Laser point cloud lava pipe skylight detection method and system based on grid division method
Through the detection method based on the grid division method, the problem of inefficiency of traditional artificial visual retrieval is solved, and the automatic detection of lava tube sunroof is realized, which improves detection efficiency and accuracy and simplifies the calculation process.
Patent Information
- Application Number
- CN202410956476.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-11-12
AI Technical Summary
The traditional lava tube skylight recognition method based on laser point cloud relies on manual visual retrieval, with low accuracy and difficulty in meeting the needs of large-scale detection.
Using the detection method based on the mesh division method, the original point cloud data of the lava pipeline is cut, cropped, preprocessed, and projected to a two-dimensional plane and divided into a uniform plane grid, which is based on whether the grid cells contain data points. Classification, traverse grid cells for detection and identification of skylights.
The automatic detection of lava tube sunroof is realized, which improves detection efficiency and accuracy, simplifies the calculation process, reduces unnecessary computing resource consumption, and ensures detection accuracy.
Smart Images

Figure CN118941514A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer graphics, and in particular to a lava tube skylight detection method and system based on a laser point cloud of a grid division method. Background Art
[0002] Lava tubes are pipes formed during the flow of lava after a volcanic eruption. The surface lava will first solidify into a shell of a certain thickness due to heat dissipation and cooling, forming an insulating protective layer, while the internal lava will continue to flow at a high temperature due to slow heat dissipation. When the volcanic eruption weakens and the magma flow stops, the last internal lava rushes out of the front edge restriction, and after flowing out, a hollow pipe is formed, which is the lava tube. The outer shell of the lava tube can withstand adverse environmental conditions such as extreme temperatures on the lunar surface, micrometeorite impacts, and solar radiation, providing a relatively stable and safe living environment. The hollow pipe structure of the lava tube has high space utilization and scalability. The lunar lava tube has been listed as the preferred area for future lunar bases many times.
[0003] Skylights are a special type of micro-topography in lava tubes. They exist at the top of the tubes and are approximately circular openings. They are named after their shape resembling the skylights of a house. The number of skylights on lava tubes is random, the sizes are different, the shapes are diverse, and the distribution is uneven and scattered. At present, most of the known skylights in lava tubes are measured and recorded after field surveys, and this method can only identify a limited number of skylights. LiDAR is a measurement method that quickly obtains the three-dimensional coordinates of a target, and can efficiently and accurately restore the true three-dimensional shape of the measured object. By using LiDAR to scan the lava tube, the detection of skylights inside the lava tube can be quickly completed based on point cloud data, and the position, size and other morphological information of the skylights can be efficiently recorded.
[0004] Traditional skylight recognition methods based on laser point clouds are mostly manual visual retrieval, which has low accuracy and efficiency that cannot meet the needs of large-scale lava tube detection. Summary of the invention
[0005] In order to solve the problems existing in the prior art, the present invention provides a lava tube skylight detection method and system based on laser point cloud of grid division method, which can realize automatic detection of lava tube skylight and improve detection efficiency and detection accuracy.
[0006] The detection method in the embodiment of the present invention specifically adopts the following technical solution: a lava tube skylight detection method based on laser point cloud of grid division method, comprising the following steps:
[0007] S1. Cut and segment the original point cloud data of the lava tube and cut out the top of the tube, including:
[0008] The original point cloud data of the lava tube is cut from its cross section and divided into several lava tube point cloud sections;
[0009] Crop out the top semicircular area of each lava tube point cloud segment;
[0010] S2, performing data preprocessing on the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q;
[0011] S3, projecting the point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P;
[0012] S4, dividing the two-dimensional point set into a uniform plane grid, where the plane grid includes a plurality of grid units;
[0013] S5, classifying the grid cells into “empty holes” and “full holes” according to whether the grid cells contain data points;
[0014] S6. Traverse all grid cells and detect and identify skylights according to the categories of the grid cells and their neighboring grid cells.
[0015] The detection system in the embodiment of the present invention specifically adopts the following technical solution: a lava tube skylight detection system based on laser point cloud of grid division method, including the following modules:
[0016] The cropping module is used to cut and segment the original point cloud data of the lava tube and crop the top of the tube, including:
[0017] The original point cloud data of the lava tube is cut from its cross section into several lava tube point cloud segments; the top semicircular area of each lava tube point cloud segment is cropped;
[0018] A preprocessing module is used to perform data preprocessing on the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q;
[0019] A data conversion module, used for projecting the point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P;
[0020] A grid division module, used for dividing the two-dimensional point set into a uniform plane grid, wherein the plane grid includes a plurality of grid units;
[0021] The classification module is used to classify the grid cells into "empty holes" and "full holes" according to whether the grid cells contain data points;
[0022] The recognition module is used to traverse all grid cells and detect and recognize skylights according to the categories of the grid cells and their neighboring grid cells.
