CN111325666B - Airborne laser point cloud processing method based on variable resolution voxel grid - Google Patents
Airborne laser point cloud processing method based on variable resolution voxel grid Download PDFInfo
- Publication number
- CN111325666B CN111325666B CN202010085191.1A CN202010085191A CN111325666B CN 111325666 B CN111325666 B CN 111325666B CN 202010085191 A CN202010085191 A CN 202010085191A CN 111325666 B CN111325666 B CN 111325666B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- compression
- resolution
- variable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 238000007906 compression Methods 0.000 claims abstract description 48
- 230000006835 compression Effects 0.000 claims abstract description 47
- 238000000034 method Methods 0.000 claims abstract description 20
- 230000011218 segmentation Effects 0.000 claims abstract description 12
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000013144 data compression Methods 0.000 claims description 9
- 230000000903 blocking effect Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 239000004576 sand Substances 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 2
- 238000013145 classification model Methods 0.000 abstract description 2
- 238000007781 pre-processing Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4092—Image resolution transcoding, e.g. by using client-server architectures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention relates to an airborne laser point cloud processing method based on a variable-resolution voxel grid and application thereof. Firstly, the variable resolution point cloud compression algorithm can compress the original laser point cloud into the variable resolution point cloud, and the variable resolution voxel grid obtained based on the regular grid meshing of the variable resolution point cloud can cover a larger plane range and can ensure that a central area has higher resolution, so that different requirements of processing of different terrain scenes on the plane coverage range and the resolution of the voxel grid can be met simultaneously. Secondly, in order to process a voxel grid with larger size and adopt a larger basic network structure, the invention builds a three-dimensional semantic segmentation network with a coding-decoding structure based on the sub-flow sparse convolution. The network has higher data processing speed and smaller video memory occupation. The variable resolution voxel grid can flexibly deal with different terrain scenes, and the robustness of the point cloud classification model is effectively improved.
Description
Technical Field
The invention relates to the field of remote sensing mapping, in particular to an airborne laser point cloud data processing method.
Background
Three-dimensional geographic data serve as important components of the intelligent earth, and a rapid acquisition technology and a data post-processing means of the three-dimensional geographic data are always the research key points in the fields of photogrammetry and remote sensing. The laser scanning technology is used as a real-time and rapid active measuring means, can acquire high-precision three-dimensional point cloud and attached physical attributes of a measured target all day long and all weather, and can acquire three-dimensional information of a sheltered area through vegetation gaps. The research of the point cloud data compression method is an important research subject for nearly two decades, and a plurality of classical and effective preprocessing algorithms are used for realizing the preprocessing of the point cloud data, such as projection transformation, voxel division and the like. The method can be divided into two types according to the difference of the dimensionality of the data after point cloud preprocessing: based on the voxel grid method, the three-dimensional model is regularized to a voxel grid, for example, to a voxel grid (Wu,2015) of 30 × 30 × 30 size or to a voxel grid (Maturana,2015) of 32 × 32 × 32. For better data indexing of the volumetric data, octree can be introduced to construct an octree-based voxel grid (Riegler, 2017; Wang, 2017); ② image-based methods, three-dimensional models are mainly rendered into a series of two-dimensional views or converted into geometric feature maps or panoramas, such as multi-angle views (Su, 2015; Qi, 2016; Shi,2015), geometric feature maps (Sinha, 2016; Nannan Qin, 2018). However, in combination with the existing research methods, the point cloud data preprocessing is still in the exploration phase, which is mainly caused by the diversity of the point clouds (different point cloud spatial distributions, for example, uneven density), the complexity of the scene (the scanned data features are of various types: such as houses, dense forests, artificial features, etc.), and the severe occlusion.
Disclosure of Invention
The invention aims to provide a method for processing data in airborne laser point cloud, which realizes point cloud compression based on variable resolution, reduces the geographic space occupied by the data, ensures the invariance of point cloud information description and provides good preprocessing data for subsequent point cloud classification.
