A high-precision granary volume calculation method and system

Abstract

The invention discloses a high-precision granary volume calculation method and system, which relate to the fields of laser radar technology and point cloud processing technology, and include the following steps: data acquisition; using laser radar equipment to perform all-round scanning on a target object to obtain a laser point cloud data set; data preprocessing; point cloud segmentation; object recognition; point cloud accuracy compensation, performing accuracy compensation on point cloud data, the accuracy compensation includes error compensation for laser radar accuracy that changes with distance and temperature; volume calculation, for identified regular objects, using a volume calculation formula to perform volume calculation; for identified irregular objects, creating a three-dimensional grid model of the object through a three-dimensional model reconstruction algorithm, adjusting the grid model, and then calculating its volume; the invention optimizes from two aspects of radar distance and temperature compensation, and establishes an error model, in which basis functions are designed from two aspects of temperature and distance in the model, and the accuracy is improved from two aspects at the same time.

Description

Technical Field

[0001] The present invention relates to the field of laser radar technology and point cloud processing technology, and in particular to a high-precision granary volume calculation method and system. Background Art

[0002] Laser radar technology (LiDAR) is a modern remote sensing technology that measures the distance, speed and other characteristics of a target by emitting a laser beam to the target and receiving the reflected beam. The LiDAR system can quickly scan the surface of the pile and obtain a large amount of data from the target points. After processing, these data can generate accurate three-dimensional stereo images to reveal the shape and volume of the pile. This technology is a non-contact measurement. When measuring the volume of the pile, it does not require direct contact with the measured object, and is suitable for dangerous or inaccessible environments. Existing LiDAR technology has been widely used in measuring the volume of objects. However, the accuracy of traditional LiDAR in measuring the volume of grain silos is only about 95%, which cannot meet the requirements of some customers for higher accuracy. In addition, due to the fluctuation of radar stability, temperature influence and missing point cloud data, the point cloud data of the material surface is inaccurate, which further affects the accuracy of volume measurement. Summary of the invention

[0003] The purpose of the present invention is to improve the accuracy of laser radar in measuring the volume of a granary, and to provide a high-precision granary volume calculation method and system.

[0004] The technical solution of the present invention is as follows:

[0005] S1. Data collection: Use laser radar equipment to perform a full-scale scan of the target object and obtain point cloud data;

[0006] S2. Data preprocessing;

[0007] S3. Rough segmentation of point cloud: Rough segmentation of the preprocessed point cloud; the rough segmentation here is the approximate area of ​​the material surface and the granary;

[0008] S4. Fine segmentation of point cloud: Use machine learning or deep learning methods to finely segment the point cloud after rough segmentation;

[0009] S5. Point cloud accuracy compensation: perform accuracy compensation on the finely segmented point cloud data. The accuracy compensation is the error compensation of the laser radar accuracy changing with distance and temperature.

[0010] S6. Calculate the volume of the point cloud after accuracy compensation.

[0011] Furthermore, the point cloud rough segmentation of S3 includes:

[0012] S31. Splitting the point cloud into shapes, where the split shapes include but are not limited to cylinders, cones, and planes;

[0013] S32. Use RANSAC to segment the cylinder and cone. If the cylinder and cone are successfully segmented, combine them to obtain a segmentation result.

[0014] S33. If the cylinder and cone are not segmented in S32, the RANSAC method is used to perform point cloud plane segmentation, and the segmented planes are combined to generate a point cloud segmentation result.

[0015] Specifically, the point cloud fine segmentation adopts PointNet model or PointNet++ model for fine segmentation.

[0016] Furthermore, the point cloud accuracy compensation step in S5 is:

[0017] By comparing the accuracy errors between different parts of the standard object and the corresponding point cloud in advance, the error model function of distance, temperature and accuracy is established;

[0018] The error value of the temperature and distance during measurement is calculated according to the established error model function and the error correction is performed;

[0019] The volume calculation in S6 is based on the corrected point cloud.

[0020] Furthermore, the step of establishing the error model function of distance, temperature and accuracy by comparing the accuracy errors between different parts of the standard object and the corresponding point cloud in advance includes the following steps:

[0021] S51. Collect point cloud data in advance and obtain the standard design dimension drawing of the granary;

[0022] S52. Preprocessing the collected point cloud data, removing noise in the point cloud data, smoothing the point cloud data, calibrating the point cloud data, and locally downsampling the calibrated point cloud data;

[0023] S53. Performing granary identification and segmentation on the downsampled point cloud to segment the granary body point cloud, where the granary body point cloud includes a material surface point cloud and a granary wall point cloud;

[0024] S54. Extracting the point cloud information at different positions from the radar optical center, the temperature information of the granary at the corresponding time, and the size information of the corresponding standard granary;

[0025] S55. According to the information in step S54, an error model function is constructed in three-dimensional space. The specific function relationship is as follows:

[0026] ;

[0027] in, represents the error model function value; represents the key points of the generated error model, m and n represent the number of key points selected in the x-axis and y-axis directions, d and t represent the collected distance value and temperature value, respectively. and is the basis function, and the expression is:

[0028] ;

[0029] in, Represents the order of the basis function.

