Method and system for checking three-dimensional point cloud image data

Abstract

The invention provides a method and a system for checking three-dimensional point cloud image data, and relates to the technical field of three-dimensional point cloud image data processing, and the method comprises the steps: carrying out the coordinate system calibration of the three-dimensional point cloud image data, and obtaining the three-dimensional point cloud image data passing the coordinate system calibration; preprocessing the three-dimensional point cloud image data; the flatness of the preprocessed three-dimensional point cloud image data is calculated, and whether the flatness meets a preset flatness standard or not is judged; counting the number of noisy points of the three-dimensional point cloud image data of which the flatness meets a preset flatness standard, and judging whether the number of noisy points is within a preset number threshold range or not; counting the number of noisy points of the three-dimensional point cloud image data of which the flatness meets a preset flatness standard, and judging whether the number of noisy points is within a preset number threshold range or not; and recording points corresponding to the plane normal vectors as abnormal points, and checking the three-dimensional point cloud image data according to the number of the abnormal points. According to the invention, the accuracy and reliability of three-dimensional point cloud image data detection are ensured, the calculation complexity of the detection process is effectively reduced, and the three-dimensional point cloud image data detection speed is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional point cloud image data processing, and in particular to a three-dimensional point cloud image data verification method and system. Background Art

[0002] With the advancement of three-dimensional imaging technology, coordinate system calibration and three-dimensional point cloud image data quality detection play a core role in three-dimensional data processing. They are indispensable for ensuring the accuracy and reliability of three-dimensional point cloud image data, ensuring that three-dimensional point cloud image data from different sources can be accurately aligned, and effectively eliminating noise and errors, thereby improving the quality of three-dimensional point cloud image data. Therefore, it is of great significance to study the detection method of three-dimensional point cloud image data quality.

[0003] At present, 3D data processing technology has made significant progress, but in the 3D data inspection process, most processes only focus on the deviation inspection of the coordinate system, while ignoring the comprehensive evaluation of the 3D point cloud image data, resulting in that even if the coordinate system is aligned correctly, the noise, redundancy, missing or uneven density in the 3D point cloud image data are still not discovered and corrected in time, thereby affecting the accuracy and reliability of subsequent 3D modeling, analysis or application. The prior art discloses a 3D point cloud quality assessment method and system, which obtains point cloud information, inputs it into a preset noise simulation function, generates random point cloud noise; constructs a point cloud scoring database based on the generated random point cloud noise and the preset point cloud quality assessment method; constructs a point cloud quality assessment model based on the point cloud scoring database; inputs the point cloud information into the point cloud quality assessment model, and outputs the quality score of the point cloud as the point cloud quality assessment result, but this method uses the constructed cloud quality assessment model to evaluate the quality of the 3D point cloud image data, and the cloud quality assessment model needs to be trained and optimized with a large amount of data, and the detection process has high computational complexity, which affects the quality detection speed of the 3D point cloud image data. Summary of the invention

[0004] In order to solve the problems that the above-mentioned existing technologies have poor accuracy and reliability in three-dimensional point cloud image data detection, the detection process has high computational complexity, and affects the detection speed of three-dimensional point cloud image data, the present invention proposes a three-dimensional point cloud image data verification method and system, which ensures the accuracy and reliability of three-dimensional point cloud image data detection by performing comprehensive detection on the three-dimensional point cloud image data, effectively reduces the computational complexity of the detection process, and improves the detection speed of three-dimensional point cloud image data.

[0005] In order to achieve the above technical effects, the technical solution of the present invention is as follows: A method for checking three-dimensional point cloud image data comprises the following steps: S1. calibrating the coordinate system of the three-dimensional point cloud image data to obtain the three-dimensional point cloud image data after the coordinate system calibration; S2. Preprocessing the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; S3. Calculate the flatness of the preprocessed three-dimensional point cloud image data, and determine whether the flatness meets the preset flatness standard. If so, the flatness check passes, and S4 is executed; if not, the flatness check fails; S4. Count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within the preset number threshold range. If so, the noise point number check passes, and S5 is executed; if not, the noise point number check fails; S5. Obtain the plane normal vector and the reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and determine whether the deviation value between the plane normal vector and the reference normal vector is within the deviation range. If so, the plane normal vector check passes, and S6 is executed; if not, the normal vector check fails; S6. The point corresponding to the plane normal vector is recorded as an outlier point, the number of the outlier points is counted, and the three-dimensional point cloud image data is verified according to the number of the outlier points to obtain a verification result of the three-dimensional point cloud image data.