[0023] After adopting the above technical solution, the present invention has the following advantages compared with the prior art:
[0024] 1. Starting from the position and morphological structure of the skylight at the top of the lava tube, the present invention combines the data representation of the skylight in the handheld laser point cloud to extract a feasible automatic detection method for the skylight of the lava tube using handheld laser point cloud. During the detection process, the point cloud at the top of the lava tube is cut out and directly projected onto a two-dimensional plane; the detection method does not change the main morphological features of the skylight, thereby simplifying the calculation process while ensuring the detection effect.
[0025] 2. The present invention divides the data points projected onto the two-dimensional plane into regular grids, and defines two types of grids, "empty holes" and "real holes", based on whether there are data points. The key data representation of the skylight on the point cloud is captured, and the skylight is identified with the help of the easy-to-retrieve regular grid category, thereby reducing unnecessary computing resource consumption and ensuring detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 Schematic diagram of the process of the lava tube skylight detection method in an embodiment of the present invention;
[0027] Figure 2 Schematic diagram of data preprocessing of a lava tube skylight detection method according to an embodiment of the present invention;
[0028] Figure 3 This is a diagram showing the detection results of the three-dimensional point cloud lava tube skylight of the present invention. DETAILED DESCRIPTION
[0029] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0030] Example 1
[0031] This embodiment provides a lava tube skylight detection method based on a laser point cloud grid division method. Figure 1 As shown, the specific process includes the following steps:
[0032] S1. Cut and segment the original point cloud data of the lava tube and cut out the top of the tube.
[0033] In this embodiment, the SLAM100 handheld laser scanner can be used to survey the lava tubes and obtain the point cloud data of the lava tubes in the experimental area. The SLAM100 handheld laser scanner is launched by Pegasus Robotics and can form a 270°×360° point cloud coverage, which can obtain three-dimensional point cloud data of the surrounding environment without light and GPS. The handheld laser scanner was used to perform a full-scale, multi-angle scan of three different lava tubes in a prefecture-level city. The detection lengths of the three lava tubes were 104 meters, 325 meters, and 40 meters, respectively.
[0034] After obtaining the original point cloud data of the lava tube, pre-processing such as cutting, segmentation, and cropping is performed. The specific steps are as follows:
[0035] S11. Cut the original point cloud data of the lava tube from its cross section into several lava tube point cloud segments. The number of points in each point cloud segment is about 15 million.
[0036] Due to the large scale of lava tube data, the clipping tool in the CloudCompare software can be used to cut the point cloud into segments from the cross section of the lava tube to obtain several lava tube point cloud segments.
[0037] S12, cropping the top semicircular area of each lava tube point cloud segment for subsequent skylight detection.
[0038] S2. Preprocess the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q. The preprocessing mainly includes filtering and denoising.
[0039] In this step, for each point in the top semicircular area point cloud, the number of neighboring points is set to 50, and the mean and standard deviation of the point set composed of each point and its neighboring points are calculated. The range is defined with the mean as the center and the standard deviation as the radius. If the point is within the defined range, it is retained; if the point is not within the defined range, it is determined to be an outlier and discarded. After filtering is completed, manual visual recognition is performed to manually trim out the noise points that are not completely filtered.
[0040] Furthermore, if Figure 2 As shown, the specific steps of step S2 are as follows:
[0041] S21, for the top semicircular area point cloud Q of the lava tube point cloud segment 0 Every point q in 0 , given a positive integer k satisfying k>1, as the number of neighborhood points of each point.
[0042] S22, for each point q 0 Query the k points closest to its Euclidean distance to form point q 0The neighborhood point set of .
[0043] S23, for each point q 0 Calculate the standard deviation of its neighborhood point set. If the standard deviation meets the preset threshold requirement, point q 0 Keep, if the standard deviation does not meet the preset threshold requirements, delete point q 0 .
[0044] S24, checking the pipe top point cloud (i.e., the top semicircular area point cloud of the lava tube point cloud section) on which the above operation has been performed, manually trimming and removing the outstanding point cloud noise that still exists, and obtaining the filtered point cloud data Q.
