithm (such as principal component analysis PCA or local binary pattern LBP). The feature vector represents the texture characteristics of the fish surface, including color, texture roughness and pattern, and finally the three-dimensional texture feature vector of the fish sample surface is obtained.

[0144] Step S154: reconstructing the three-dimensional scanned surface structure smoothing model of the fish sample based on the three-dimensional texture feature vector of the fish sample surface to obtain a three-dimensional scanned reconstructed model of the fish sample.

[0145] In an embodiment of the present invention, the three-dimensional texture feature vector of the fish sample surface obtained by combining the previous conversion is applied to the three-dimensional scanning surface structure smooth model of the fish sample generated after the previous smoothing process, and the texture feature vector is mapped to the surface of the smoothed model by using a texture mapping tool (such as the texture mapping function of Blender or the UV unfolding of 3ds Max). Through the texture mapping algorithm, the texture features are matched and reconstructed with the model surface according to the feature vector to generate a final three-dimensional scanning and reconstruction model. This model includes the detailed surface texture features of the fish sample and accurately reproduces the actual appearance of the fish body, and finally a three-dimensional scanning and reconstruction model of the fish sample is obtained.

[0146] Further, step S2 includes the following steps:

[0147] Step S21: performing fish area grid subdivision on the fish sample three-dimensional scanning and reconstruction model to obtain a fish sample three-dimensional scanning area subdivision grid model;

[0148] In an embodiment of the present invention, a computer-aided design (CAD) software is used to subdivide the regional grid of the previously reconstructed three-dimensional scanning reconstruction model of the fish sample. Specifically, by setting the parameters of the grid subdivision, it is ensured that the surface of the fish body is uniformly subdivided to improve the accuracy of the model. The subdivision process uses a grid subdivision algorithm, such as the Loop subdivision algorithm, to decompose the original coarse grid into more fine grid units, and finally obtain the fish sample three-dimensional scanning regional subdivision grid model.

[0149] Step S22: determining the key grid points of regional characteristic attributes for each fish area grid unit in the fish sample three-dimensional scanning area subdivision grid model, so as to obtain the fish sample head area grid starting point, the fish sample head area grid end point, the fish sample trunk area grid starting point, the fish sample trunk area grid end point, the fish sample tail fin area grid starting point and the fish sample tail fin area grid end point;

[0150] In an embodiment of the present invention, a grid analysis tool is applied to each fish area grid unit within the grid model of the three-dimensional scanned area of ​​the fish sample after grid subdivision, and the key grid points of the corresponding fish body area are identified by setting specific feature attributes (including feature attributes such as fish head, trunk and tail fin) screening conditions. These points represent the boundaries of the fish head, trunk and tail fin in the grid, that is, the starting and ending points of the corresponding fish body parts, so that point identification is performed by utilizing an automated tool, and the starting and ending points of the fish head area grid, the starting and ending points of the trunk area grid, and the starting and ending points of the tail fin area grid are identified therefrom, and by matching these points with the anatomical structure of the fish body, it is ensured that the boundaries of each area are accurately determined, and finally the starting point of the fish sample head area grid, the end point of the fish sample head area grid, the starting point of the fish sample trunk area grid, the end point of the fish sample trunk area grid, the starting point of the fish sample tail fin area grid, and the end point of the fish sample tail fin area grid are determined.

[0151] Step S23: performing mesh segmentation of the fish sample head region on the fish sample three-dimensional scanning region subdivision mesh model based on the fish sample head region mesh starting point and the fish sample head region mesh end point to obtain the fish sample head three-dimensional mesh segmentation sub-region;

[0152] In an embodiment of the present invention, a mesh segmentation tool is used to segment the head area of ​​the fish sample three-dimensional scanning area subdivided mesh model by combining the previously determined mesh starting point of the fish sample head area and the mesh end point of the fish sample head area. In a specific operation, a polygonal mesh is automatically generated to cover the head area by inputting the start and end point coordinates in the software. A segmentation algorithm such as Voronoi segmentation or boundary-based segmentation separates the head area from the overall model to form an independent three-dimensional mesh sub-area of ​​the head, and finally a three-dimensional mesh segmentation sub-area of ​​the fish head is obtained.