[0006] Preferably, the coordinate system calibration of the three-dimensional point cloud image data includes: S11. Using a calibration plate with a pattern, aligning the calibration plate with a world coordinate system; S12. Using a camera to identify the pattern of the calibration plate and calculate a standard reference matrix from the camera coordinate system of the calibration plate to the world coordinate system; S13. Compare the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration, and check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the standard range of the coordinate system deviation. If so, the coordinate system calibration of the three-dimensional point cloud image data is passed; if not, the coordinate system calibration of the three-dimensional point cloud image data is failed.

[0007] Preferably, the standard reference matrix and the latest recognition matrix are both 4x4 homogeneous matrices, and the calculation expression of the standard reference matrix is:

[0008] in, represents the standard reference matrix determined in a theoretical or preset calibration process, represents a 3x3 rotation matrix, Represents a 3x1 translation vector; The calculation expression of the latest recognition matrix is:

[0009] in, It indicates the latest recognition matrix calculated by recognizing the calibration plate pattern in real time. represents the 3x3 calibration rotation matrix, Represents a 3x1 calibration translation vector.

[0010] Preferably, comparing the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration to check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the coordinate system deviation standard range includes: S131. Based on the translation vector and the calibration translation vector, calculate the translation vector difference between the translation vector and the calibration translation vector in the X, Y, and Z directions to obtain the translation vector difference results between the translation vector and the calibration translation vector in the X, Y, and Z directions; S132. Based on the rotation matrix and the calibration rotation matrix, respectively calculate the rotation difference between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes to obtain the rotation difference results between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes; S133. If the translation vector difference result exceeds the translation vector difference threshold or the rotation difference result exceeds the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​outside the standard range of coordinate system deviation; if the translation vector difference result does not exceed the translation vector difference threshold or the rotation difference result does not exceed the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​within the standard range of coordinate system deviation.

[0011] Preferably, the preprocessing of the three-dimensional point cloud image data includes: cropping the three-dimensional point cloud image data according to the actual size of the calibration plate, and retaining the three-dimensional point cloud image data on the plane part of the calibration plate as the preprocessed three-dimensional point cloud image data.

[0012] Preferably, calculating the flatness of the preprocessed three-dimensional point cloud image data and judging whether the flatness meets a preset flatness standard comprises: S31. Define a plane passing through the origin , and the normal vector is The plane is the standard plane; S32. Calculate each point P ( x , y ,z ) to the basic distance of the standard plane d as follows:

[0013] S33. Based on the distance d , calculate the average distance as follows:

[0014] in For point P( x , y , z ) S34. Based on the average distance , calculate the first standard deviation as follows:

[0015] S35. If the first standard deviation Greater than the first standard deviation T1 , then the flatness does not meet the preset flatness standard; if then, the flatness meets the preset flatness standard.

[0016] Preferably, counting the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard includes: S41. Obtaining points in the three-dimensional point cloud image data whose flatness meets the preset flatness standard , by searching for the point A point within a sphere or cube with the center as the center. The number of points in the neighborhood; S42. Calculation point The average distance to other points in the neighborhood as follows:

[0017] in, Indicate point The number of points in the domain; Indicate point and Point The Euclidean distance between i Represents a specific point in the 3D point cloud image data, that is, the center point under investigation; Represents the i-th point in the three-dimensional point cloud image data; j Indicates that this represents any other point in the 3D point cloud image data, which may be a point One of the neighbors; Represents a set including all points points in the neighborhood of ; S43. Based on the average distance , calculate the second standard deviation as follows:

[0018] S44. Based on the second standard deviation , calculate the outlier factor as follows:

[0019] in, represents the standard deviation ratio; S45. Calculation point To point The distance between the nearest neighbor points in the neighborhood, if the distance is greater than the outlier factor , then point is identified as a noise point; otherwise, the point Not recognized as noise; S46. Count all points The number of noise points identified as noise points .

