Simulation plant modeling method and system based on image processing

Abstract

The invention relates to the field of image processing, and discloses a simulation plant modeling method and system based on image processing, and the method comprises the steps: obtaining a plant image in a natural environment, extracting geometric features in the plant image through employing an image recognition technology, converting the extracted plant features into a three-dimensional model through employing three-dimensional modeling software, and carrying out the modeling of a simulation plant. Whether the model is matched with the actual plant or not is judged by comparing and analyzing the matching degree between the model and the actual plant, further analysis is carried out according to a matching analysis result, an error deviation degree coefficient is calculated, and whether an error between the three-dimensional simulation model and the actual plant needs to be corrected or not is judged; and the error of the model is dynamically corrected by using iterative correction according to an error detection result, so that the microscopic characteristics of the plant can be accurately captured, the dynamic change of the plant can be reflected in time, the precision of the model is further optimized according to real-time dynamic change adjustment, and the complexity of expansion and maintenance is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and more specifically to a simulation plant modeling method and system based on image processing. Background Art

[0002] Traditional simulation plant modeling methods mainly rely on physical models and manual drawing, which are inefficient and difficult to adapt to the complex dynamic changes of plant morphology. They cannot directly extract key features from real plants and it is difficult to achieve high-precision and dynamic simulation. With the rapid development of computer technology, artificial intelligence and image processing technology, simulation plant modeling methods based on image processing have gradually become an important tool for studying plant morphology, growth dynamics and environmental response. This method achieves high-precision simulation of plant morphology, dynamics and environmental response by acquiring plant image data and combining computer vision and image processing technology; However, the above process still has the following disadvantages: First, the existing methods lack three-dimensional modeling and multi-source sensors, and cannot accurately capture the microscopic characteristics of plants, reflect the dynamic changes of plants, and further optimize the accuracy of the model according to environmental changes; Second, the existing methods lack the matching analysis between the constructed model and the actual plant image, and cannot accurately reflect the actual morphology and details of the plant. The error defects are then detected based on the results of the matching analysis, which can easily lead to inaccurate correction of model errors. There is also a lack of iterative correction and comparative verification, and it is unable to adapt to dynamic changes, which increases the complexity of expansion and maintenance. Summary of the invention

[0003] In order to overcome the above-mentioned defects of the prior art, the present invention provides a simulated plant modeling method and system based on image processing to solve the problems existing in the above-mentioned background technology.

[0004] The present invention provides the following technical solution: a simulated plant modeling method based on image processing, comprising: S1: used to collect images of natural plants, by acquiring plant images in natural environments and uploading the acquired plant images to an image processing center; S2: Based on feature extraction of plant images uploaded to the image processing center, the geometric features in the plant images are extracted by using image recognition technology, and the lighting conditions in the images are analyzed to identify the shadow areas; S3: Constructing a three-dimensional simulation model of the plant by converting the extracted plant features into a three-dimensional model; S4: Analyze the matching degree between the three-dimensional simulation model and the actual plant through multi-view comparison, calculate the matching degree, and judge whether the three-dimensional simulation model matches the actual plant; S5: Based on the results of the comparative analysis, the errors of all corresponding points between the three-dimensional simulation model and the actual plant are further analyzed to obtain the errors between the corresponding points and the maximum error; S6: By analyzing the error between corresponding points and the maximum error, the error deviation coefficient is calculated to determine whether the error between the three-dimensional simulation model and the actual plant needs to be corrected in detail and position; S7: Iteratively correct the three-dimensional simulation model and compare the corrected model with the actual plant until the correction of the three-dimensional simulation model is verified.

[0005] Preferably, the S1 uses a high-resolution camera to shoot the plant from multiple angles, including the top, bottom, side, front and back, and continuously shoots multiple images to form a set of image sequences, and uploads the image sequences to an image processing center.

[0006] Preferably, S2 preprocesses the plant images uploaded to the image processing center, including denoising, contrast enhancement and size adjustment. The feature extraction process for the plant images includes: using the Canny edge detection algorithm to find the plant outline, dividing the image into different areas according to the color features, texture features and shape features of the images of various parts of the plant, calculating the leaf shape according to the length, width, area, circumference and shape factor of the leaves, calculating the stem structure according to the diameter, length and direction of the stem extracted from the image, calculating the flower distribution according to the position, size and shape of the flowers extracted from the image, estimating the lighting conditions by analyzing the global and local brightness distribution of the image, and then using the threshold segmentation method to identify the shadow area of ​​the image.