[0045] S3. Project the preprocessed point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P.
[0046] Furthermore, the specific steps of step S3 are as follows:
[0047] Each point q in the preprocessed point cloud data Q i =(x i ,y i ,z i ), projected onto a two-dimensional plane to obtain a two-dimensional point set P, where each point p i =(x i ,y i ).
[0048] That is, in this step, the point cloud data Q is a three-dimensional point cloud, which can be expressed as q = {q 1 ,q 2 …q n}; projected onto the XOY plane, the obtained two-dimensional point set P can be expressed as p = {p 1 ,p 2 …p n}.
[0049] S4. Divide the two-dimensional point set into a uniform plane grid, where the plane grid includes a plurality of grid units.
[0050] Further, the specific steps of step S4 are as follows:
[0051] Traverse all points in the two-dimensional point set P and calculate the maximum value x of the x and y coordinates in the two-dimensional point set P max 、x min ,y max ,y min , determine the search resolution and the number of grid cells m used to divide the uniform plane grid; based on the number of grid cells m, calculate the side length L of the grid cell, the formula is as follows:
[0052]
[0053] After determining the side length of the grid unit, calculate the number of grid units x in the x direction respectively. _num and the number of grid cells in the y direction y _num :
[0054]
[0055] In this embodiment, the number m of grid units may be 150.
[0056] S5. Classify the grid cells into “empty holes” and “full holes” according to whether the grid cells contain data points.
[0057] Furthermore, the specific steps of step S5 are as follows:
[0058] S51. Traverse all data points p in the two-dimensional point set P i =(x i ,y i ), and put the data points into the grid cells (u, v) where the data points are located in sequence; the grid cells are allocated according to the coordinate values of each point in the two-dimensional point set P and the side length of the grid cell, and the formula is as follows:
[0059]
[0060] where u∈[0,x _num ],v∈[0,y _num ].
[0061] S52, after allocating the data points to the grid cells in sequence, determine whether each grid cell contains a data point, and classify the grid cells into two categories: "empty holes" and "full holes" accordingly. If the grid cell contains a data point, it is classified as a "full hole", and if it does not contain a data point, it is classified as an "empty hole".
[0062] S6. Traverse all grid cells and detect and identify skylights according to the categories of the grid cells and their neighboring grid cells.
[0063] Furthermore, the specific steps of step S6 are as follows:
[0064] S61, traversing each grid cell in turn to determine whether the grid cell is a "real hole". If it is a "real hole", continue to determine the categories of the eight neighboring grid cells of the grid cell. If there is one or more "empty holes" among the eight neighboring grid cells, determine that the grid cell is a skylight grid.
[0065] S62, restoring the data points in the skylight grid to three-dimensional space to obtain the final result of the skylight detection.
[0066] At this point, the detection of the lava tube skylight using the handheld laser point cloud is completed. The detection results are as follows: Figure 3 shown.
[0067] Example 2
[0068] Based on the same inventive concept as Example 1, this embodiment provides a lava tube skylight detection system based on a laser point cloud based on a grid division method, comprising the following modules:
[0069] The cropping module is used to cut and segment the original point cloud data of the lava tube and crop the top of the tube, including:
[0070] The original point cloud data of the lava tube is cut from its cross section into several lava tube point cloud segments; the top semicircular area of each lava tube point cloud segment is cropped;
[0071] A preprocessing module is used to perform data preprocessing on the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q;
[0072] A data conversion module, used for projecting the point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P;
[0073] A grid division module, used for dividing the two-dimensional point set into a uniform plane grid, wherein the plane grid includes a plurality of grid units;
[0074] The classification module is used to classify the grid cells into "empty holes" and "full holes" according to whether the grid cells contain data points;
[0075] The recognition module is used to traverse all grid cells and detect and recognize skylights according to the categories of the grid cells and their neighboring grid cells.
[0076] In this embodiment, the preprocessing of the preprocessing module includes filtering and denoising. For each point in the point cloud of the top semicircular area, the number of neighborhood points is set, and the mean and standard deviation of the point set composed of each point and its neighborhood points are calculated; a range is defined with the mean as the center and the standard deviation as the radius. If the point is within the defined range, the point is retained; if the point is not within the defined range, it is determined to be an outlier and discarded.