[0020] Preferably, obtaining the plane normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and determining whether a deviation value between the plane normal vector and a preset normal vector is within a deviation range, comprises: S51. Perform a dot product operation on the plane normal vector and the reference normal vector to obtain a dot product result; S52. Based on the dot product result, calculate the angle between the plane normal vector and the reference normal vector; S53. If the angle is not greater than the angle threshold, the deviation value between the plane normal vector and the preset normal vector is within the deviation range; otherwise, the deviation value between the plane normal vector and the preset normal vector exceeds the deviation range.

[0021] The present invention also proposes a three-dimensional point cloud image data verification system, comprising: A coordinate system calibration module is used to perform coordinate system calibration on the three-dimensional point cloud image data to obtain three-dimensional point cloud image data that has passed the coordinate system calibration; A preprocessing module, used to preprocess the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; A flatness checking module is used to calculate the flatness of the preprocessed three-dimensional point cloud image data and determine whether the flatness meets a preset flatness standard. If so, the flatness check passes; if not, the flatness check fails; A noise point checking module is used to count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within a preset number threshold range. If so, the noise point number check passes; if not, the noise point number check fails; A normal vector checking module is used to obtain a plane normal vector and a reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and to determine whether a deviation value between the plane normal vector and the reference normal vector is within a deviation range. If so, the plane normal vector check passes; if not, the normal vector check fails; The verification module is used to record the points corresponding to the plane normal vectors as abnormal points, count the number of the abnormal points, and verify the three-dimensional point cloud image data according to the number of the abnormal points to obtain the verification result of the three-dimensional point cloud image data.

[0022] The present invention also proposes a computer device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the operation of the three-dimensional point cloud image data verification method.

[0023] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: The present invention proposes a method and system for verifying 3D point cloud image data, firstly, coordinate system calibration is performed on the 3D point cloud image data, and the 3D point cloud image data is preprocessed after the coordinate system calibration is passed, and then the flatness of the preprocessed 3D point cloud image data is calculated and the flatness is checked; then the number of noise points of the 3D point cloud image data whose flatness meets a preset flatness standard is checked, and then the plane normal vector of each point in the 3D point cloud image data whose number of noise points is within a preset number threshold range is checked, and after the plane normal vector check is passed, the 3D point cloud image data is verified according to the number of abnormal points to obtain the 3D point cloud image data verification result; the present invention verifies the 3D point cloud through automated coordinate system calibration and 3D point cloud image data quality verification process. The image data is fully inspected to ensure the accuracy and reliability of the 3D point cloud image data detection, and data deviations caused by equipment aging, environmental interference and other factors are prevented, thereby ensuring the quality and consistency of the data in subsequent applications; the 3D point cloud image data quality verification process is implemented in sequence through flatness inspection, noise point quantity inspection, plane normal vector inspection and 3D point cloud image data verification, which effectively reduces the computational complexity of the detection process and improves the 3D point cloud image data detection speed; the present invention also significantly improves the stability and predictability of data application results through strict quality control of coordinate system calibration and 3D point cloud image data quality verification, ensuring that data changes in any field will not lead to deviations in application results, thereby providing users with a stable and reliable 3D data acquisition and processing tool. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart showing a method for checking three-dimensional point cloud image data proposed in an embodiment of the present invention; Figure 2 A structural diagram showing a calibration plate proposed in an embodiment of the present invention; Figure 3 A structural diagram showing a three-dimensional point cloud image data verification system proposed in an embodiment of the present invention; Figure 4 A block diagram of a computer device proposed in an embodiment of the present invention is shown.

[0025] 401. Processor; 402. Memory; 403. Communication interface; 404. Communication bus; 405. Executable instructions. DETAILED DESCRIPTION

[0026] The drawings are for illustrative purposes only and should not be construed as limiting the present patent; It is understandable to those skilled in the art that some well-known contents may be omitted in the drawings; The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0027] Example 1 like Figure 1 As shown, this embodiment proposes a method for checking three-dimensional point cloud image data, comprising the following steps: S1. calibrating the coordinate system of the three-dimensional point cloud image data to obtain the three-dimensional point cloud image data after the coordinate system calibration; In S1, see Figure 1 and Figure 2 , the coordinate system calibration of the three-dimensional point cloud image data includes: S11. Using a calibration plate with a pattern, aligning the calibration plate with a world coordinate system; S12. Using a camera to identify the pattern of the calibration plate and calculate a standard reference matrix from the camera coordinate system of the calibration plate to the world coordinate system; wherein the camera is a 3D depth camera; S13. Compare the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration, and check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the coordinate system deviation standard range. If so, the coordinate system calibration of the three-dimensional point cloud image data is passed; if not, the coordinate system calibration of the three-dimensional point cloud image data is failed; In S13, the standard reference matrix and the latest recognition matrix are both 4x4 homogeneous matrices, representing the transformation matrix from the camera's coordinate system to the world coordinate system represented by the calibration plate placement position. The calculation expression of the standard reference matrix is:

[0028] in, represents the standard reference matrix determined in a theoretical or preset calibration process, represents a 3x3 rotation matrix, Represents a 3x1 translation vector; The latest identification matrix is ​​obtained through an actual calibration process, and the calculation expression of the latest identification matrix is:

[0029] in, It indicates the latest recognition matrix calculated by recognizing the calibration plate pattern in real time. represents the 3x3 calibration rotation matrix, Represents a 3x1 calibration translation vector.

[0030] The step of comparing the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration to check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the coordinate system deviation standard range includes: S131. Based on the translation vector and the calibration translation vector, calculate the translation vector difference between the translation vector and the calibration translation vector in the X, Y, and Z directions to obtain the translation vector difference results between the translation vector and the calibration translation vector in the X, Y, and Z directions; In S131, the calculation expression of the translation vector difference result is as follows:

[0031]

[0032]

[0033] in, Represents the translation vector difference result in the X direction, Represents the translation vector difference result in the Y direction, Represents the translation vector difference result in the Z direction, It represents the translation component along the X direction in the standard reference matrix determined in the theoretical or preset calibration process. This value is obtained and stored from the previous calibration process as a reference standard. represents the translation component of the standard reference matrix in the Y direction, represents the component of the standard reference matrix in the Z direction, It indicates the translation component in the X direction of the latest recognition matrix calculated by recognizing the calibration plate pattern in real time. This value is calculated in real time from the latest calibration process. represents the translation component of the latest recognition matrix along the Y direction, Represents the translation component of the latest identified matrix along the Z direction; S132. Based on the rotation matrix and the calibration rotation matrix, respectively calculate the rotation difference between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes to obtain the rotation difference results between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes; In S132, the calculation expression of the rotation difference result is as follows:

[0034]

[0035]

[0036] The angle calculation function is as follows: in, Represents the rotation difference result around the X axis, Represents the rotation difference result around the Y axis, Represents the rotation difference result around the Z axis, represents the rotation matrix around the X axis in the standard reference matrix, represents the rotation matrix around the Y axis in the standard reference matrix, represents the rotation matrix around the Z axis in the standard reference matrix, represents the calibration rotation matrix around the X axis in the latest recognition matrix, represents the calibration rotation matrix around the Y axis in the latest recognition matrix, represents the calibration rotation matrix around the Z axis in the latest recognition matrix, is the angle calculation function, It represents the angle value converted from the rotation matrix around each axis in the standard reference matrix to the axis angle form. Indicates the angle value of the rotation matrix around each axis in the latest identified matrix converted into the axis-angle form; S133. If the translation vector difference result exceeds the translation vector difference threshold or the rotation difference result exceeds the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​outside the standard range of coordinate system deviation; if the translation vector difference result does not exceed the translation vector difference threshold or the rotation difference result does not exceed the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​within the standard range of coordinate system deviation.

[0037] S2. Preprocessing the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; In S2, the preprocessing of the three-dimensional point cloud image data includes: cropping the three-dimensional point cloud image data according to the real size of the calibration plate, and retaining the three-dimensional point cloud image data on the plane part of the calibration plate as the preprocessed three-dimensional point cloud image data.

[0038] S3. Calculate the flatness of the preprocessed three-dimensional point cloud image data, and determine whether the flatness meets the preset flatness standard. If so, the flatness check passes, and S4 is executed; if not, the flatness check fails; and the verification of the three-dimensional point cloud image data ends; In S3, a smaller standard deviation value means that the point cloud is more evenly distributed on the plane and has a higher flatness; the calculating the flatness of the preprocessed three-dimensional point cloud image data and judging whether the flatness meets the preset flatness standard include: S31. Define a plane passing through the origin , and the normal vector is The plane is the standard plane; S32. Calculate each point P ( x , y , z ) to the basic distance of the standard plane d as follows:

[0039] S33. Based on the distance d , calculate the average distance as follows:

[0040] in For point P( x , y , z ) S34. Based on the average distance , calculate the first standard deviation as follows:

[0041] S35. If the first standard deviation Greater than the first standard deviation T1 , then the flatness does not meet the preset flatness standard; if then, the flatness meets the preset flatness standard.