[0030] Furthermore, the error value calculation and error correction of the temperature and distance during measurement according to the established error model function are specifically as follows:

[0031] According to the current point Generate a lookup point based on the distance and temperature , find and calculate the corresponding error value in the model , while searching from the point cloud The normal vector of , adjust the position of the current point according to the normal vector:

[0032] ;

[0033] in, For the current point Corrected point.

[0034] Preferably, in the point cloud segmentation, segmentation is performed according to the shape and / or color features of the object, and the segmentation algorithm is selected from RANSAC or an algorithm based on morphology.

[0035] Specifically, the step S6 is as follows: for the identified irregular object, a convex hull model is calculated for the corrected point cloud, the convex hull model is continuously refined and fitted to the point cloud, and the volume of the three-dimensional model is calculated according to the fitted point cloud.

[0036] Furthermore, the specific steps of S6 include:

[0037] S61. Get the three vertex coordinates of a specified triangle in the convex hull model and the normal vector corresponding to the triangle face, the normal vector is recorded as ;

[0038] S62. Move the point cloud along the normal vector of the triangle Perform projection, extract the points projected into the triangle, and calculate the projection distance of each point inside the triangle;

[0039] S63. Subdivide the points inside the triangle into a mesh, and then mesh each vertex of the subdivided triangle along the normal direction Move in the opposite direction, and the moving distance is the projection distance of each point;

[0040] S64. Repeat steps S61 to S63 until all triangles in the convex hull model are traversed and the fitting of the convex hull model is completed;

[0041] S65. Split the fitted convex hull model into small tetrahedrons, calculate the volume of each tetrahedron and perform accumulation operation to obtain the total volume.

[0042] Furthermore, the method also includes: S7. 3D rendering, presenting various states of the granary to the user in the form of 3D interactive models and animations.

[0043] At the same time, the present invention also discloses a granary volume calculation system based on material surface point cloud compensation using a laser radar, the system being used to calculate the granary volume, and the system comprising:

[0044] The data acquisition module is used to transmit a laser beam through the laser radar and receive the reflected signal, calculate the distance using the time difference or phase difference, and obtain point cloud data;

[0045] Data preprocessing module, used to preprocess point cloud data;

[0046] Point cloud coarse segmentation module, used to perform coarse segmentation on the preprocessed point cloud data;

[0047] Point cloud fine segmentation module: used to perform fine segmentation on the point cloud after rough segmentation;

[0048] Point cloud accuracy compensation module, used to perform accuracy compensation on finely segmented point cloud data;

[0049] The volume calculation module is used to calculate the volume of the point cloud after accuracy compensation.

[0050] After adopting the above scheme, the beneficial effects of the present invention are:

[0051] (1) The present invention achieves high-precision volume calculation through multi-faceted precision optimization and data correction. Specifically, the present invention optimizes the radar distance and temperature compensation. Because the point cloud accuracy of the laser radar is different at different distances and temperatures when scanning the target object, the present invention establishes an error model. In the model, basis functions are designed from the two aspects of temperature and distance, thereby improving the accuracy from both aspects.

[0052] (2) In a specific error model, it has strong approximation and fitting capabilities, can accurately fit the local characteristics of the error model, and construct a complete and reliable error model.

[0053] (3) The present invention continuously refines the fitting point cloud through the convex hull model, thereby avoiding the volume error caused by noise points, sparse point clouds and partial incomplete point clouds, and further improving the calculation accuracy of the granary volume. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is the main flow chart of the present invention;

[0055] Figure 2 It is a radar accuracy distribution diagram at different distances in a specific implementation manner;

[0056] Figure 3 It is a distribution diagram of radar accuracy at different distances after correction in a specific implementation manner;

[0057] Figure 4 is a precision deviation diagram before temperature compensation in a specific implementation manner;

[0058] Figure 5 It is a precision deviation diagram after temperature compensation in a specific implementation mode;

[0059] Figure 6 is a schematic diagram of points inside a triangle in a specific implementation manner;

[0060] Figure 7 This is a schematic diagram of mesh subdivision inside a triangle in a specific implementation manner;

[0061] Figure 8 A point cloud diagram showing that a granary has an arching tendency in a specific implementation manner;

[0062] Fig. 9 This is an interactive interface diagram of a granary belt 3D model rendering in a specific implementation method. DETAILED DESCRIPTION

[0063] Now, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the accompanying drawings are only exemplary and are intended to explain the principles and spirit of the present invention, rather than to limit the scope of the present invention.

[0064] The present invention will be further explained more clearly and completely below in conjunction with specific implementation methods.

[0065] A high-precision method for calculating the volume of a granary, such as Figure 1 As shown, the following steps are included:

[0066] S1. Data acquisition; use a laser radar device to perform an all-round scan of the target object to obtain a laser point cloud data set; the laser radar device emits a laser beam through the laser radar and receives the reflected signal, calculates the distance using the time difference or phase difference, and obtains data on a large number of target points; for example, the H13 three-dimensional profile scanning laser radar can scan a three-dimensional area covering a large field of view; for example, the H13 three-dimensional profile scanning laser radar can scan a three-dimensional area covering a large field of view and output information such as the height and volume of the material cargo; the laser radar device emits a laser beam through the laser radar and receives the reflected signal, calculates the distance using the time difference or phase difference, and obtains data on a large number of target points;

[0067] S2. Data preprocessing; the data preprocessing includes filtering and denoising, data smoothing, and data registration; filtering and denoising is used to remove irrelevant point cloud data and noise to improve data quality, and specifically, mean filtering and median filtering can be used; in order to reduce outliers and sharp discontinuities, data smoothing is used to make the data more suitable for subsequent analysis;

[0068] S3. Rough segmentation of point cloud: Rough segmentation of the preprocessed point cloud; segment the preprocessed point cloud data into different object parts, where the rough segmentation is the approximate area of ​​the material surface and the granary; specifically, segmentation can be performed based on the shape and color characteristics of the object, and common segmentation algorithms include RANSAC and morphology-based algorithms; the specific method used in this specific implementation is:

[0069] S31. Splitting the point cloud into shapes, where the split shapes include but are not limited to cylinders, cones, and planes;

[0070] S32. Use RANSAC to segment the cylinder and cone. If the cylinder and cone are successfully segmented, combine them to obtain a segmentation result.

[0071] S33. If the cylinder and cone are not segmented in S32, the RANSAC method is used to perform point cloud plane segmentation, and the segmented planes are combined to generate a point cloud segmentation result.

[0072] S4. Fine segmentation of point cloud; use machine learning or deep learning methods to finely segment the point cloud after rough segmentation; this specific implementation uses the PointNet model or the PointNet++ model for fine segmentation. Specifically, the PointNet model and the PointNet++ model are existing technologies and are not described in detail here.

[0073] S5. Point cloud accuracy compensation; perform accuracy compensation on point cloud data, including compensation for the error of laser radar accuracy as it changes with distance and temperature; the error of radar accuracy increases nonlinearly with distance, for example, the scanning distance within 0-5 meters has the highest accuracy, which is 1-1.5 cm, the scanning distance between 5-9 meters has an accuracy of 1.5-2 cm, and the scanning distance above 9 meters has an accuracy of 2-3 cm; similarly, different temperature zones have different effects on the accuracy of laser radar, especially under high temperature conditions, the accuracy of radar fluctuates greatly, that is, temperature drift phenomenon; the present invention fixes the problem of temperature drift accuracy fluctuation through accuracy compensation technology, ensuring high-precision measurement under various temperature conditions. The S5 specifically includes the following steps:

[0074] In the point cloud accuracy compensation in S5, when the laser radar scans the target object, the point cloud accuracy level changes significantly under different distances and temperatures; S5 includes the following steps:

[0075] By comparing the accuracy errors between different parts of the standard object and the corresponding point cloud, an error model function of distance, temperature and accuracy is established;

[0076] The error value of the temperature and distance during measurement is calculated according to the established error model and the error correction is performed;

[0077] Subsequent volume calculation is performed based on the corrected point cloud.

[0078] Furthermore, the specific steps of establishing the error model function of distance, temperature and accuracy in S5 include:

[0079] S51. Collect a large amount of point cloud data in advance and obtain the standard design dimension drawing of the granary;

[0080] S52. Preprocess the collected point cloud data, remove noise in the point cloud data, smooth the point cloud data, calibrate the point cloud data, and locally downsample the calibrated point cloud data; in this example, the number of collected point cloud samples is 3000;

[0081] S53. Performing granary identification and segmentation on the downsampled point cloud to segment the granary body point cloud;

[0082] S54. extract the point cloud information at different positions from the radar optical center and the size information of the corresponding standard granary, and obtain the temperature information of the granary at that time; the number of error information collected at different positions and different temperature conditions is about 10,000;

[0083] S55. Based on the information in step S54, a smooth error model function is constructed in the three-dimensional space. The model function calculates the corresponding error according to the current temperature and distance, and has good flexibility and strong fitting ability. The specific function relationship is as follows:

[0084] ;

[0085] This function achieves its powerful prediction ability through its basis functions, among which, Represents the error model function value, parameter axis , The node vectors are , ; represents the key points of the generated error model, also known as control points, m and n represent the number of key points selected in the x-axis and y-axis directions, respectively. , , d and t represent the collected distance value and temperature value respectively, and is the basis function, and the expression is:

[0086] ;

[0087] This function is a second-order basis function, which can be simply understood as the current basis function value. Calculated from the previous order basis function value, such as , and so on; the role of this basis function is to smoothly fit the model function. At the same time, the basis function is a non-negative function, which can avoid the discontinuity of the model function.

[0088] Furthermore, the calculation of the error value and the error correction of the temperature and distance during measurement according to the established error model described in S5 are specifically as follows:

[0089] According to the current point Generate a lookup point based on the distance and temperature , find the corresponding error value in the model , while searching from the point cloud The normal vector of , adjust the position of the current point according to the normal vector:

[0090] ;

[0091] in, For the current point Corrected point.

[0092] Figure 2 is a distribution diagram of radar accuracy at different distances in this specific implementation mode, Figure 3is the distribution map of radar accuracy after correction; Figure 4 This is the accuracy deviation diagram before temperature compensation in this specific implementation mode. Figure 5 It can be seen from these figures that after this step, the accuracy of the point cloud is significantly improved.

[0093] S6. Volume calculation; the specific steps of S6 include:

[0094] S61. The convex hull model is composed of many triangles; obtain the three vertex coordinates of a specified triangle in the convex hull model and the normal vector corresponding to the triangle face, and the normal vector is recorded as ;

[0095] S62. Move the point cloud along the normal vector of the triangle Perform projection and extract the points projected into the triangle, such as Figure 6 As shown: and calculate the projection distance of each point inside the triangle;

[0096] S63. Figure 7 As shown, the points inside the triangle are meshed (the large triangle is subdivided into many small triangles according to the points inside), and then each vertex of the subdivided triangle is meshed along the normal direction. Move in the opposite direction, and the moving distance is the projection distance of each point;

[0097] S64. Repeat steps S61 to S63 until all triangles in the convex hull model are traversed, completing the fitting of the convex hull model;

[0098] S65. Split the convex hull model after fitting into small tetrahedrons (composed of four points, a triangle and a point outside the triangle), and obtain the volume of the complete model by calculating the volume of each tetrahedron and accumulating them.

[0099] Furthermore, the method further comprises:

[0100] S7.3D rendering presents the various states of the granary (such as feeding, unloading, empty, full, arching, dust) to users in the form of 3D interactive models and animations. Fig. 9 The calculated object volume results are displayed through a graphical interface or other forms to generate a three-dimensional visualization model to display the detailed structure and volume distribution of the object; this is convenient for users to intuitively understand and further analyze. Furthermore, the high-precision material surface generated by algorithm optimization provides a good data basis for judging the state of the granary; through material surface shape analysis, common arch problems (such as Figure 8 and issue early warning to prevent abnormal operation of the granary.

[0101] This specific implementation also discloses a granary volume calculation system based on material surface point cloud compensation using a laser radar. The system is used to calculate the granary volume. Based on the above-mentioned high-precision granary volume calculation method, the system includes:

[0102] Data preprocessing module, used to preprocess point cloud data;

[0103] Point cloud coarse segmentation module, used to perform coarse segmentation on the preprocessed point cloud data;

[0104] Point cloud fine segmentation module: used to perform fine segmentation on the point cloud after rough segmentation;

[0105] Point cloud accuracy compensation module, used to perform accuracy compensation on finely segmented point cloud data;

[0106] The volume calculation module is used to calculate the volume of the point cloud after accuracy compensation.

[0107] Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent replacements for some of the technical features therein; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A high-precision method for calculating the volume of a granary, characterized in that: The steps include: S1. Data collection: Use laser radar equipment to perform a full-scale scan of the target object and obtain point cloud data; S2. Data preprocessing; S3. Rough segmentation of point cloud: Rough segmentation of the preprocessed point cloud; S4. Fine segmentation of point cloud: Use machine learning or deep learning methods to finely segment the point cloud after rough segmentation; S5. Point cloud accuracy compensation: perform accuracy compensation on the finely segmented point cloud data. The accuracy compensation is the error compensation of the laser radar accuracy changing with distance and temperature. S6. Calculate the volume of the point cloud after accuracy compensation; The point cloud rough segmentation of S3 includes: S31. Splitting the point cloud into shapes, where the split shapes include but are not limited to cylinders, cones, and planes; S32. Use RANSAC to segment the cylinder and cone. If the cylinder and cone are successfully segmented, combine them to obtain a segmentation result. S33. If the cylinder and cone are not segmented in S32, the RANSAC method is used to perform point cloud plane segmentation, and the segmented planes are combined to generate a point cloud segmentation result; The point cloud accuracy compensation steps in S5 are: By comparing the accuracy errors between different parts of the standard object and the corresponding point cloud in advance, the error model function of distance, temperature and accuracy is established; The error value of the temperature and distance during measurement is calculated according to the established error model function and the error correction is performed; The volume calculation in S6 is based on the corrected point cloud; The step of establishing the error model function of distance, temperature and accuracy by comparing the accuracy errors between different parts of the standard object and the corresponding point cloud in advance includes the following steps: S51. Collect point cloud data in advance; S52. Preprocessing the collected point cloud data, removing noise in the point cloud data, smoothing the point cloud data, calibrating the point cloud data, and locally downsampling the calibrated point cloud data; S53. Performing granary identification and segmentation on the point cloud after local downsampling, segmenting the granary body point cloud, the granary body point cloud includes the material surface point cloud and the granary wall point cloud; S54. Extracting the point cloud information at different positions from the radar optical center, the temperature information of the granary at the corresponding time, and the size information of the corresponding standard granary; S55. According to the information in step S54, an error model function is constructed in three-dimensional space. The specific function relationship is as follows: ； in, represents the error model function value; represents the key points of the generated error model, m and n represent the number of key points selected in the x-axis and y-axis directions, d and t represent the collected distance value and temperature value, respectively. and is the basis function, and the expression is: ； in, Represents the order of the basis function.

2. A high-precision granary volume calculation method according to claim 1, characterized in that: The point cloud fine segmentation adopts PointNet model or PointNet++ model for fine segmentation.

3. A high-precision granary volume calculation method according to claim 1, characterized in that: The error value calculation and error correction of the temperature and distance during measurement according to the established error model function are specifically as follows: According to the current point The distance and temperature generate a point , find and calculate the corresponding error value in the error model , while finding points from the point cloud The normal vector of , adjust the position of the current point according to the normal vector: ； in, For the current point Corrected point.

4. A high-precision granary volume calculation method according to claim 1, characterized in that: The step S6 is as follows: for the identified irregular object, a convex hull model is calculated for the corrected point cloud, the convex hull model is continuously refined and fitted to the point cloud, and the volume of the three-dimensional model is calculated based on the fitted point cloud.

5. A high-precision granary volume calculation method according to claim 4, characterized in that: The specific steps of S6 include: S61. Get the three vertex coordinates of a specified triangle in the convex hull model and the normal vector corresponding to the triangle face, the normal vector is recorded as ; S62. Move the point cloud along the normal vector of the triangle Perform projection, extract the points projected into the triangle, and calculate the projection distance of each point inside the triangle; S63. Subdivide the points inside the triangle into a mesh, and then mesh each vertex of the subdivided triangle along the normal direction Move in the opposite direction, and the moving distance is the projection distance of each point; S64. Repeat steps S61 to S63 until all triangles in the convex hull model are traversed and the fitting of the convex hull model is completed; S65. Split the fitted convex hull model into small tetrahedrons, calculate the volume of each tetrahedron and perform accumulation operation to obtain the total volume.

6. A high-precision granary volume calculation system, the system is used to calculate the volume of a granary, characterized in that: The system is based on the high-precision granary volume calculation method according to any one of claims 1 to 5, and the system comprises: The data acquisition module is used to transmit a laser beam through the laser radar and receive the reflected signal, calculate the distance using the time difference or phase difference, and obtain point cloud data; Data preprocessing module, used to preprocess point cloud data; Point cloud coarse segmentation module, used to perform coarse segmentation on the preprocessed point cloud data; Point cloud fine segmentation module: used to perform fine segmentation on the point cloud after rough segmentation; Point cloud accuracy compensation module, used to perform accuracy compensation on finely segmented point cloud data; The volume calculation module is used to calculate the volume of the point cloud after accuracy compensation.

Images