[0153] Step S24: performing mesh segmentation of the fish sample trunk region on the fish sample 3D scanning region subdivision mesh model based on the fish sample trunk region mesh starting point and the fish sample trunk region mesh ending point to obtain the fish trunk 3D mesh segmentation sub-region;

[0154] In an embodiment of the present invention, a mesh segmentation tool is used to segment the trunk area of ​​the fish sample three-dimensional scan area subdivision mesh model by combining the previously determined mesh starting point of the fish sample trunk area and the mesh end point of the fish sample trunk area. In a specific operation, a polygonal mesh is automatically generated to cover the trunk area by inputting the starting point and end point coordinates in the software. A segmentation algorithm such as Voronoi segmentation or boundary-based segmentation separates the trunk area from the overall model to form an independent three-dimensional mesh sub-area of ​​the trunk, and finally a three-dimensional mesh segmentation sub-area of ​​the fish trunk is obtained.

[0155] Step S25: performing fish sample tail fin region grid segmentation on the fish sample three-dimensional scanning region subdivision grid model based on the fish sample tail fin region grid starting point and the fish sample tail fin region grid end point to obtain the fish tail fin three-dimensional grid segmentation sub-region.

[0156] In an embodiment of the present invention, the caudal fin region is segmented by using a grid segmentation tool to subdivide the grid model of the three-dimensional scanning area of ​​the fish sample in combination with the previously determined grid starting point of the caudal fin region of the fish sample and the grid end point of the caudal fin region of the fish sample. In a specific operation, a polygonal grid is automatically generated to cover the caudal fin region by inputting the starting point and end point coordinates in the software. A segmentation algorithm such as Voronoi segmentation or boundary-based segmentation separates the caudal fin region from the overall model to form an independent caudal fin three-dimensional grid sub-region, and finally the fish caudal fin three-dimensional grid segmentation sub-region is obtained.

[0157] Further, step S3 includes the following steps:

[0158] Step S31: measuring the body length of the fish body in the different three-dimensional network segmentation sub-regions of the fish samples to obtain the body length of the fish body in the different fish segmentation sub-regions;

[0159] In an embodiment of the present invention, a measuring tool is used to measure the body length of different previously divided three-dimensional network segmentation sub-regions of fish samples (including sub-regions such as fish head, trunk and tail fin), so as to measure and record the corresponding body length data. This process can use a three-dimensional scanner or a digital measuring tool to measure and record the longest distance from the start to the end position of each fish body sub-region on a computer, that is, to measure the body length of the corresponding fish head, trunk and tail fin, and finally obtain the body length of the fish body sub-regions of different fish segmentation sub-regions.

[0160] Step S32: measuring the body width of the fish body in the different three-dimensional grid segmentation sub-regions of the fish sample to obtain the body width of the fish body in the different fish segmentation sub-regions;

[0161] In an embodiment of the present invention, the body width of the fish body is similarly measured by using a measuring tool to measure the width of the fish body of different previously divided three-dimensional grid segmentation sub-regions of fish samples (including sub-regions such as fish head, trunk and tail fin), so as to measure and record the corresponding body width data, that is, to measure the maximum width of the corresponding fish head, trunk and tail fin, and finally obtain the body width of the fish body in different fish segmentation sub-regions.

[0162] Step S33: measuring the body height of fish in different fish sample three-dimensional grid segmentation sub-regions to obtain the body height of fish in different fish segmentation sub-regions;

[0163] In an embodiment of the present invention, the body height of the fish body is similarly measured by using a measuring tool to measure the different previously divided three-dimensional grid segmentation sub-regions of the fish samples (including sub-regions such as the fish head, trunk and tail fin), so as to measure and record the corresponding body height data, that is, to measure the maximum height of the corresponding fish head, trunk and tail fin, and finally obtain the body height of the fish body in different fish segmentation sub-regions.