[0042] S4. Count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within the preset number threshold range. If so, the noise point number check passes, and S5 is executed; if not, the noise point number check fails; and the verification of the three-dimensional point cloud image data ends; In S4, local statistics are used to identify points far away from other points, which are considered as noise points; and counting the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard includes: S41. Obtaining points in the three-dimensional point cloud image data whose flatness meets the preset flatness standard , by searching for the point A point within a sphere or cube with the center as the center. The number of points in the neighborhood; S42. Calculation point The average distance to other points in the neighborhood as follows:

[0043] in, Indicate point The number of points in the domain; Indicate point and Point The Euclidean distance between i Represents a specific point in the 3D point cloud image data, that is, the center point under investigation; Represents the i-th point in the three-dimensional point cloud image data; j Indicates that this represents any other point in the 3D point cloud image data, which may be a point One of the neighbors; Represents a set including all points points in the neighborhood of ; S43. Based on the average distance , calculate the second standard deviation as follows:

[0044] S44. Based on the second standard deviation , calculate the outlier factor as follows:

[0045] in, Represents the ratio of standard deviations; Use the ratio of standard deviations To define the outlier factor , usually a value greater than 1, such as 2 or 3, indicating how many standard deviations away from the mean a point is considered an anomaly: S45. Calculation point To point The distance between the nearest neighbor points in the neighborhood, if the distance is greater than the outlier factor , then point is identified as a noise point; otherwise, the point Not recognized as noise; S46. Count all points The number of noise points identified as noise points ; when When the number of noise points is outside the preset number threshold range, , the number of noise points is within a preset number threshold range; S5. Obtain the plane normal vector and the reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within the preset number threshold range, and determine whether the deviation value between the plane normal vector and the reference normal vector is within the deviation range. If so, the plane normal vector check passes, and S6 is executed; if not, the normal vector check fails; and the verification of the three-dimensional point cloud image data ends; In S5, the obtaining of the plane normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and determining whether a deviation value between the plane normal vector and a preset normal vector is within a deviation range, includes: S51. Perform a dot product operation on the plane normal vector and the reference normal vector to obtain a dot product result; Each point in the three-dimensional point cloud image data described in S5 is The plane normal vector is , the reference normal vector is recorded as , dot product result The calculation expression is: = ) S52. Based on the dot product result, calculate the angle between the plane normal vector and the reference normal vector; Calculating the angle between the plane normal vector and the reference normal vector includes: Using the dot product result , the cosine of the angle between the plane normal vector and the reference normal vector is calculated as follows:

[0046] is the modulus of the plane normal vector, are the modulus of the reference normal vector respectively; Magnitude of the plane normal vector The calculation expression is as follows:

[0047] The modulus of the reference normal vector The calculation expression is as follows:

[0048] The cosine value of the angle is converted into the corresponding angle as follows:

[0049] S53. If the angle Not greater than the angle threshold , the deviation value between the plane normal vector and the preset normal vector is within the deviation range; otherwise, the deviation value between the plane normal vector and the preset normal vector exceeds the deviation range, and the point corresponding to the plane normal vector is regarded as an abnormal point; S6. The point corresponding to the plane normal vector is recorded as an outlier point, the number of the outlier points is counted, and the three-dimensional point cloud image data is verified according to the number of the outlier points to obtain a verification result of the three-dimensional point cloud image data.

[0050] In S6, the number of abnormal points is counted ,when When , the three-dimensional point cloud image data quality verification result fails; when The three-dimensional point cloud image data quality verification result passes.