[0007] Preferably, S3 constructs a three-dimensional geometric model of the plant by converting the extracted two-dimensional image features into three-dimensional space coordinates, inputting the converted three-dimensional coordinate data using three-dimensional modeling software, generating a three-dimensional point cloud of the plant by using a stereo matching algorithm, reconstructing the three-dimensional surface of the plant from the point cloud data by using a Poisson reconstruction algorithm, and then mapping the texture information of the original image onto the reconstructed three-dimensional model.

[0008] Preferably, S4 identifies and matches corresponding key feature points on the three-dimensional simulation model and the actual plant image, including leaf tips, leaf bases, stems, flowers and fruits, and compares corresponding areas of the three-dimensional simulation model and the actual plant image from multiple perspectives, thereby analyzing whether the three-dimensional simulation model matches the actual plant image; The specific analysis method of the matching degree is as follows: Step S411: The point cloud of the simulation model is registered with the scanned point cloud of the actual plant by using the iterative closest point algorithm, thereby analyzing the average distance error between the key feature points of the actual plant image and the 3D simulation model, and calculating the geometric matching degree: ,in, represents the coordinates of the i-th feature point on the actual plant image, represents the coordinates of the i-th feature point on the 3D simulation model, and N represents the number of matching feature points; Step S412: By using the local binary pattern to extract texture features, the texture features of the actual plant image and the three-dimensional simulation model rendering image are compared and analyzed, and the texture matching degree is calculated as , where X represents the actual plant image and Y represents the 3D simulation model rendering image. represents the local mean of image X, represents the local mean value of image Y, represents the local variance of image X, represents the local variance of image Y, represents the local covariance of images X and Y, and represents a small constant introduced to avoid the denominator being zero; Step S413: Based on the geometric matching degree and the texture matching degree, the matching degree is calculated as , and is the weight coefficient; Set a matching threshold By comparing the matching degree with the matching threshold, it is determined whether the 3D simulation model matches the actual plant. Matching threshold , then it is determined that the 3D simulation model matches the actual plant well, and no error correction analysis is required. If the matching degree M Matching threshold , it is determined that the 3D simulation model does not match the actual plant, and further error analysis is required.

[0009] Preferably, the S5 calculates the error between the corresponding points by using the coordinates of the corresponding points between the three-dimensional simulation model and the actual plant and combining the light and wind influence factors. ,in, represents the error between the jth corresponding point of the 3D simulation model and the actual plant, represents the coordinates of the jth corresponding point on the actual plant image, represents the coordinates of the jth corresponding point on the three-dimensional simulation model, represents the light impact factor, represents the wind force influence factor, and the specific value is obtained by simulation based on experimental data; the maximum error is .

[0010] Preferably, the specific analysis method of S6 for the error deviation degree coefficient is: Step S611: Calculate the average error of the corresponding points based on the errors between the corresponding points: , m represents the total number of corresponding points; Step S612: Based on the average error of the corresponding points and the error between the corresponding points, the error deviation coefficient is calculated as ; Through the maximum error The error deviation coefficient R is used to judge whether the error between the 3D simulation model and the actual plant needs to be corrected, and a maximum error threshold is set. and error deviation threshold If the maximum error Maximum error threshold And the error deviation coefficient R Error deviation threshold , then the error is considered to be within the acceptable range and the model does not need to be modified. If the error deviation coefficient R Error deviation threshold or maximum error Maximum error threshold , it indicates that the relative fluctuation of the error is large, and the three-dimensional simulation model needs to be further corrected.

[0011] Preferably, S7 uses an iterative correction method to correct the three-dimensional simulation model based on the judgment result of the model correction, and compares the corrected model with the actual plant until the maximum error and the error deviation coefficient are both lower than the threshold value, thereby ending the correction process and confirming that the three-dimensional simulation model has been verified.