[0164] Step S34: performing recognition and analysis of the external contour shape of the fish body part of the sub-regions of the three-dimensional grid segmentation sub-regions of different fish samples to obtain the external contour shape of the fish body part of the different fish segmentation sub-regions;

[0165] In an embodiment of the present invention, the three-dimensional grid segmentation sub-regions corresponding to different fish samples are processed by using image processing software (such as MATLAB's image processing toolbox or OpenCV), and the external contour shape of the fish body is extracted by applying a shape recognition algorithm (such as edge detection and contour extraction), and finally the external contour shape of the fish body part of different fish segmentation sub-regions is obtained.

[0166] Step S35: Integrate the fish phenotypic characteristics of the fish body length, fish body width, fish body height and fish body external contour shape of different fish segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions.

[0167] In an embodiment of the present invention, the data of body length, body width, body height and external contour shape previously measured are integrated into a corresponding fish phenotypic characteristic data set, and the measurement results are input into a data processing platform (such as Excel or Python's Pandas library), and the detailed data of body length, body width, body height and external contour shape of each segmented sub-region of each fish sample are organized into a tabular format. Through the data integration operation, all the data are integrated into a comprehensive data set for analysis, which includes all the body shape characteristics of different fish segmented sub-regions, and finally the fish phenotypic characteristic data sets of different fish segmented sub-regions are obtained.

[0168] Further, step S4 includes the following steps:

[0169] Step S41: Calculating the volume of the fish body according to the length, width and height of the fish body in different fish segmentation sub-regions, and obtaining volume parameters of the fish body in different fish segmentation sub-regions;

[0170] In an embodiment of the present invention, the volume of each fish body part is quantitatively calculated by using the body length, body width and body height of different fish segmentation sub-regions obtained in the previous analysis, and the volume calculation formula is V=L*W*H, wherein L is the body length, W is the body width, and H is the body height. The calculated volume parameters can be used to evaluate the size and space occupancy of the fish body, and finally the volume parameters of the fish body parts in different fish segmentation sub-regions are obtained.

[0171] Step S42: Calculating the fish body surface area according to the length and height of the fish body parts in different fish segmentation sub-regions, and obtaining the fish body surface area parameters of the different fish segmentation sub-regions;

[0172] In an embodiment of the present invention, the surface area of ​​each fish body part is similarly calculated by using the body length and height of the fish body part of different fish segmentation sub-regions obtained in the previous analysis, so as to calculate the surface area of ​​the corresponding fish body part using the surface area calculation formula A=2*L*H. The calculated surface area parameters are helpful to analyze the complexity and characteristics of the fish body surface, and finally the surface area parameters of the fish body parts in different fish segmentation sub-regions are obtained.

[0173] Step S43: calculating the fish body shape factors of the external contours of the fish body parts of the different fish segmentation sub-regions based on the fish body part volume parameters and the fish body part surface area parameters of the different fish segmentation sub-regions, to obtain the fish body part shape factors of the different fish segmentation sub-regions;

[0174] In the embodiment of the present invention, the shape factor of the corresponding external contour of the fish body part is quantitatively calculated by combining the fish body part volume parameters and the fish body part surface area parameters of the different fish segmentation sub-regions obtained by previous quantitative calculation, wherein the calculation of the shape factor is generally performed using the formula , is the volume, The calculated shape factor reflects the shape characteristics and surface smoothness of the fish body, and finally the fish body shape factors of different fish segmentation sub-regions are obtained.

[0175] Step S44: merging the fish body volume parameters, fish body surface area parameters and fish body shape factors of different fish segmentation sub-regions into fish phenotypic parameters to obtain fish phenotypic parameter sets of different fish segmentation sub-regions;

[0176] In an embodiment of the present invention, the volume, surface area and shape factor data of different fish segmentation sub-regions previously calculated are integrated into a data table by using a data processing tool (such as Excel or the Pandas library in Python), where each row represents a segmentation sub-region and each column contains the corresponding volume, surface area and shape factor values, ultimately obtaining a set of fish phenotypic parameters for different fish segmentation sub-regions.