[0051] In the present embodiment, firstly, a coordinate system calibration is performed on the 3D point cloud image data. After the coordinate system calibration is passed, the 3D point cloud image data is preprocessed. Secondly, the flatness of the preprocessed 3D point cloud image data is calculated and the flatness is checked. Then, the number of noise points of the 3D point cloud image data whose flatness meets a preset flatness standard is checked. Then, the plane normal vector of each point in the 3D point cloud image data whose number of noise points is within a preset number threshold range is checked. After the plane normal vector check is passed, the 3D point cloud image data is checked according to the number of abnormal points to obtain the 3D point cloud image data verification result. The present invention comprehensively calibrates the 3D point cloud image data through an automated coordinate system calibration and 3D point cloud image data quality verification process. Detection ensures the accuracy and reliability of 3D point cloud image data detection, prevents data deviation caused by equipment aging, environmental interference and other factors, thereby ensuring the quality and consistency of data in subsequent applications; the 3D point cloud image data quality verification process is implemented in sequence through flatness inspection, noise number inspection, plane normal vector inspection and 3D point cloud image data verification, which effectively reduces the computational complexity of the detection process and improves the 3D point cloud image data detection speed; the present invention also significantly improves the stability and predictability of data application results through strict quality control of coordinate system calibration and 3D point cloud image data quality verification, ensuring that data changes in any field will not lead to deviations in application results, thereby providing users with a stable and reliable 3D data acquisition and processing tool.

[0052] Example 2 See also Figure 3 This embodiment proposes a three-dimensional point cloud image data verification system, including: A coordinate system calibration module is used to perform coordinate system calibration on the three-dimensional point cloud image data to obtain three-dimensional point cloud image data that has passed the coordinate system calibration; A preprocessing module, used to preprocess the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; A flatness checking module is used to calculate the flatness of the preprocessed three-dimensional point cloud image data and determine whether the flatness meets a preset flatness standard. If so, the flatness check passes; if not, the flatness check fails; A noise point checking module is used to count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within a preset number threshold range. If so, the noise point number check passes; if not, the noise point number check fails; A normal vector checking module is used to obtain a plane normal vector and a reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and to determine whether a deviation value between the plane normal vector and the reference normal vector is within a deviation range. If so, the plane normal vector check passes; if not, the normal vector check fails; The verification module is used to record the points corresponding to the plane normal vectors as abnormal points, count the number of the abnormal points, and verify the three-dimensional point cloud image data according to the number of the abnormal points to obtain the point cloud verification result.

[0053] Example 3 See also Figure 4 , this embodiment further proposes a computer device, including: a processor 401, a memory 402, a communication interface 403 and a communication bus 404, wherein the processor 401, the memory 402 and the communication interface 403 communicate with each other through the communication bus 404; Wherein: the processor 401, the memory 402 and the communication interface 403 communicate with each other through the communication bus 404. The communication interface 403 is used for network communication with other devices such as a client or other servers. The processor 401 is used to execute the executable instruction 405, which can specifically execute the relevant steps in the embodiment of the three-dimensional point cloud image data verification method.

[0054] Specifically, the executable instructions 405 may include program codes. The processor 401 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0055] The memory 402 is used to store executable instructions 405. The memory 402 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0056] The executable instruction 405 may be specifically called by the processor 401 to cause the computer device to perform the following operations: S1. calibrating the coordinate system of the three-dimensional point cloud image data to obtain the three-dimensional point cloud image data after the coordinate system calibration; S2. Preprocessing the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; S3. Calculate the flatness of the preprocessed three-dimensional point cloud image data, and determine whether the flatness meets the preset flatness standard. If so, the flatness check passes, and S4 is executed; if not, the flatness check fails; and the verification of the three-dimensional point cloud image data ends; S4. Count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within the preset number threshold range. If so, the noise point number check passes, and S5 is executed; if not, the noise point number check fails; and the verification of the three-dimensional point cloud image data ends; S5. Obtain the plane normal vector and the reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within the preset number threshold range, and determine whether the deviation value between the plane normal vector and the reference normal vector is within the deviation range. If so, the plane normal vector check passes, and S6 is executed; if not, the normal vector check fails; and the verification of the three-dimensional point cloud image data ends; S6. The point corresponding to the plane normal vector is recorded as an outlier point, the number of the outlier points is counted, and the three-dimensional point cloud image data is verified according to the number of the outlier points to obtain a verification result of the three-dimensional point cloud image data.