The technical scheme of the invention is that the airborne laser point cloud processing method based on the variable resolution voxel grid comprises the following steps:
step 1, point cloud input and blocking: reading in original point cloud data, and partitioning the point cloud by using a sliding window method, wherein the size of the partition is w multiplied by h, the sliding step length is s, and w is h;
step 2, compression strategy based on equal ratio transformation: dividing each block into a compressed area and an internal area, wherein the center point of each block is PcAnd the radius of the maximum circumscribed circle of the segment is denoted as R1Wherein the inner region is denoted by PcAs a circle center and has a radius of R3The remaining is a compressed area; adopting an equal proportion linear compression method from outside to inside for the point cloud in the compressed area, namely compressing the horizontal coordinate of the point only and keeping the original value for the elevation coordinate; the internal area is not compressed, and the three-dimensional coordinates of the points of the internal area adopt a processing mode of keeping original values;
step 3, performing point cloud compression area data compression for each block: performing data compression on the point cloud in the compressed area according to the compression strategy;
any original point in a given compression region is Pi(Xi，Yi，Zi) The point after compression is P'i(X′i，Y′i，Z′i) Point PiAnd point P'iThe intersection point of the connecting line between the two and the boundary circle of the inner region is Ps(Xs，Ys，Zs) The radius of the outer original boundary circle is R1The radius of the boundary circle after external compression is R2The radius of the boundary circle of the inner region is R3The central point of the boundary circle of the inner region is Pc(Xc，Yc，Zc) Wherein
R2＝0.5×R1(ii) a At the same time, let point PiAnd point PcThe included angle between the connecting line and the horizontal line is alpha, and then the following formula is obtained according to the geometric relationship of similar triangles:
the linear compression is scaled as follows:
wherein,
representative point PiAnd PcThe horizontal distance between the two is calculated as follows:
the above formulas are combined to obtain:
transforming it can result in:
in addition to this, the present invention is,
Xs＝R3×cosα+Xc
combining the formulas to obtain:
in the same way, the following Y 'is obtained'iThe calculation formula of (2):
since only the plane coordinates are compressed, the elevation coordinates of the laser spot before and after compression are unchanged, namely:
Z′i＝Zi
and then regularly gridding the variable-resolution point cloud obtained by compression to obtain a variable-resolution regular voxel grid which is coupled with the variable-resolution point cloud.
Further, the specific implementation manner of step 1 is as follows,
assuming that the geographic spaces occupied by the point cloud in the horizontal direction are X and Y respectively, the sliding step length is s, the size of the intercepting window is w X h, and the minimum value of the horizontal coordinate of the whole data block is X0，y0To obtainIf the obtained block set is P, the block steps are as follows:
(1) transversely sliding the X direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (X)0+n×s，y0)，(x0+n×s+w，y0+ h), n belongs to {0,1,2, … }, until all points in the X direction are blocked, putting the blocked data into P;
(2) longitudinally sliding in the Y direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (x)0，y0+n×s)，(x0+w，y0+ n × s + h), n ∈ {0,1,2, … }, until all points in the Y direction are blocked, and the blocked data is put into P.
The invention also provides an application of the airborne laser point cloud processing method based on the variable resolution voxel grid in the technical scheme on point cloud classification, and the method also comprises the following steps,
step 4, performing regular grid formation on the variable-resolution point cloud obtained by compression to obtain a dual variable-resolution regular voxel grid, and inputting the three-dimensional grid point cloud into a pre-trained encoding-decoding structure semantic segmentation network based on sub-stream sparse convolution to obtain a point cloud classification result of each classification;
and 5, merging the classified partitioned point clouds.
Further, the network structure of the encoding-decoding structure semantic segmentation network based on the sub-stream sparse convolution in the step 4 is as follows,
the coding-decoding structure semantic segmentation network based on the sub-stream sparse convolution is composed of a coding layer component and a decoding layer component; the coding layer component consists of 5 groups of convolutional layers, the number of characteristic graphs output by the 5 groups of convolutional layers is respectively 32, 64, 96, 128 and 160, each group of convolutional layers consists of two convolutional layers with the same structure and a down-sampling convolutional layer, and each convolutional layer is provided with a batch normalization layer and a ReLU layer in front; the structure of the decoding layer part is symmetrical to that of the coding layer part, so that the decoding layer part also comprises 5 groups of convolution layers, and the number of characteristic diagrams output by the 5 groups of convolution layers is respectively 160, 128, 96, 64 and 32; each group of convolution layers consists of a reverse convolution layer for up-sampling and two convolution layers with the same structure; meanwhile, the output characteristics of the convolution layer of the coding layer component are subjected to channel merging with the output characteristics of the convolution layer corresponding to the decoding layer component through a jump connection structure; finally, the output of the decoding layer component is passed to a softmax layer for voxel-by-voxel classification.