[0177] Step S45: performing a graphical visualization conversion on the fish phenotypic parameter sets of different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters of different fish segmentation sub-regions.

[0178] In an embodiment of the present invention, various charts, including bar charts, scatter plots and three-dimensional surface plots, are generated from parameters such as volume, surface area and shape factor by using data visualization tools (such as Tableau, Matplotlib or Excel). Through these charts, the morphological characteristics of different fish segmentation sub-regions can be intuitively displayed to facilitate further analysis and comparison, thereby converting fish characteristic parameter data into easy-to-understand visual information, and finally drawing and generating fish characteristic parameter visualization phenotypes of different fish segmentation sub-regions.

[0179] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0180] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for automatically extracting fish phenotypes based on a three-dimensional scanning model, characterized in that: The following steps are involved: Step S1: performing three-dimensional scanning processing on the fish sample by a three-dimensional scanning device to obtain three-dimensional point cloud data of the fish sample surface and three-dimensional texture data of the fish sample surface; performing three-dimensional scanning model reconstruction on the fish sample based on the three-dimensional point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstruction model of the fish sample; wherein step S1 includes the following steps: Step S11: optimizing and adjusting the scanning position and posture of the fish sample by a high-precision positioning device and an automatic adjustment mechanism to obtain the fish sample in a standardized scanning position and posture; Step S12: performing three-dimensional scanning processing on the fish sample in the standardized scanning position and posture by a three-dimensional scanning device to obtain three-dimensional point cloud data of the surface of the fish sample; Step S13: performing noise sparse filtering on the three-dimensional point cloud data of the fish sample surface to obtain three-dimensional denoised sparse point cloud data of the fish sample surface; Step S14: performing a three-dimensional texture feature analysis on the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface to obtain three-dimensional texture data of the fish sample surface; Step S15: reconstructing a three-dimensional scanning model of the fish sample at the standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface and the three-dimensional texture data of the fish sample surface to obtain a three-dimensional scanning reconstruction model of the fish sample; wherein step S15 includes the following steps: Step S151: Preliminarily establish a three-dimensional scanning model of the fish sample at a standardized scanning position and posture based on the three-dimensional denoised sparse point cloud data of the fish sample surface, and obtain a preliminary three-dimensional scanning mapping model of the fish sample; Step S152: performing model surface smoothing processing on the preliminary mapping model of the three-dimensional scanning of the fish sample to obtain a smoothed surface structure model of the three-dimensional scanning of the fish sample; Step S153: performing texture feature vector conversion on the three-dimensional texture data of the fish sample surface to obtain the three-dimensional texture feature vector of the fish sample surface; Step S154: reconstructing the three-dimensional scanned surface structure smoothing model of the fish sample based on the three-dimensional texture feature vector of the fish sample surface, so as to obtain a three-dimensional scanned reconstructed model of the fish sample; Step S2: performing fish region grid segmentation processing on the fish sample 3D scanning and reconstruction model to obtain different fish sample 3D grid segmentation sub-regions, wherein the different fish sample 3D grid segmentation sub-regions include a fish head 3D grid segmentation sub-region, a fish trunk 3D grid segmentation sub-region, and a fish tail fin 3D grid segmentation sub-region; Step S3: performing sub-region fish phenotypic characteristic analysis on different fish sample three-dimensional grid segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions, wherein the fish phenotypic characteristic data sets of different fish segmentation sub-regions include the body length, body width, body height and external contour shape of the fish body part of the different fish segmentation sub-regions; Step S4: Calculate fish phenotypic parameters for fish phenotypic feature data sets of different fish segmentation sub-regions to obtain fish phenotypic parameter sets of different fish segmentation sub-regions; perform chart visualization conversion on the fish phenotypic parameter sets of different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters of different fish segmentation sub-regions; wherein step S4 includes the following steps: Step S41: Calculating the volume of the fish body according to the length, width and height of the fish body in different fish segmentation sub-regions, and obtaining volume parameters of the fish body in different fish segmentation sub-regions; Step S42: Calculating the fish body surface area according to the length and height of the fish body parts in different fish segmentation sub-regions, and obtaining the fish body surface area parameters of the different fish segmentation sub-regions; Step S43: calculating the fish body shape factors of the external contours of the fish body parts of the different fish segmentation sub-regions based on the fish body part volume parameters and the fish body part surface area parameters of the different fish segmentation sub-regions, to obtain the fish body part shape factors of the different fish segmentation sub-regions; Step S44: merging the fish body volume parameters, fish body surface area parameters and fish body shape factors of different fish segmentation sub-regions into fish phenotypic parameters to obtain fish phenotypic parameter sets of different fish segmentation sub-regions; Step S45: performing a graphical visualization conversion on the fish phenotypic parameter sets of different fish segmentation sub-regions to generate visualized phenotypes of fish characteristic parameters of different fish segmentation sub-regions.

2. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 1, characterized in that: Step S13 includes the following steps: Step S131: performing noise point distribution detection and annotation on the three-dimensional point cloud data of the fish sample surface to obtain the three-dimensional noise point distribution annotated point cloud data on the fish sample surface; Step S132: performing two-dimensional section segmentation processing on the three-dimensional noise point distribution annotated point cloud data on the surface of the fish sample to obtain the noise point distribution annotated point cloud data under the two-dimensional segmented section of each fish sample; Step S133: measuring and calculating the spatial distance between any two fish sample noise points in the noise point distribution annotated point cloud data under each fish sample two-dimensional segmentation section, and obtaining the spatial position distance value between any two fish sample noise points under each fish sample two-dimensional segmentation section; Step S134: comparing and judging the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section according to a preset spatial position spacing threshold value; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is greater than or equal to the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is increased by 1; when the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is less than the preset spatial position spacing threshold value, the sparsity at the point cloud spatial position of the corresponding fish sample noise point is reduced by 1, until the spatial position spacing value between any two fish sample noise points under each fish sample two-dimensional segmentation section is compared and judged, so as to construct a sparse matrix of the spatial distribution of fish sample noise points under each fish sample two-dimensional segmentation section; Step S135: Based on the sparse matrix of the spatial distribution of fish sample noise points under each two-dimensional segmentation section, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to noise point sparse filtering processing to obtain the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section of the fish sample; the fish sample noise point sparse filtering point cloud data under each two-dimensional segmentation section is subjected to cross-section three-dimensional integration and restoration processing to obtain the three-dimensional denoised sparse point cloud data of the fish sample surface.

3. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 2, characterized in that: Step S132 includes the following steps: A global noise distribution spatial model is established for the annotated point cloud data of the three-dimensional noise distribution on the surface of the fish sample, and a three-dimensional global noise distribution spatial model is obtained on the surface of the fish sample; The three-dimensional global noise distribution spatial model on the surface of fish samples is used to identify and analyze the three-dimensional morphology of fish samples and the distribution characteristics of noise points, and the three-dimensional distribution morphology of the fish sample surface and the distribution characteristics of three-dimensional noise points on the surface of fish samples are obtained; Based on the three-dimensional distribution morphology of the fish sample surface and the three-dimensional noise distribution characteristics of the fish sample surface, a two-dimensional cross-section cutting design is performed on the three-dimensional global noise distribution space model of the fish sample surface to generate a two-dimensional cross-section cutting design plan for the fish sample surface; According to the two-dimensional cross-section cutting design plan of the fish sample surface, the three-dimensional noise distribution annotated point cloud data on the fish sample surface is subjected to two-dimensional cross-section segmentation processing to obtain the noise distribution annotated point cloud data under the two-dimensional segmentation cross-section of each fish sample.

4. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 2, characterized in that: The step S135 described in which the noise point sparse filtering process is performed on the noise point distribution annotated point cloud data under the two-dimensional segmentation section of the corresponding fish sample based on the fish sample noise point spatial distribution sparse matrix under each fish sample two-dimensional segmentation section includes the following steps: Based on the sparse matrix of the spatial distribution of the noise points of the fish samples under the two-dimensional segmentation section of each fish sample, the noise point distribution annotation point cloud data under the corresponding two-dimensional segmentation section of the fish sample is subjected to point-by-point zeroing filtering processing to obtain the initial sparse filtering point cloud data of the noise point distribution annotation under the two-dimensional segmentation section of each fish sample; The local noise density is verified and estimated on the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample, so as to obtain the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample; Based on the local noise density distribution estimation data under the two-dimensional segmentation section of each fish sample, the initial sparse filtered point cloud data of the noise distribution annotation under the two-dimensional segmentation section of each fish sample is adaptively noise suppressed to obtain the fish sample noise sparse filtered point cloud data under the two-dimensional segmentation section of each fish sample.

5. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 1, characterized in that: Step S14 includes the following steps: Step S141: performing a fish sample point cloud surface structure topology analysis on the fish sample at a standardized scanning position and posture based on the fish sample surface three-dimensional denoised sparse point cloud data to obtain a fish sample three-dimensional point cloud surface topology structure model; Step S142: performing detail contrast enhancement processing on the three-dimensional point cloud surface topological structure model of the fish sample to obtain a three-dimensional surface structure detail contrast enhancement model of the fish sample; Step S143: performing three-dimensional texture feature analysis on the three-dimensional surface structure detail contrast enhancement model of the fish sample to obtain three-dimensional texture data of the fish sample surface.

6. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing fish area grid subdivision on the fish sample three-dimensional scanning and reconstruction model to obtain a fish sample three-dimensional scanning area subdivision grid model; Step S22: determining the key grid points of regional characteristic attributes for each fish area grid unit in the fish sample three-dimensional scanning area subdivision grid model, so as to obtain the fish sample head area grid starting point, the fish sample head area grid end point, the fish sample trunk area grid starting point, the fish sample trunk area grid end point, the fish sample tail fin area grid starting point and the fish sample tail fin area grid end point; Step S23: performing mesh segmentation of the fish sample head region on the fish sample three-dimensional scanning region subdivision mesh model based on the fish sample head region mesh starting point and the fish sample head region mesh end point to obtain the fish sample head three-dimensional mesh segmentation sub-region; Step S24: performing mesh segmentation of the fish sample trunk region on the fish sample 3D scanning region subdivision mesh model based on the fish sample trunk region mesh starting point and the fish sample trunk region mesh ending point to obtain the fish trunk 3D mesh segmentation sub-region; Step S25: performing fish sample tail fin region grid segmentation on the fish sample three-dimensional scanning region subdivision grid model based on the fish sample tail fin region grid starting point and the fish sample tail fin region grid end point to obtain the fish tail fin three-dimensional grid segmentation sub-region.

7. The method for automatically extracting fish phenotypes based on a three-dimensional scanning model according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: measuring the body length of the fish body in the different three-dimensional network segmentation sub-regions of the fish samples to obtain the body length of the fish body in the different fish segmentation sub-regions; Step S32: measuring the body width of the fish body in the different three-dimensional grid segmentation sub-regions of the fish sample to obtain the body width of the fish body in the different fish segmentation sub-regions; Step S33: measuring the body height of fish in different fish sample three-dimensional grid segmentation sub-regions to obtain the body height of fish in different fish segmentation sub-regions; Step S34: performing recognition and analysis of the external contour shape of the fish body part of the sub-regions of the three-dimensional grid segmentation sub-regions of different fish samples to obtain the external contour shape of the fish body part of the different fish segmentation sub-regions; Step S35: Integrate the fish phenotypic characteristics of the fish body length, fish body width, fish body height and fish body external contour shape of different fish segmentation sub-regions to obtain fish phenotypic characteristic data sets of different fish segmentation sub-regions.

Images