[0057] In the present embodiment, firstly, a coordinate system calibration is performed on the 3D point cloud image data. After the coordinate system calibration is passed, the 3D point cloud image data is preprocessed. Secondly, the flatness of the preprocessed 3D point cloud image data is calculated and the flatness is checked. Then, the number of noise points of the 3D point cloud image data whose flatness meets a preset flatness standard is checked. Then, the plane normal vector of each point in the 3D point cloud image data whose number of noise points is within a preset number threshold range is checked. After the plane normal vector check is passed, the 3D point cloud image data is checked according to the number of abnormal points to obtain the 3D point cloud image data verification result. The present invention comprehensively calibrates the 3D point cloud image data through an automated coordinate system calibration and 3D point cloud image data quality verification process. Detection ensures the accuracy and reliability of 3D point cloud image data detection, prevents data deviation caused by equipment aging, environmental interference and other factors, thereby ensuring the quality and consistency of data in subsequent applications; the 3D point cloud image data quality verification process is implemented in sequence through flatness inspection, noise number inspection, plane normal vector inspection and 3D point cloud image data verification, which effectively reduces the computational complexity of the detection process and improves the 3D point cloud image data detection speed; the present invention also significantly improves the stability and predictability of data application results through strict quality control of coordinate system calibration and 3D point cloud image data quality verification, ensuring that data changes in any field will not lead to deviations in application results, thereby providing users with a stable and reliable 3D data acquisition and processing tool.

[0058] Obviously, the above embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for checking three-dimensional point cloud image data, characterized in that: The following steps are involved: S1. calibrating the coordinate system of the three-dimensional point cloud image data to obtain the three-dimensional point cloud image data after the coordinate system calibration; S2. Preprocessing the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; S3. Calculate the flatness of the preprocessed three-dimensional point cloud image data, and determine whether the flatness meets the preset flatness standard. If so, the flatness check passes, and S4 is executed; if not, the flatness check fails; S4. Count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within the preset number threshold range. If so, the noise point number check passes, and S5 is executed; if not, the noise point number check fails; S5. Obtain the plane normal vector and the reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and determine whether the deviation value between the plane normal vector and the reference normal vector is within the deviation range. If so, the plane normal vector check passes, and S6 is executed; if not, the normal vector check fails; S6. The point corresponding to the plane normal vector is recorded as an outlier point, the number of the outlier points is counted, and the three-dimensional point cloud image data is verified according to the number of the outlier points to obtain a verification result of the three-dimensional point cloud image data.

2. The method for checking three-dimensional point cloud image data according to claim 1, characterized in that: The coordinate system calibration of the three-dimensional point cloud image data includes: S11. Using a calibration plate with a pattern, aligning the calibration plate with a world coordinate system; S12. Using a camera to identify the pattern of the calibration plate and calculate a standard reference matrix from the camera coordinate system of the calibration plate to the world coordinate system; S13. Compare the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration, and check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the standard range of the coordinate system deviation. If so, the coordinate system calibration of the three-dimensional point cloud image data is passed; if not, the coordinate system calibration of the three-dimensional point cloud image data is failed.

3. The method for checking three-dimensional point cloud image data according to claim 2, characterized in that: The standard reference matrix and the latest recognition matrix are both 4x4 homogeneous matrices, and the calculation expression of the standard reference matrix is: in, represents the standard reference matrix determined in a theoretical or preset calibration process, represents a 3x3 rotation matrix, Represents a 3x1 translation vector; The calculation expression of the latest recognition matrix is: in, It indicates the latest recognition matrix calculated by recognizing the calibration plate pattern in real time. represents the 3x3 calibration rotation matrix, Represents a 3x1 calibration translation vector.

4. The method for checking three-dimensional point cloud image data according to claim 3, characterized in that: The step of comparing the latest identification matrix obtained according to the real-time placement position identification with the standard reference matrix saved during calibration to check whether the coordinate system deviation between the standard reference matrix and the latest identification matrix is ​​within the coordinate system deviation standard range includes: S131. Based on the translation vector and the calibration translation vector, calculate the translation vector difference between the translation vector and the calibration translation vector in the X, Y, and Z directions to obtain the translation vector difference results between the translation vector and the calibration translation vector in the X, Y, and Z directions; S132. Based on the rotation matrix and the calibration rotation matrix, respectively calculate the rotation difference between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes to obtain the rotation difference results between the rotation matrix and the calibration rotation matrix around the X, Y, and Z axes; S133. If the translation vector difference result exceeds the translation vector difference threshold or the rotation difference result exceeds the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​outside the standard range of coordinate system deviation; if the translation vector difference result does not exceed the translation vector difference threshold or the rotation difference result does not exceed the rotation difference threshold, the coordinate system deviation between the standard reference matrix and the latest identified matrix is ​​within the standard range of coordinate system deviation.