The invention directly utilizes the existing mature semantic segmentation network framework in image classification, and can simultaneously avoid the problems of doping and overlapping of points of different classes in a two-dimensional projection graph and the like. Firstly, the variable resolution point cloud compression algorithm can compress the original laser point cloud into the variable resolution point cloud, and the variable resolution voxel grid obtained based on the regular grid meshing of the variable resolution point cloud can cover a larger plane range and can ensure that a central area has higher resolution, so that different requirements of processing of different terrain scenes on the plane coverage range and the resolution of the voxel grid can be met simultaneously. Secondly, in order to process a voxel grid with larger size and adopt a larger basic network structure, the invention builds a three-dimensional semantic segmentation network with a coding-decoding structure based on the sub-flow sparse convolution. The network has higher data processing speed and smaller video memory occupation. The variable resolution voxel grid can flexibly deal with different terrain scenes, and the robustness of the point cloud classification model is effectively improved.
Drawings
FIG. 1 is a schematic diagram of the transformation of point coordinates within a compression zone according to the present invention.
FIG. 2 shows the airborne laser point cloud data compression result of the present invention. (a) The (b) and (c) are all exemplary diagrams of variable resolution compression
FIG. 3 is a semantic segmentation network of an encoding-decoding structure based on UNet-like substream sparse convolution of the present invention.
FIG. 4 shows the classification result of airborne laser point cloud data according to the present invention.
Detailed Description
The method of the present invention will be further described with reference to the accompanying drawings.
The embodiment of the invention provides an airborne laser point cloud data compression method, which specifically comprises the following steps:
step 1, point cloud input and blocking: reading in original point cloud data, and partitioning the point cloud by using a sliding window method. The purpose of blocking the input point cloud based on a sliding window is to split the data with too large an original range into target processing units covering sufficient spatial information.
Reading in original airborne laser point cloud data, and partitioning the point cloud by using a sliding window algorithm. Assuming that the geographic spaces occupied by the point cloud in the horizontal direction are X and Y respectively, the sliding step length is s, the size of the intercepting window is w X h, and the minimum value of the horizontal coordinate of the whole data block is X0，y0And if the obtained block set is P, the blocking steps are as follows:
(3) transversely sliding the X direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (X)0+n×s，y0)，(x0+n×s+w，y0+ h), n ∈ {0,1,2, … }, until all points in the X direction are blocked, and the blocked data is put into P.
(4) Longitudinally sliding in the Y direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (x)0，y0+n×s)，(x0+w，y0+ n × s + h), n ∈ {0,1,2, … }, until all points in the Y direction are blocked, and the blocked data is put into P.
In the present invention, w ═ h.
Step 2, compression strategy based on equal ratio transformation: dividing each block into a compression area and an internal area, and adopting an equal proportion linear compression method from outside to inside for point cloud in the compression area, namely compressing only the horizontal coordinate of the point, and adopting a mode of keeping the original value for the elevation coordinate; the three-dimensional coordinates of the points in the internal region are processed in a manner of keeping original values.
The same piece of point cloud data generally has the same density. The point cloud compression compresses the point cloud representing the spatial context information to a certain range, and the density of the part of point cloud is improved to keep the spatial content represented by the part of point cloud unchanged, but the compression strength cannot be too high, otherwise, the point and point are too close to each other, so that the expression information is lost, for example, vegetation is compressed into a vertical straight line, a building is compressed into a horizontal straight line, and the like. Therefore, considering the situation that the compression rate setting is needed to be carried out on the compression area of the input point cloud, the information loss situation is reduced to the minimum while the spatial distribution of the points is compressed.