[0077] The preprocessing process of the preprocessing module specifically includes:
[0078] For the top semicircular area of the lava tube point cloud segment, point cloud Q 0 Every point q in 0 , given a positive integer k satisfying k>1, as the number of neighboring points of each point;
[0079] For each point q 0 Query the k points closest to its Euclidean distance to form point q 0The neighborhood point set of ;
[0080] For each point q 0 Calculate the standard deviation of its neighborhood point set. If the standard deviation meets the preset threshold requirement, point q 0 Keep, if the standard deviation does not meet the preset threshold requirements, delete point q 0 ;
[0081] Check the point cloud of the top semicircular area that has been retained or deleted, and manually trim and remove the prominent point cloud noise to obtain the filtered point cloud data.
[0082] In addition, the plane meshing process in the meshing module includes the following steps:
[0083] Traverse all points in the two-dimensional point set P and calculate the maximum value x of the x and y coordinates in the two-dimensional point set P max 、x min ,y max ,y min , determine the search resolution and the number of grid cells m used to divide the uniform plane grid; based on the number of grid cells m, calculate the side length L of the grid cell:
[0084]
[0085] Calculate the number of grid cells x in the x direction respectively _num and the number of grid cells in the y direction y _num :
[0086]
[0087] The classification process of grid cells in the classification module includes the following steps:
[0088] Traverse all data points p in the two-dimensional point set P i =(x i ,y i ), and put the data points into the grid cells (u, v) where the data points are located in sequence; the grid cells are allocated according to the coordinate values of each point in the two-dimensional point set P and the side length of the grid cell:
[0089]
[0090] where u∈[0,x _num ],v∈[0,y _num ].
[0091] After allocating the data points to the grid cells in turn, it is determined whether each grid cell contains a data point, and the grid cells are classified accordingly; if the grid cell contains a data point, it is classified as a "real hole", and if it does not contain a data point, it is classified as an "empty hole".
[0092] The above modules of this embodiment are respectively used to execute the corresponding steps in Embodiment 1, and the specific implementation process can refer to Embodiment 1.
[0093] The present invention can realize fast and efficient lava tube skylight detection of point cloud data, and provide users with skylight distribution and morphological information of lava tube exploration areas.
[0094] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A lava tube skylight detection method based on laser point cloud using a grid division method, characterized in that: The following steps are involved: S1. Cut and segment the original point cloud data of the lava tube and cut out the top of the tube, including: Cut the original point cloud data of the lava tube from its cross section and divide it into several lava tube point cloud segments: Crop out the top semicircular area of each lava tube point cloud segment; S2, performing data preprocessing on the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q; S3, projecting the point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P; S4, dividing the two-dimensional point set into a uniform plane grid, where the plane grid includes a plurality of grid units; S5. Classify the grid cells into "empty holes" and "full holes" according to whether the grid cells contain data points; S6. Traverse all grid cells and detect and identify skylights according to the categories of the grid cells and their neighboring grid cells.
2. The lava tube skylight detection method according to claim 1, characterized in that: The preprocessing of step S2 includes filtering and denoising. For each point in the point cloud of the top semicircular area, the number of neighborhood points is set, and the mean and standard deviation of the point set composed of each point and its neighborhood points are calculated; a range is defined with the mean as the center and the standard deviation as the radius. If the point is within the defined range, the point is retained; if the point is not within the defined range, it is determined to be an outlier and discarded.
3. The lava tube skylight detection method according to claim 2, characterized in that: Step S2 includes the following steps: S21. For each point q0 in the top semicircular area point cloud Q0 of the lava tube point cloud segment, a positive integer k is given to satisfy k>1 as the number of neighboring points of each point; S22, for each point q0, query the k points that are closest to its Euclidean distance to form the neighborhood point set of point q0; S23, for each point q0, calculate the standard deviation of its neighborhood point set, if the standard deviation meets the preset threshold requirement, then the point q0 is retained, if the standard deviation does not meet the preset threshold requirement, then the point q0 is deleted; S24, checking the point cloud of the top semicircular area on which the retention or deletion operation has been performed, and manually cutting out and removing the outstanding point cloud noise that still exists, to obtain filtered point cloud data.