[0012] To achieve the above object, the present invention provides the following technical solution: a simulated plant modeling system based on image processing, implementing the above-mentioned simulated plant modeling method based on image processing, comprising: Image collection module: used to collect images of natural plants by acquiring plant images in natural environments and uploading the acquired plant images to the image processing center; Image feature extraction module: Based on feature extraction of plant images uploaded to the image processing center, the geometric features in the plant images are extracted by using image recognition technology, and the lighting conditions in the images are analyzed to identify shadow areas. The extracted plant features are then transmitted to the model building module; Model building module: by using 3D modeling software, the extracted plant features are converted into 3D models, thereby building a 3D simulation model of the plant; Comparative analysis module: Through multi-view comparative analysis, the matching degree between the 3D simulation model and the actual plant is calculated to determine whether the 3D simulation model matches the actual plant, and the comparative analysis results are transmitted to the error analysis module; Error analysis module: Based on the results of comparative analysis, further analyze the errors of all corresponding points between the 3D simulation model and the actual plant to obtain the errors between corresponding points and the maximum error; Error detection module: By analyzing the error between corresponding points and the maximum error, the error deviation coefficient is calculated to determine whether the error between the 3D simulation model and the actual plant needs to be corrected in detail and position, and the judgment result is transmitted to the correction verification module; Correction verification module: It iteratively corrects the three-dimensional simulation model and compares the corrected model with the actual plant until the correction verification of the three-dimensional simulation model is passed.

[0013] Technical effects and advantages of the present invention: The present invention collects images of natural plants, obtains plant images in natural environments, extracts geometric features in plant images by using image recognition technology, analyzes lighting conditions in images, identifies shadow areas, and converts the extracted plant features into three-dimensional models by using three-dimensional modeling software, thereby constructing a three-dimensional simulation model of the plant, which is conducive to capturing the microscopic features of the plant in real time and accurately, reflecting the dynamic changes of the plant in a timely manner, and further optimizing the accuracy of the model according to environmental changes. The matching degree between the three-dimensional simulation model and the actual plant is analyzed by multi-perspective comparison, the matching degree is calculated, and it is determined whether the three-dimensional simulation model matches the actual plant. According to the matching analysis results, all corresponding points between the three-dimensional simulation model and the actual plant are further analyzed. The error between the corresponding points and the maximum error are obtained, and then the error deviation degree coefficient is calculated to determine whether the error size between the 3D simulation model and the actual plant needs to be corrected in detail and position. The 3D simulation model is iteratively corrected and the corrected model is compared with the actual plant until the 3D simulation model correction is verified. By matching and analyzing the constructed model with the actual plant image, it is beneficial to accurately reflect the actual morphology and details of the plant. The error defects are detected according to the results of the matching analysis, and the error defects of the model can be accurately corrected, which is beneficial to dynamically adjust the error area of ​​the model, so that it can better adapt to dynamic changes and make targeted corrections, reducing the complexity of expansion and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a diagram of the method steps of the present invention.

[0015] Figure 2 It is a system structure block diagram of the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are merely illustrative. The image processing-based simulation plant modeling method and system involved in the present invention are not limited to the various structures recorded in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0017] like Figure 1 This embodiment provides a simulated plant modeling method based on image processing, comprising: S1: used to collect images of natural plants by acquiring plant images in natural environments and uploading the acquired plant images to an image processing center.

[0018] In this embodiment, the S1 uses a high-resolution camera to shoot the plant from multiple angles, including the top, bottom, side, front and back, and continuously shoots multiple images to form a set of image sequences, which are then uploaded to an image processing center.

[0019] Specifically, by selecting representative plant growth locations and periods of uniform lighting, adjusting high-resolution camera parameters, and using the camera to continuously capture plant images from different angles, including top, bottom, side, front, and back, the images are taken in the following order: (1) Top perspective shooting: photograph the entire plant from above to ensure that the top and overall shape of the plant are captured; (2) Side perspective shooting: photograph from the side to show the side view of the plant and the arrangement of leaves; (3) Bottom perspective shooting: photograph from below to capture the roots and bottom structure of the plant; (4) Close-up shooting: take close-up shots of detailed parts, such as flowers or fruits; (5) Dynamic perspective shooting: if the plant is dynamically changing, a series of continuous photos or videos can be taken to capture the dynamic effects of the plant in the wind; record the environmental parameters during shooting, including lighting conditions, ambient temperature, and weather conditions, and check whether the shooting data covers all key parts of the plant, such as leaves, stems, flowers, and fruits. If some parts are not fully covered, they need to be supplemented with photos. The acquired data are classified according to plant species and stored in the image processing center.