The algorithm adopts a linear compression method, namely, an equal proportion linear compression method from outside to inside is adopted for point clouds in a compressed area. In consideration of the particularity of airborne laser point cloud, the difference of different ground object targets is mainly determined by the elevations of the points, so that the algorithm only compresses the horizontal coordinates of the points when compressing, and the elevation coordinates are kept in the original value mode. The processed point cloud can reduce the occupied space on the geographical layer, and simultaneously, the consistency of the contents of all ground objects on the elevation is ensured. In order to ensure that the point cloud resolution of the central area is unchanged, a processing mode of keeping an original value is adopted for the three-dimensional coordinates of the points of the internal area.
Step 3, point cloud compression area data compression: and performing data compression on the point cloud in the compressed area according to the compression strategy.
Any original point in a given compression region is Pi(Xi,Yi,Zi) The point after compression is P'i(X′i,Y′i,Z′i) Point PiAnd point P'iThe intersection point of the connecting line between the two and the inner non-compression boundary circle is Ps(Xs,Ys,Zs) The radius of the outer original boundary circle is R1The radius of the boundary circle after external compression is R2The radius of the boundary circle of the inner region is R3The central point of the boundary circle of the inner region is Pc(Xc,Yc,Zc). Wherein,
R2＝0.5×R1at the same time, let point PiAnd point PcThe angle between the connecting line and the horizontal line is α, and the following formula can be obtained according to the geometric relationship of the similar triangles shown in fig. 1:
the linear compression is scaled as follows:
wherein,
representative point PiAnd PcThe horizontal distance between the two is calculated as follows:
Combining the above formulas can obtain:
transforming it can result in:
furthermore, from the geometric relationship shown in fig. 1, the following formula can be obtained:
Xs＝R3×cosα+Xc
combining the above formulas to obtain:
similarly, the following Y 'can be obtained'iThe calculation formula of (2):
since only the plane coordinates are compressed, the elevation coordinates of the laser spot before and after compression are unchanged, namely:
Z′i＝Zi
by the above compression algorithm, will be at R1-R2Compression of points in range to R2-R3In (1).
And 4, performing regular grid formation on the variable-resolution point cloud obtained by compression to obtain a dual variable-resolution regular voxel grid. Inputting the three-dimensional gridded point cloud into a coding-decoding structure semantic segmentation network based on sub-stream sparse convolution to obtain a point cloud classification result;
after the compressed point cloud is obtained, the data is input into a semantic segmentation network of a pre-trained encoding-decoding structure based on UNet-like substream sparse convolution. The network can carry out feature learning and automatic classification on the input data, and the correctness of the classification result is ensured.
The UNet-like network consists of an encoding layer component and a decoding layer component. The coding layer component comprises 5 groups of convolutional layers, the number of characteristic graphs output by the 5 groups of convolutional layers is respectively 32, 64, 96, 128 and 160, and each group of convolutional layers comprises two convolutional layers with the same structure and one downsampling convolutional layer. Each convolutional layer is preceded by a batch normalization and a ReLU layer. The structure of the decoding layer part is symmetrical to that of the encoding layer part, so that the decoding layer part also comprises 5 sets of convolution layers, and the number of characteristic diagrams output by the 5 sets of convolution layers is 160, 128, 96, 64 and 32 respectively. Each set of convolutional layers consists of one anti-convolutional layer for up-sampling and two convolutional layers with the same structure. Meanwhile, the output characteristics of the convolution layer of the coding layer component and the output characteristics of the convolution layer corresponding to the decoding layer component are subjected to channel merging through a jump connection structure. Finally, the output of the decoding layer component is passed to a softmax layer for voxel-by-voxel classification.