4. The lava tube skylight detection method according to claim 1, characterized in that: Step S4 includes the following steps: Traverse all points in the two-dimensional point set P and calculate the maximum value x of the x and y coordinates in the two-dimensional point set P max 、x min ,y max ,y min , determine the search resolution and the number of grid cells m used to divide the uniform plane grid; based on the number of grid cells m, calculate the side length L of the grid cell: Calculate the number of grid cells x in the x direction respectively _num and the number of grid cells in the y direction y -num :
5. The lava tube skylight detection method according to claim 4, characterized in that: Step S5 includes the following steps: S51. Traverse all data points p in the two-dimensional point set P i =(x i ,y i ), and put the data points into the grid cells (u, v) where the data points are located in sequence; the grid cells are allocated according to the coordinate values of each point in the two-dimensional point set P and the side length of the grid cell: where u∈[0,x -num ],v∈[0,y -num ]; S52, after allocating the data points to the grid cells in sequence, determine whether each grid cell contains a data point, and classify the grid cells accordingly; if the grid cell contains a data point, classify it as a "real hole", if not, classify it as an "empty hole".
6. The lava tube skylight detection method according to claim 1, characterized in that: Step S6 includes the following steps: S61, traversing each grid cell in turn to determine whether the grid cell is a "real hole". If it is a "real hole", continue to determine the category of the neighboring grid cells of the grid cell. If there is one or more "empty holes" in the neighboring grid cells, determine that the grid cell is a skylight grid; S62, restoring the data points in the skylight grid to three-dimensional space to obtain the final result of the skylight detection.
7. A lava tube skylight detection system based on laser point cloud of grid division method, characterized in that: Includes the following modules: The cropping module is used to cut and segment the original point cloud data of the lava tube and crop the top of the tube, including: The original point cloud data of the lava tube is cut from its cross section into several lava tube point cloud segments; the top semicircular area of each lava tube point cloud segment is cropped; A preprocessing module is used to perform data preprocessing on the top semicircular area of the cropped lava tube point cloud segment to obtain preprocessed point cloud data Q; A data conversion module, used for projecting the point cloud data Q onto a two-dimensional plane to obtain a two-dimensional point set P; A grid division module, used for dividing the two-dimensional point set into a uniform plane grid, wherein the plane grid includes a plurality of grid units; The classification module is used to classify the grid cells into "empty holes" and "full holes" according to whether the grid cells contain data points; The recognition module is used to traverse all grid cells and detect and recognize skylights according to the categories of the grid cells and their neighboring grid cells.
8. The lava tube skylight detection system according to claim 7, characterized in that: The preprocessing in the preprocessing module includes filtering and denoising. For each point in the point cloud in the top semicircular area, the number of neighborhood points is set, and the mean and standard deviation of the point set composed of each point and its neighborhood points are calculated; a range is defined with the mean as the center and the standard deviation as the radius. If the point is within the defined range, the point is retained; if the point is not within the defined range, it is determined to be an outlier and discarded.
9. The lava tube skylight detection system according to claim 8, characterized in that: The preprocessing process of the preprocessing module includes: For each point q0 in the top semicircular area point cloud Q0 of the lava tube point cloud segment, a positive integer k is given to satisfy k>1 as the number of neighboring points of each point; For each point q0, query the k points with the closest Euclidean distance to it to form the neighborhood point set of point q0; For each point q0, calculate the standard deviation of its neighborhood point set. If the standard deviation meets the preset threshold requirement, point q0 is retained. If the standard deviation does not meet the preset threshold requirement, point q0 is deleted. Check the point cloud of the top semicircular area that has been retained or deleted, and manually trim and remove the prominent point cloud noise to obtain the filtered point cloud data.
10. The lava tube skylight detection system according to claim 7, characterized in that: The process of plane meshing in the meshing module includes the following steps: Traverse all points in the two-dimensional point set P and calculate the maximum value x of the x and y coordinates in the two-dimensional point set P max 、x min ,y max ,y min , determine the search resolution and the number of grid cells m used to divide the uniform plane grid; based on the number of grid cells m, calculate the side length L of the grid cell: Calculate the number of grid cells x in the x direction respectively _num and the number of grid cells in the y direction y -num : The classification process of grid cells in the classification module includes the following steps: Traverse all data points p in the two-dimensional point set P i =(x i ,y i ), and put the data points into the grid cells (u, v) where the data points are located in sequence; the grid cells are allocated according to the coordinate values of each point in the two-dimensional point set P and the side length of the grid cell: where u∈[0,x -num ],v∈[0,y -num ]; After allocating the data points to the grid cells in turn, it is determined whether each grid cell contains a data point, and the grid cells are classified accordingly; if the grid cell contains a data point, it is classified as a "real hole", and if it does not contain a data point, it is classified as an "empty hole".
Citation Information
Patent Citations
Volume calculation method, system and equipment based on point cloud data in sinkhole and medium
CN117876465A