[0020] S2: Based on feature extraction of plant images uploaded to the image processing center, the geometric features in the plant images are extracted by using image recognition technology, and the lighting conditions in the images are analyzed to identify the shadow areas.

[0021] In this embodiment, S2 preprocesses the plant images uploaded to the image processing center, including denoising, contrast enhancement and size adjustment. The feature extraction process of the plant images includes: using the Canny edge detection algorithm to find the plant outline, dividing the image into different areas according to the color characteristics, texture characteristics and shape characteristics of the images of various parts of the plant, calculating the leaf shape according to the length, width, area, circumference and shape factor of the leaves, calculating the stem structure according to the diameter, length and direction of the stem extracted from the image, calculating the flower distribution according to the position, size and shape of the flowers extracted from the image, estimating the lighting conditions by analyzing the global and local brightness distribution of the image, and then using the threshold segmentation method to identify the shadow area of ​​the image.

[0022] It should be specifically noted that the specific steps of extracting flower images through features include: Step 1: Use Canny edge detection or other edge detection algorithms to find the boundaries of the flowers, use color histogram to locate the flower area, and use Harris corner detection to locate the feature points of the flowers; Step 2: Use Fourier transform to calculate and describe the shape of the flower based on the length, width, area, and perimeter of the extracted flower; Step 3: Use a color-based segmentation algorithm to separate the flower area and extract the color histogram of the flower area to describe the color characteristics of the flower. Then use a deep learning model to classify the color of the flower. Step 4: Extract the number and arrangement density of flowers through image segmentation, point cloud generation and statistical analysis techniques; Step 5: Extract the arrangement of flowers by analyzing their relative position, orientation, and spacing.

[0023] S3: By converting the extracted plant features into a three-dimensional model, a three-dimensional simulation model of the plant is constructed.

[0024] In this embodiment, the S3 constructs a three-dimensional geometric model of the plant by converting the extracted two-dimensional image features into three-dimensional space coordinates, using three-dimensional modeling software to input the converted three-dimensional coordinate data, generates a three-dimensional point cloud of the plant by using a stereo matching algorithm, reconstructs the three-dimensional surface of the plant from the point cloud data by using a Poisson reconstruction algorithm, and then maps the texture information of the original image to the reconstructed three-dimensional model.

[0025] It should be specifically noted that the specific steps for constructing a three-dimensional geometric model of a plant include: Step 1: Use Canny edge detection, color, texture and shape feature extraction methods to process the two-dimensional image; Step 2: Determine the 2D coordinates of each feature point in the image. If there are images from multiple perspectives, use the stereo matching algorithm to find the corresponding points of the same object in images from different perspectives. Then, based on the matching results and the camera's internal and external parameters, calculate the 3D coordinates of each feature point to generate a 3D point cloud of the plant. Step 3: Use the Poisson reconstruction algorithm to input the point cloud data, construct a continuous scalar function, and then use the iterative method to solve the Poisson equation to obtain a smooth three-dimensional surface, thereby reconstructing the three-dimensional surface of the plant from the point cloud data; Step 4: Extract texture information from the original image, determine the 2D image texture coordinates corresponding to each vertex on the 3D model, use the texture mapping algorithm to associate the texture coordinates with the 3D model vertices to render the 3D model so that the texture information is correctly displayed on the model surface.

[0026] S4: By comparing and analyzing the matching degree between the three-dimensional simulation model and the actual plant from multiple perspectives, the matching degree is calculated to determine whether the three-dimensional simulation model matches the actual plant.