Step 5, merging the classified sub-block point clouds, wherein each sub-block point cloud only takes a classification result which is not compressed in the middle;
and 4, merging the sub-block point cloud classification results obtained in the step 4, taking only the classification results of each sub-block point cloud, which are not compressed in the middle, and finally splicing into a whole block of point cloud.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (4)
1. The airborne laser point cloud processing method based on the variable resolution voxel grid is characterized by comprising the following steps:
step 1, point cloud input and blocking: reading in original point cloud data, and partitioning the point cloud by using a sliding window method, wherein the size of the partition is w multiplied by h, the sliding step length is s, and w is h;
step 2, compression strategy based on equal ratio transformation: dividing each block into a compressed area and an internal area, wherein the center point of each block is PcAnd the radius of the maximum circumscribed circle of the segment is denoted as R1Wherein the inner region is denoted by PcAs a circle center and has a radius of R3The remaining is a compressed area; adopting an equal proportion linear compression method from outside to inside for the point cloud in the compressed area, namely compressing the horizontal coordinate of the point only and keeping the original value for the elevation coordinate; the internal area is not compressed, and the three-dimensional coordinates of the points of the internal area adopt a processing mode of keeping original values;
step 3, performing point cloud compression area data compression for each block: performing data compression on the point cloud in the compressed area according to the compression strategy;
any original point in a given compression region is Pi(Xi,Yi,Zi) The point after compression is P'i(X′i,Y′i,Z′i) Point PiAnd point P'iThe intersection point of the connecting line between the two and the boundary circle of the inner region is Ps(Xs,Ys,Zs) The radius of the outer original boundary circle is R1The radius of the boundary circle after external compression is R2The radius of the boundary circle of the inner region is R3The central point of the boundary circle of the inner region is Pc(Xc,Yc,Zc) Wherein
at the same time, let point PiAnd point PcThe included angle between the connecting line and the horizontal line is alpha, and then the following formula is obtained according to the geometric relationship of similar triangles:
the linear compression is scaled as follows:
wherein,
representative point PiAnd PcThe horizontal distance between the two is calculated as follows:
the above formulas are combined to obtain:
transforming it can result in:
in addition to this, the present invention is,
Xs＝R3×cosα+Xc
combining the formulas to obtain:
by the same token, the following Y is obtainediThe calculation formula of `:
since only the plane coordinates are compressed, the elevation coordinates of the laser spot before and after compression are unchanged, namely:
Z'i＝Zi
and then regularly gridding the variable-resolution point cloud obtained by compression to obtain a variable-resolution regular voxel grid which is coupled with the variable-resolution point cloud.
2. The variable resolution voxel grid-based airborne laser point cloud processing method according to claim 1, characterized in that: the specific implementation of step 1 is as follows,
assuming that the geographic spaces occupied by the point cloud in the horizontal direction are X and Y respectively, the sliding step length is s, the size of the intercepting window is w X h, and the minimum value of the horizontal coordinate of the whole data block is X0，y0And if the obtained block set is P, the blocking steps are as follows:
(1) transversely sliding the X direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (X)0+n×s，y0)，(x0+n×s+w，y0+ h), n belongs to {0,1,2, … }, until all points in the X direction are blocked, putting the blocked data into P;
(2) longitudinally sliding in the Y direction according to the step length s, and taking the point cloud of a fixed window, wherein the coordinates of the upper left corner and the lower right corner of the window are respectively (x)0，y0+n×s)，(x0+w，y0+ n × s + h), n ∈ {0,1,2, … }, until all points in the Y direction are blocked, and the blocked data is put into P.
3. The variable resolution voxel grid-based airborne laser point cloud processing method according to claim 1, characterized in that: the method also comprises the following steps of,
step 4, performing regular grid formation on the variable-resolution point cloud obtained by compression to obtain a dual variable-resolution regular voxel grid, and inputting the three-dimensional grid point cloud into a pre-trained encoding-decoding structure semantic segmentation network based on sub-stream sparse convolution to obtain a point cloud classification result of each classification;
and 5, merging the classified partitioned point clouds.