5. The method for checking three-dimensional point cloud image data according to claim 4, characterized in that: The preprocessing of the three-dimensional point cloud image data includes: cutting the three-dimensional point cloud image data according to the real size of the calibration plate, and retaining the three-dimensional point cloud image data on the plane part of the calibration plate as the preprocessed three-dimensional point cloud image data.

6. The method for checking three-dimensional point cloud image data according to claim 1, characterized in that: The calculating the flatness of the preprocessed three-dimensional point cloud image data and judging whether the flatness meets a preset flatness standard include: S31. Define a plane passing through the origin , and the normal vector is The plane is the standard plane; S32. Calculate each point P ( x , y , z ) to the basic distance of the standard plane d as follows: S33. Based on the distance d , calculate the average distance as follows: in For point P( x , y , z ) S34. Based on the average distance , calculate the first standard deviation as follows: S35. If the first standard deviation Greater than the first standard deviation T1 , then the flatness does not meet the preset flatness standard; if then, the flatness meets the preset flatness standard.

7. The method for checking three-dimensional point cloud image data according to claim 1, characterized in that: The counting of the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard includes: S41. Obtaining points in the three-dimensional point cloud image data whose flatness meets the preset flatness standard , by searching for the point A point within a sphere or cube with the center as the center. The number of points in the neighborhood; S42. Calculation point The average distance to other points in the neighborhood as follows: in, Indicate point The number of points in the domain; Indicate point and Point The Euclidean distance between i Represents a specific point in the 3D point cloud image data, that is, the center point under investigation; Represents the i-th point in the three-dimensional point cloud image data; j Indicates that this represents any other point in the 3D point cloud image data, which may be a point One of the neighbors; Represents a set including all points points in the neighborhood of ; S43. Based on the average distance , calculate the second standard deviation as follows: S44. Based on the second standard deviation , calculate the outlier factor as follows: in, represents the standard deviation ratio; S45. Calculation point To point The distance between the nearest neighbor points in the neighborhood, if the distance is greater than the outlier factor , then point is identified as a noise point; otherwise, the point Not recognized as noise; S46. Count all points The number of noise points identified as noise points .

8. The method for checking three-dimensional point cloud image data according to claim 1, characterized in that: The obtaining of the plane normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and determining whether a deviation value between the plane normal vector and a preset normal vector is within a deviation range, comprises: S51. Perform a dot product operation on the plane normal vector and the reference normal vector to obtain a dot product result; S52. Based on the dot product result, calculate the angle between the plane normal vector and the reference normal vector; S53. If the angle is not greater than the angle threshold, the deviation value between the plane normal vector and the preset normal vector is within the deviation range; otherwise, the deviation value between the plane normal vector and the preset normal vector exceeds the deviation range.

9. A three-dimensional point cloud image data verification system, characterized in that: include: A coordinate system calibration module is used to perform coordinate system calibration on the three-dimensional point cloud image data to obtain three-dimensional point cloud image data that has passed the coordinate system calibration; A preprocessing module, used to preprocess the three-dimensional point cloud image data to obtain preprocessed three-dimensional point cloud image data; A flatness checking module is used to calculate the flatness of the preprocessed three-dimensional point cloud image data and determine whether the flatness meets a preset flatness standard. If so, the flatness check passes; if not, the flatness check fails; A noise point checking module is used to count the number of noise points of the three-dimensional point cloud image data whose flatness meets the preset flatness standard, and determine whether the number of noise points is within a preset number threshold range. If so, the noise point number check passes; if not, the noise point number check fails; A normal vector checking module is used to obtain a plane normal vector and a reference normal vector of each point in the three-dimensional point cloud image data whose number of noise points is within a preset number threshold range, and to determine whether a deviation value between the plane normal vector and the reference normal vector is within a deviation range. If so, the plane normal vector check passes; if not, the normal vector check fails; The verification module is used to record the points corresponding to the plane normal vectors as abnormal points, count the number of the abnormal points, and verify the three-dimensional point cloud image data according to the number of the abnormal points to obtain the verification result of the three-dimensional point cloud image data.

10. A computer device, characterized in that: include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the operation of the three-dimensional point cloud image data verification method as described in any one of claims 1-8.

Images