[0027] In this embodiment, S4 identifies and matches corresponding key feature points on the three-dimensional simulation model and the actual plant image, including leaf tips, leaf bases, stems, flowers, and fruits, and compares the corresponding areas of the three-dimensional simulation model and the actual plant image from multiple perspectives, thereby analyzing whether the three-dimensional simulation model matches the actual plant image; The specific analysis method of the matching degree is as follows: Step S411: The point cloud of the simulation model is registered with the scanned point cloud of the actual plant by using the iterative closest point algorithm, thereby analyzing the average distance error between the key feature points of the actual plant image and the 3D simulation model, and calculating the geometric matching degree: ,in, represents the coordinates of the i-th feature point on the actual plant image, represents the coordinates of the i-th feature point on the 3D simulation model, and N represents the number of matching feature points; Step S412: By using the local binary pattern to extract texture features, the texture features of the actual plant image and the three-dimensional simulation model rendering image are compared and analyzed, and the texture matching degree is calculated as , where X represents the actual plant image and Y represents the 3D simulation model rendering image. represents the local mean of image X, represents the local mean value of image Y, represents the local variance of image X, represents the local variance of image Y, represents the local covariance of images X and Y, and represents a small constant introduced to avoid the denominator being zero; Step S413: Based on the geometric matching degree and the texture matching degree, the matching degree is calculated as , and is the weight coefficient; Set a matching threshold By comparing the matching degree with the matching threshold, it is determined whether the 3D simulation model matches the actual plant. Matching threshold , then it is determined that the 3D simulation model matches the actual plant well, and no error correction analysis is required. If the matching degree M Matching threshold , it is determined that the 3D simulation model does not match the actual plant, and further error analysis is required.

[0028] It should be specifically noted that the actual plant image and the simulation model rendering image are processed in blocks, a window of appropriate size is selected, and the local average value of image X is calculated based on each window. , the local mean of image Y , local variance of image X , local variance of image Y , local covariance of images X and Y ; For constant and The value of depends on the dynamic range value K of the image, which is usually , , where K is usually set to 255.

[0029] S5: Based on the results of the comparative analysis, the errors of all corresponding points between the three-dimensional simulation model and the actual plant are further analyzed to obtain the errors between the corresponding points and the maximum error.

[0030] In this embodiment, S5 calculates the error between the corresponding points by using the coordinates of the corresponding points between the three-dimensional simulation model and the actual plant and combining the light and wind influence factors. ,in, represents the error between the jth corresponding point of the 3D simulation model and the actual plant, represents the coordinates of the jth corresponding point on the actual plant image, represents the coordinates of the jth corresponding point on the three-dimensional simulation model, represents the light impact factor, represents the wind force influence factor, and the specific value is obtained by simulation based on experimental data; the maximum error is .

[0031] S6: By analyzing the errors between corresponding points and the maximum error, the error deviation degree coefficient is calculated to determine whether the error size between the three-dimensional simulation model and the actual plant requires detail and position correction.

[0032] In this embodiment, the specific analysis method of S6 for the error deviation degree coefficient is: Step S611: Calculate the average error of the corresponding points based on the errors between the corresponding points: , m represents the total number of corresponding points; Step S612: Based on the average error of the corresponding points and the error between the corresponding points, the error deviation coefficient is calculated as ; Through the maximum error The error deviation coefficient R is used to judge whether the error between the 3D simulation model and the actual plant needs to be corrected, and a maximum error threshold is set. and error deviation threshold If the maximum error Maximum error threshold And the error deviation coefficient R Error deviation threshold , then the error is considered to be within the acceptable range and the model does not need to be modified. If the error deviation coefficient R Error deviation threshold or maximum error Maximum error threshold , it indicates that the relative fluctuation of the error is large, and the three-dimensional simulation model needs to be further corrected.

[0033] S7: Iteratively correct the three-dimensional simulation model and compare the corrected model with the actual plant until the correction of the three-dimensional simulation model is verified.

[0034] In this embodiment, S7 uses an iterative correction method to correct the three-dimensional simulation model based on the judgment result of the model correction, and compares the corrected model with the actual plant until the maximum error and the error deviation coefficient are both lower than the threshold. The correction process is ended and the three-dimensional simulation model verification is confirmed to have passed.

[0035] like Figure 2The present embodiment shown provides an implementation system corresponding to an anti-counterfeiting method for articles based on local feature visual information, including an image collection module, an image feature extraction module, a model construction module, a comparison and analysis module, an error analysis module, an error detection module and a correction and verification module, wherein the image collection module is connected to the image feature extraction module, the image feature extraction module is connected to the model construction module, the model construction module is connected to the comparison and analysis module, the comparison and analysis module is connected to the error analysis module, the error analysis module is connected to the error detection module, and the error detection module is connected to the correction and verification module.