4. The method for processing the airborne laser point cloud based on the variable resolution voxel grid according to claim 3, wherein: the network structure of the encoding-decoding structure semantic segmentation network based on the sub-stream sparse convolution in the step 4 is as follows,
the coding-decoding structure semantic segmentation network based on the sub-stream sparse convolution is composed of a coding layer component and a decoding layer component; the coding layer component consists of 5 groups of convolutional layers, the number of characteristic graphs output by the 5 groups of convolutional layers is respectively 32, 64, 96, 128 and 160, each group of convolutional layers consists of two convolutional layers with the same structure and a down-sampling convolutional layer, and each convolutional layer is provided with a batch normalization layer and a ReLU layer in front; the structure of the decoding layer part is symmetrical to that of the coding layer part, so that the decoding layer part also comprises 5 groups of convolution layers, and the number of characteristic diagrams output by the 5 groups of convolution layers is respectively 160, 128, 96, 64 and 32; each group of convolution layers consists of a reverse convolution layer for up-sampling and two convolution layers with the same structure; meanwhile, the output characteristics of the convolution layer of the coding layer component are subjected to channel merging with the output characteristics of the convolution layer corresponding to the decoding layer component through a jump connection structure; finally, the output of the decoding layer component is passed to a softmax layer for voxel-by-voxel classification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010085191.1A CN111325666B (en) | 2020-02-10 | 2020-02-10 | Airborne laser point cloud processing method based on variable resolution voxel grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010085191.1A CN111325666B (en) | 2020-02-10 | 2020-02-10 | Airborne laser point cloud processing method based on variable resolution voxel grid |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111325666A CN111325666A (en) | 2020-06-23 |
| CN111325666B true CN111325666B (en) | 2021-11-02 |
Family
ID=71167336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010085191.1A Active CN111325666B (en) | 2020-02-10 | 2020-02-10 | Airborne laser point cloud processing method based on variable resolution voxel grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111325666B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111598803B (en) * | 2020-05-12 | 2023-05-09 | 武汉慧点云图信息技术有限公司 | Point cloud filtering method based on variable-resolution voxel grid and sparse convolution |
| CN112597989B (en) * | 2020-12-18 | 2021-09-14 | 中国科学院上海微系统与信息技术研究所 | Millimeter wave three-dimensional holographic image concealed article detection method and system |
| WO2022213376A1 (en) * | 2021-04-09 | 2022-10-13 | 深圳市大疆创新科技有限公司 | Obstacle detection method and apparatus, and movable platform and storage medium |
| CN116797704B (en) * | 2023-08-24 | 2024-01-23 | 山东云海国创云计算装备产业创新中心有限公司 | Point cloud data processing method, system, device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102419794A (en) * | 2011-10-31 | 2012-04-18 | 武汉大学 | Rapid filtering method for airborne laser point cloud data |
| CN102663237A (en) * | 2012-03-21 | 2012-09-12 | 武汉大学 | Point cloud data automatic filtering method based on grid segmentation and moving least square |
| CN105182359A (en) * | 2015-07-31 | 2015-12-23 | 武汉大学 | Satellite-borne Lidar hyperchaotic compressed sensing high-spatial-resolution imaging method |
| CN110570429A (en) * | 2019-08-30 | 2019-12-13 | 华南理工大学 | Lightweight real-time semantic segmentation method based on three-dimensional point cloud |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10796457B2 (en) * | 2018-06-26 | 2020-10-06 | Intel Corporation | Image-based compression of LIDAR sensor data with point re-ordering |
-
2020
- 2020-02-10 CN CN202010085191.1A patent/CN111325666B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102419794A (en) * | 2011-10-31 | 2012-04-18 | 武汉大学 | Rapid filtering method for airborne laser point cloud data |
| CN102663237A (en) * | 2012-03-21 | 2012-09-12 | 武汉大学 | Point cloud data automatic filtering method based on grid segmentation and moving least square |
| CN105182359A (en) * | 2015-07-31 | 2015-12-23 | 武汉大学 | Satellite-borne Lidar hyperchaotic compressed sensing high-spatial-resolution imaging method |