[0036] The image collection module is used to collect images of natural plants by acquiring plant images in a natural environment and uploading the acquired plant images to an image processing center; The image feature extraction module extracts features from plant images uploaded to the image processing center, extracts geometric features from plant images by using image recognition technology, analyzes lighting conditions in the images, identifies shadow areas, and transmits the extracted plant features to the model building module; The model building module converts the extracted plant features into a three-dimensional model by using three-dimensional modeling software, thereby building a three-dimensional simulation model of the plant; The comparison and analysis module compares and analyzes the matching degree between the three-dimensional simulation model and the actual plant through multiple perspectives, calculates the matching degree, determines whether the three-dimensional simulation model matches the actual plant, and transmits the comparison and analysis results to the error analysis module; The error analysis module further analyzes the errors of all corresponding points between the three-dimensional simulation model and the actual plant based on the results of the comparative analysis, and obtains the errors between the corresponding points and the maximum error; The error detection module calculates the error deviation coefficient by analyzing the error between corresponding points and the maximum error, determines whether the error between the three-dimensional simulation model and the actual plant needs to be corrected in detail and position, and transmits the judgment result to the correction verification module; The correction verification module iteratively corrects the three-dimensional simulation model and compares the corrected model with the actual plant until the correction verification of the three-dimensional simulation model is passed.

[0037] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0038] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A simulated plant modeling method based on image processing, characterized in that: include: S1: used to collect images of natural plants, by acquiring plant images in natural environments and uploading the acquired plant images to an image processing center; S2: Based on feature extraction of plant images uploaded to the image processing center, the geometric features in the plant images are extracted by using image recognition technology, and the lighting conditions in the image are analyzed to identify the shadow area; S3: By converting the extracted plant features into a three-dimensional model, a three-dimensional simulation model of the plant is constructed; S4: Through multi-view comparison and analysis of the matching degree between the three-dimensional simulation model and the actual plant, the matching degree is calculated to determine whether the three-dimensional simulation model matches the actual plant; S5: Based on the results of the comparative analysis, the errors of all corresponding points between the three-dimensional simulation model and the actual plant are further analyzed to obtain the errors between the corresponding points and the maximum error; S6: By analyzing the error between corresponding points and the maximum error, the error deviation coefficient is calculated to determine whether the error between the three-dimensional simulation model and the actual plant needs to be corrected in detail and position; S7: Iteratively correct the three-dimensional simulation model and compare the corrected model with the actual plant until the correction of the three-dimensional simulation model is verified.

2. The method for modeling simulated plants based on image processing according to claim 1, characterized in that: The S1 uses a high-resolution camera to shoot plants from multiple angles, including the top, bottom, side, front, and back, and continuously shoots multiple images to form a set of image sequences, which are then uploaded to an image processing center.

3. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The S2 preprocesses the plant images uploaded to the image processing center, including denoising, contrast enhancement and size adjustment. The feature extraction process of the plant images includes: using the Canny edge detection algorithm to find the plant outline, dividing the image into different areas according to the color characteristics, texture characteristics and shape characteristics of the images of various parts of the plant, calculating the leaf shape according to the length, width, area, circumference and shape factor of the leaf, calculating the stem structure according to the diameter, length and direction of the stem extracted from the image, calculating the flower distribution according to the position, size and shape of the flower extracted from the image, estimating the lighting conditions by analyzing the global and local brightness distribution of the image, and then using the threshold segmentation method to identify the shadow area of ​​the image.

4. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The S3 constructs a three-dimensional geometric model of the plant by converting the extracted two-dimensional image features into three-dimensional space coordinates, inputting the converted three-dimensional coordinate data using three-dimensional modeling software, generating a three-dimensional point cloud of the plant by using a stereo matching algorithm, reconstructing the three-dimensional surface of the plant from the point cloud data by using a Poisson reconstruction algorithm, and then mapping the texture information of the original image onto the reconstructed three-dimensional model.

5. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The S4 identifies and matches corresponding key feature points on the three-dimensional simulation model and the actual plant image, including leaf tips, leaf bases, stems, flowers and fruits, and compares the corresponding areas of the three-dimensional simulation model and the actual plant image from multiple perspectives, thereby analyzing whether the three-dimensional simulation model matches the actual plant image; the specific analysis method of the matching degree is as follows: Step S411: By using an iterative closest point algorithm to align the point cloud of the simulation model with the scanned point cloud of the actual plant, the average distance error between the key feature points of the actual plant image and the three-dimensional simulation model is analyzed, and the geometric matching degree is calculated as ,in, represents the coordinates of the i-th feature point on the actual plant image, represents the coordinates of the i-th feature point on the three-dimensional simulation model, and N represents the number of matching feature points; Step S412: by using the local binary pattern to extract texture features, the texture features of the actual plant image and the three-dimensional simulation model rendering image are compared and analyzed, and the texture matching degree is calculated as , where X represents the actual plant image and Y represents the 3D simulation model rendering image. represents the local mean of image X, represents the local mean value of image Y, represents the local variance of image X, represents the local variance of image Y, represents the local covariance of images X and Y, and represents a small constant introduced to avoid the denominator being zero; Step S413: Based on the geometric matching degree and the texture matching degree, the matching degree is calculated as , and is the weight coefficient; set a matching threshold By comparing the matching degree with the matching threshold, it is determined whether the 3D simulation model matches the actual plant. Matching threshold , then it is determined that the 3D simulation model matches the actual plant well, and no error correction analysis is required. If the matching degree M Matching threshold , it is determined that the 3D simulation model does not match the actual plant, and further error analysis is required.

6. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The S5 calculates the error between the corresponding points by using the coordinates of the corresponding points between the three-dimensional simulation model and the actual plant and combining the light and wind influence factors: ,in, represents the error between the jth corresponding point of the 3D simulation model and the actual plant, represents the coordinates of the jth corresponding point on the actual plant image, represents the coordinates of the jth corresponding point on the three-dimensional simulation model, represents the light impact factor, represents the wind force influence factor, and the specific value is obtained by simulation based on experimental data; the maximum error is .

7. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The specific analysis method of the error deviation coefficient in step S6 is as follows: Step S611: Based on the errors between corresponding points, the average error of corresponding points is calculated as , m represents the total number of corresponding points; Step S612: Based on the average error of the corresponding points and the error between the corresponding points, the error deviation coefficient is calculated as ; Through the maximum error The error deviation coefficient R is used to judge whether the error between the 3D simulation model and the actual plant needs to be corrected, and a maximum error threshold is set. and error deviation threshold If the maximum error Maximum error threshold And the error deviation coefficient R Error deviation threshold , then the error is considered to be within the acceptable range and the model does not need to be modified. If the error deviation coefficient R Error deviation threshold or maximum error Maximum error threshold , it indicates that the relative fluctuation of the error is large, and the three-dimensional simulation model needs to be further corrected.

8. The method for simulated plant modeling based on image processing according to claim 1, characterized in that: The S7 uses an iterative correction method to correct the three-dimensional simulation model based on the judgment result of the model correction, and compares the corrected model with the actual plant until the maximum error and the error deviation coefficient are both lower than the threshold, then ends the correction process and confirms that the three-dimensional simulation model has been verified.

9. A simulated plant modeling system based on image processing, implementing a simulated plant modeling method based on image processing as claimed in any one of claims 1 to 8, characterized in that: include: Image collection module: used to collect images of natural plants by acquiring plant images in natural environments and uploading the acquired plant images to the image processing center; Image feature extraction module: based on feature extraction of plant images uploaded to the image processing center, the geometric features in the plant images are extracted by using image recognition technology, and the lighting conditions in the images are analyzed to identify the shadow areas, and the extracted plant features are transmitted to the model construction module; Model construction module: by using 3D modeling software, the extracted plant features are converted into 3D models, thereby constructing a 3D simulation model of the plant; Comparative analysis module: by comparing and analyzing the matching degree between the 3D simulation model and the actual plant from multiple perspectives, the matching degree is calculated, and it is determined whether the 3D simulation model matches the actual plant, and the comparative analysis results are transmitted to the error analysis module; Error analysis module: Based on the results of comparative analysis, further analyze the errors of all corresponding points between the 3D simulation model and the actual plant to obtain the errors between the corresponding points and the maximum error; Error detection module: By analyzing the errors between the corresponding points and the maximum error, calculate the error deviation degree coefficient, judge whether the error size between the 3D simulation model and the actual plant needs to be corrected in detail and position, and transmit the judgment result to the correction verification module; Correction verification module: It iteratively corrects the three-dimensional simulation model and compares the corrected model with the actual plant until the correction verification of the three-dimensional simulation model is passed.

Images