| CN110570429A (en) * | 2019-08-30 | 2019-12-13 | 华南理工大学 | Lightweight real-time semantic segmentation method based on three-dimensional point cloud |
Non-Patent Citations (12)
| Title |
|---|
| "Deep fusion of multi-view and multimodal representation of ALS point cloud for 3D terrain scene recognition";Nannan Qin et al.;《ISPRS Journal of Photogrammetry and Remote Sensing》;20180326;第205-212页 * |
| "Deep-Learning-Based Classification for DTM Extraction from ALS Point Cloud";Xiangyun Hu et al.;《remote sensing》;20160905;第1-16页 * |
| "DEM Extraction from ALS Point Clouds in Forest Areas via Graph Convolution Network";Jinming Zhang et al.;《remote sensing》;20200103;第1-18页 * |
| "Fast and Accurate Plane Segmentation of Airborne LiDAR Point Cloud Using Cross-Line Elements";Teng Wu et al.;《remote sensing》;20160505;第1-26页 * |
| "Learning Dual Multi-Scale Manifold Ranking for Semantic Segmentation of High-Resolution Images";Mi Zhang et al.;《remote sensing》;20170519;第1-30页 * |
| "Semi-Global Filtering of Airborne LiDAR Data for Fast Extraction of Digital Terrain Models";Xiangyun Hu et al.;《remote sensing》;20150824;第10996-11015页 * |
| "一种结合了栅格化和特征判断的点云压缩方法";胡志胜等;《辽宁工程技术大学学报（自然科学版）》;20150831;第34卷(第8期);第958-962页 * |
| "保留路面特征的车载激光点云非均匀压缩方法";刘如飞等;《遥感信息》;20170228;第32卷(第1期);第100-103页 * |
| "大数据量机载LiDAR点云的快速区域分割处理";罗胜等;《测绘科学技术学报》;20181231;第35卷(第5期);第491-496页 * |
| "机载激光点云与影像数据融合方法的研究";隆华平等;《地理空间信息》;20180228;第16卷(第2期);第14-15页 * |
| "顾及矢量特征的机载LiDAR点云数据压缩方法";杨玲等;《地理与地理信息科学》;20090731;第25卷(第4期);第25-29页 * |
| Grzegorz Józ＇ków et al.."Compression strategies for LiDAR waveform cube".《ISPRS Journal of Photogrammetry and Remote Sensing》.2014,第1-13页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111325666A (en) | 2020-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111325666B (en) | Airborne laser point cloud processing method based on variable resolution voxel grid | |
| US11995886B2 (en) | Large-scale environment-modeling with geometric optimization | |
| CN109242862B (en) | Real-time digital surface model generation method | |
| CN106898045B (en) | Large-area true three-dimensional geographic scene self-adaptive construction method based on SGOG tiles | |
| CN115564926B (en) | Three-dimensional patch model construction method based on image building structure learning | |
| CN114332366A (en) | Digital city single house point cloud facade 3D feature extraction method | |
| CN111950658B (en) | Deep learning-based LiDAR point cloud and optical image priori coupling classification method | |
| CN113378756B (en) | Three-dimensional human body semantic segmentation method, terminal device and storage medium | |
| Zhang et al. | A dual attention neural network for airborne LiDAR point cloud semantic segmentation | |
| CN116543117A (en) | High-precision large-scene three-dimensional modeling method for unmanned aerial vehicle images | |
| CN117115359B (en) | Multi-view power grid three-dimensional space data reconstruction method based on depth map fusion | |
| CN114511846A (en) | Real-time three-dimensional target detection method based on point cloud cross-view feature conversion | |
| CN114170311A (en) | Binocular stereo matching method | |
| CN115546760A (en) | Point cloud sequence data processing method and device, computer equipment and storage medium | |
| CN114067060A (en) | Dense point cloud rapid generation method for three-dimensional modeling | |
| Li et al. | New methodologies for precise building boundary extraction from LiDAR data and high resolution image | |
| Li et al. | A hierarchical deformable deep neural network and an aerial image benchmark dataset for surface multiview stereo reconstruction | |
| CN117809016A (en) | Cloud layer polarization removal orientation method based on deep learning | |
| CN114821074B (en) | Airborne LiDAR point cloud semantic segmentation method, electronic equipment and storage medium | |
| CN116148857A (en) | Low signal-to-noise ratio interference phase unwrapping method based on space four-way memory network | |
| CN116152512A (en) | Height measuring and calculating method based on building shadow restoration | |
| Byrne et al. | VOLA: A Compact Volumetric Format for 3D Mapping and Embedded Systems. | |
| Chen et al. | Ortho-NeRF: generating a true digital orthophoto map using the neural radiance field from unmanned aerial vehicle images | |
| Lv et al. | Fast Digital Orthophoto Generation: A Comparative Study of Explicit and Implicit Methods | |
| Zhang et al. | CURL: Continuous, Ultra-compact Representation for LiDAR |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |