A Dimensionality Reduction Method, Device, Equipment and Storage Medium for Laser Point Cloud Data

Abstract

The present application discloses a method, device, equipment and storage medium for reducing the dimension of lidar point cloud data, which solves the technical problem that although the existing dimension reduction of lidar point cloud data can achieve a certain dimension reduction effect, the dimension reduction effect is not ideal, resulting in high-dimensional data still existing in the dimension-reduced data. Among them, the method includes: decomposing the point cloud data to be dimension-reduced of the lidar according to a preset decomposition criterion to obtain a number of locally decomposed data; using a principal component analysis algorithm to calculate the local coordinates corresponding to each of the locally decomposed data; calculating the corresponding global coordinates according to the local coordinates of each of the locally decomposed data; and calculating the corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data.

Description

Technical Field

[0001] This application relates to the technical field of lidar, and in particular, to a method, device, equipment, and storage medium for reducing the dimension of lidar point cloud data. Background Art

[0002] With the development of technology, lidar, as an active remote sensing tool, is widely used in various industries, such as topographic surveying, atmospheric monitoring, driverless, and other industries.

[0003] The sampling frequency of lidar is generally as high as 30 kHz, resulting in a large amount of data in the radar system. Therefore, it is necessary to reduce the dimension of the lidar point cloud data. Although the existing dimension reduction for lidar point cloud data can achieve a certain dimension reduction effect, the dimension reduction effect is not ideal, resulting in high-dimensional data still existing in the data after dimension reduction.

[0004] Therefore, providing an effective method for reducing the dimension of lidar point cloud data is a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0005] This application provides a method, device, equipment, and storage medium for reducing the dimension of lidar point cloud data, which can effectively reduce the dimension of the lidar point cloud data, and solves the technical problem that although the existing dimension reduction for lidar point cloud data can achieve a certain dimension reduction effect, the dimension reduction effect is not ideal, resulting in high-dimensional data still existing in the data after dimension reduction.

[0006] In view of this, the first aspect of this application provides a method for reducing the dimension of lidar point cloud data, including:

[0007] Decompose the lidar point cloud data to be dimension-reduced according to a preset decomposition criterion to obtain a number of locally decomposed data;

[0008] Use the principal component analysis algorithm to calculate the local coordinates corresponding to each of the locally decomposed data;

[0009] Calculate the corresponding global coordinates according to the local coordinates of each of the locally decomposed data;

[0010] Calculate the corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data.

[0011] Optionally, the decomposition criterion includes: data full coverage of the point cloud data to be dimension-reduced and overlap between the locally decomposed data;

[0012] The step of decomposing the lidar point cloud data to be dimension-reduced according to a preset decomposition criterion to obtain a number of locally decomposed data specifically includes:

[0013] Obtain the lidar point cloud data to be dimension-reduced;

[0014] Decompose all the data in the point cloud data to be dimension-reduced to obtain several local decomposition data, where one local decomposition data overlaps with at least one local decomposition data.

[0015] Optionally, using the principal component analysis algorithm to calculate the local coordinates corresponding to each of the local decomposition data specifically includes:

[0016] Calculate the centered matrix corresponding to each of the local decomposition data;

[0017] Perform singular value decomposition on each of the centered matrices to obtain the corresponding decomposition matrix;

[0018] Calculate the corresponding local coordinates according to the decomposition matrix corresponding to each of the local decomposition data.

[0019] Optionally, the calculating the corresponding local coordinates according to the decomposition matrix corresponding to each of the local decomposition data specifically includes:

[0020] Based on the first preset formula, calculate the corresponding local coordinates according to the decomposition matrix corresponding to each of the local decomposition data, and the first preset formula is:

[0021]

[0022] In the formula, Θ m is the local coordinate of the local decomposition data X m , is the centered matrix of the local decomposition data X m , U m,d is the matrix composed of the first d column vectors of U m , U m is the decomposition matrix of the local decomposition data X m .

[0023] Optionally, the calculating the corresponding global coordinates according to the local coordinates of each of the local decomposition data specifically includes:

[0024] Based on the second preset formula, calculate the corresponding global coordinates according to the local coordinates of each of the local decomposition data, and the second preset formula is:

[0025]

[0026] In the formula, Θ m is the local coordinate of the local decomposition data X m , is the global coordinate of the local decomposition data X m , A m is the rotation and scaling matrix.

[0027] Optionally, calculating the corresponding low-dimensional data according to the global coordinates of each of the local decomposition data specifically includes:

[0028] Obtaining the conversion coefficient between the low-dimensional data and the global coordinates;

[0029] Based on the third preset formula, calculating the corresponding low-dimensional data according to the global coordinates of each of the local decomposition data, and the third preset formula is:

[0030]

[0031] In the formula, is the low-dimensional data of the local decomposition data X m , is the global coordinate of the local decomposition data X m , w m,i is the conversion coefficient between the low-dimensional data and the global coordinate .

[0032] Optionally, the expression of the conversion coefficient w m,i is:

[0033]

[0034] In the formula, N m is the number of high-dimensional data included in the local decomposition data X m , is the centering matrix, is the all-ones vector, s m,i is the selection matrix, where the element in the N m -th row and the i-th column is 1, and the others are 0, Θ m is the local coordinate of the local decomposition data X m , is the right pseudo-inverse of Θ m .

[0035] The second aspect of the present application provides a method for dimensionality reduction of lidar point cloud data, including:

[0036] A decomposition unit, configured to decompose the point cloud data to be dimensionally reduced of the lidar according to a preset decomposition criterion to obtain a plurality of local decomposition data;

[0037] A first calculation unit, configured to calculate the corresponding local coordinates of each of the local decomposition data by using a principal component analysis algorithm;

[0038] A second calculation unit, configured to calculate the corresponding global coordinates according to the local coordinates of each of the local decomposition data;

[0039] A third computing unit, configured to calculate corresponding low-dimensional data according to the global coordinates of each piece of the locally decomposed data.

[0040] A third aspect of the present application provides a dimensionality reduction device for lidar point cloud data. The device includes a processor and a memory.

[0041] The memory is configured to store program code and transmit the program code to the processor.

[0042] The processor is configured to execute the dimensionality reduction method for lidar point cloud data according to any one of the instructions in the program code in the first aspect.

[0043] A fourth aspect of the present application provides a storage medium, which is configured to store program code, and the program code is used to execute the dimensionality reduction method for lidar point cloud data according to any one of the first aspects.

[0044] As can be seen from the above technical solutions, the present application has the following advantages:

[0045] In the dimensionality reduction method for lidar point cloud data in the present application, first, the point cloud data to be dimensionally reduced of the lidar is decomposed according to a preset decomposition criterion to obtain a plurality of locally decomposed data. Then, the principal component analysis algorithm is used to calculate the local coordinates corresponding to each piece of the locally decomposed data. Next, according to the local coordinates of each piece of the locally decomposed data, the corresponding global coordinates are calculated. Finally, according to the global coordinates of each piece of the locally decomposed data, the corresponding low-dimensional data is calculated. The present application can effectively reduce the dimension of lidar point cloud data, and solves the technical problem that although the dimensionality reduction of existing lidar point cloud data can achieve a certain dimensionality reduction effect, the dimensionality reduction effect is not ideal, resulting in high-dimensional data still existing in the dimensionally reduced data. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 It is a schematic flowchart of Embodiment 1 of a dimensionality reduction method for lidar point cloud data in an embodiment of the present application;

[0048] Figure 2 It is a schematic flowchart of Embodiment 2 of a dimensionality reduction method for lidar point cloud data in an embodiment of the present application;

[0049] Figure 3 It is a schematic structural diagram of an embodiment of a dimensionality reduction device for lidar point cloud data in an embodiment of the present application. Specific Embodiments

[0050] The embodiments of the present application provide a method, device, equipment, and storage medium for dimensionality reduction of lidar point cloud data, which solve the technical problem that although the existing dimensionality reduction of lidar point cloud data can achieve a certain dimensionality reduction effect, the dimensionality reduction effect is not ideal, resulting in high-dimensional data still existing in the data after dimensionality reduction.

[0051] To make the objectives, features, and advantages of the present application more obvious and understandable, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0052] Please refer to Figure 1 , Figure 1 , which is a schematic flowchart of Embodiment 1 of a method for dimensionality reduction of lidar point cloud data in the embodiments of the present application.

[0053] A method for dimensionality reduction of lidar point cloud data in this embodiment may specifically include the following steps:

[0054] Step 101: Decompose the point cloud data to be dimensionally reduced by the lidar according to a preset decomposition criterion to obtain a number of locally decomposed data.

[0055] After the point cloud data to be dimensionally reduced is decomposed into a number of locally decomposed data, each locally decomposed data is dimensionally reduced respectively. After each locally decomposed data is dimensionally reduced, the dimensionality reduction of the point cloud data to be dimensionally reduced can be completed.

[0056] Step 102: Use the principal component analysis algorithm to calculate the local coordinates corresponding to each locally decomposed data.

[0057] When dimensionally reducing each locally decomposed data, first calculate the local coordinates corresponding to each locally decomposed data. It can be understood that the local coordinates only consider the individual dimensionality reduction of each locally decomposed data itself.

[0058] Step 103: Calculate the corresponding global coordinates according to the local coordinates of each locally decomposed data.

[0059] After calculating the local coordinates of each locally decomposed data, it is also necessary to consider the relationship between each locally decomposed data and other locally decomposed data. Therefore, in this embodiment, the relationship between each locally decomposed data and other locally decomposed data is calculated through the global coordinates of each locally decomposed data.

[0060] Step 104: Calculate the corresponding low-dimensional data according to the global coordinates of each locally decomposed data.

[0061] After calculating the global coordinates of each local decomposition data, the low-dimensional data of each local decomposition data can be calculated. After the low-dimensional data of each local decomposition data are all calculated, the dimensionality reduction of the point cloud data to be dimensionally reduced is completed.

[0062] In the dimensionality reduction method of the lidar point cloud data in this embodiment, first, the point cloud data to be dimensionally reduced of the lidar is decomposed according to a preset decomposition criterion to obtain a number of local decomposition data. Then, the principal component analysis algorithm is used to calculate the local coordinates corresponding to each local decomposition data. Then, according to the local coordinates of each local decomposition data, the corresponding global coordinates are calculated. Finally, according to the global coordinates of each local decomposition data, the corresponding low-dimensional data are calculated. This application can effectively reduce the dimension of the lidar point cloud data, solving the technical problem that although the existing dimensionality reduction of the lidar point cloud data can achieve a certain dimensionality reduction effect, the dimensionality reduction effect is not ideal, resulting in high-dimensional data still existing in the data after dimensionality reduction.

[0063] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of the second embodiment of a dimensionality reduction method for lidar point cloud data in an embodiment of this application.

[0064] The dimensionality reduction method of a lidar point cloud data in this embodiment may specifically include the steps:

[0065] Step 201, obtain the point cloud data to be dimensionally reduced of the lidar.

[0066] It can be understood that the point cloud data to be dimensionally reduced may be the original data corresponding to the lidar or the data after the original data deletes the outliers. This embodiment does not make specific limitations on this.

[0067] Assume that the point cloud data to be dimensionally reduced in this embodiment is X = [x 1 …x N ∈ R D×N , where D is the dimension of the point cloud data to be dimensionally reduced, and N is the number of data in the point cloud data to be dimensionally reduced.

[0068] Let Y = [y 1 …y N ∈ R d×N represent the low-dimensional data after dimensionality reduction of the high-dimensional data set X. Here, d << D, and d is the dimension of the low-dimensional data. Since the low-dimensional data and the high-dimensional data are in one-to-one correspondence, therefore, after the high-dimensional data set X is locally decomposed, the low-dimensional data Y is also correspondingly locally decomposed: Y 1 , …, Y M , where the local of the dimensionality reduction data contains the dimensionality reduction data corresponding to the local X m of the high-dimensional data, and m = 1, …, M.

[0069] Step 202: Decompose all the data in the point cloud data to be dimension-reduced to obtain a number of local decomposition data, where one local decomposition data overlaps with at least one local decomposition data.

[0070] By decomposing the point cloud data to be dimension-reduced, a number of local decomposition data can be obtained. Specifically, during the decomposition, there is an overlap between one local decomposition data and at least one other local data. At the same time, all the data in the point cloud data to be dimension-reduced need to be decomposed, that is, full data coverage.

[0071] Suppose the decomposed local point cloud data is: X 1 ,…,X M , where here N m represents the number of high-dimensional data included in X m , m = 1,…,M. Since the aforementioned decomposition is a full coverage, so

[0072] Step 203: Calculate the centered matrix corresponding to each local decomposition data.

[0073] Solve for the centered matrix of the local decomposition data X m

[0074] Since the local decomposition data so the centered matrix of the local decomposition data X m is

[0075]

[0076] In the formula, is the centering matrix, is a vector of all 1s, is the center of X m . That is, through the centering matrix centering the local decomposition data X m , the obtained centered matrix is

[0077] It should be noted that the centered matrix refers to the matrix obtained by translating the high-dimensional data to the origin as the center. N Here is the center of X m , is the centering matrix. That is, centering is geometrically equivalent to translating the local decomposition data X m ​Translate the center to the origin of the coordinate axes. For example, for a column vector [1, 3, 5], after centering, it becomes [-2, 0, 2].

[0078] Step 204: Perform singular value decomposition on each centered matrix to obtain the corresponding decomposition matrix.

[0079] For each local decomposition data X m of the centered matrix perform singular value decomposition to obtain the decomposition matrix. Specifically, solve the SVD decomposition of:

[0080]

[0081] where U m ∈R D×D and are both standard orthogonal matrices.

[0082] Step 205: Calculate the corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data.

[0083] It can be understood that in one implementation, calculating the corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data specifically includes:

[0084] Based on the first preset formula, calculate the corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data. The first preset formula is:

[0085]

[0086] where Θ m is the local coordinate of the local decomposition data X m , is the centered matrix of the local decomposition data X m , U m,d is the matrix composed of the first d column vectors of U m , and U m is the decomposition matrix of the local decomposition data X m .

[0087] Let U m,d ∈R D×d be the matrix composed of the first d column vectors of U m . In theory, the column vectors of U m,d are the standard orthogonal basis of the tangent space of the midpoint m of X . Thus, the coordinates of the projection of X m on the tangent space are:

[0088]

[0089] It should be noted that the center of the local coordinate Θ m is 0, that is, the origin of the d-dimensional Euclidean space R d .

[0090] Step 206: Based on the second preset formula, calculate the corresponding global coordinates according to the local coordinates of each local decomposition data.

[0091] In one implementation, the above-mentioned second preset formula is:[[]]

[0092]

[0093] In the formula, Θ m is the local coordinate of the local decomposition data X m , is the global coordinate of the local decomposition data X m , A m is the rotation and scaling matrix. It can be understood that Y m is the global coordinate of the local decomposition data X m , is the matrix after centering Y m . Geometrically, it is equivalent to moving Y m to the data centered at the origin through the centering matrix . Y m is the global coordinate,[[]] is also the global coordinate, but only after centering processing, without changing the geometric relationship structure inside Y m .

[0094] The local coordinate Θ m only considers the separate dimensionality reduction of X m . During the dimensionality reduction process, the relationship between X m and other local decomposition data is not considered. Therefore, Θ m is not the global coordinate Y m of X m . However, Θ m and Y m are both datasets in the d-dimensional Euclidean space R d , and both are derived from X m . Therefore, there must be a certain relationship between them. Assuming the relationship between Θ m and Y m is an affine relationship, that is:[[]]

[0095] Here represents the matrix after centering Y m . Geometrically, it is equivalent to translating Y m to the origin of the d-dimensional Euclidean space R d , so that The center of coincides with Θ m The center coincides, A m ∈R d×d is a rotation and scaling matrix. Geometrically interpreted, it is equivalent to rotating and scaling Θ m and approaching So there is:

[0096]

[0097] Here is the right pseudo-inverse of Θ m So there is:

[0098]

[0099] In the formula, represents a data point in the low-dimensional space, is the local remaining data points divided by as the center, represents the selection vector, where the i-th element is 1 and the other elements are 0, i = 1,..., N m .

[0100] Step 207, obtain the conversion coefficient between the low-dimensional data and the global coordinates.

[0101] It should be noted that the expression of the conversion coefficient w m,i is:

[0102]

[0103] In the formula, N m is the number of high-dimensional data contained in the local decomposition data X m , s m,i is the selection matrix where the element in the N m -th row and the i-th column is 1 and the others are 0, Θ m is the local coordinate of the local decomposition data X m , is the right pseudo-inverse of Θ m .

[0104] {w m,i | i = 1,..., N m} is the local linear pattern learned under the local homeomorphism criterion. Through the prediction of the i-th low-dimensional coordinate formed by and its neighboring points can be obtained.

[0105] Step 208, based on the third preset formula, calculate the corresponding low-dimensional data according to the global coordinates of each local decomposition data.

[0106] Among them, the third preset formula is:

[0107]

[0108] In the formula, is the low-dimensional data X m of the local decomposition data X, is the global coordinate of the local decomposition data X m of w, m,i is the low-dimensional data and the global coordinate conversion coefficient.

[0109] Among them, w m,i is the linear mode of the local coordinate Y m of, is the selection matrix, where the element in the Nth m row and the ith column is 1, and the others are 0. Therefore, the objective function can be written as:

[0110]

[0111] Among them, the above problem can be transformed into a Rayleigh quotient problem, and the d orthonormal eigenvectors of the matrix L m corresponding to the smallest eigenvalue form the column vectors of the global coordinate matrix Y.

[0112] In this embodiment, by solving the objective function, the global coordinates, that is, the low-dimensional data representation, can be obtained.

[0113] In the dimensionality reduction method of the lidar point cloud data in this embodiment, first, according to a preset decomposition criterion, the lidar point cloud data to be dimensionally reduced is decomposed to obtain a number of local decomposition data, then the principal component analysis algorithm is used to calculate the local coordinates corresponding to each local decomposition data, and then according to the local coordinates of each local decomposition data, the corresponding global coordinates are calculated, and finally, according to the global coordinates of each local decomposition data, the corresponding low-dimensional data is calculated. This application can effectively reduce the dimensionality of lidar point cloud data, and solves the technical problem that although the dimensionality reduction of existing lidar point cloud data can achieve a certain dimensionality reduction effect, the dimensionality reduction effect is not ideal, resulting in high-dimensional data still existing in the dimensionally reduced data.

[0114] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of an embodiment of a dimensionality reduction device for lidar point cloud data in an embodiment of this application.

[0115] A dimensionality reduction device for lidar point cloud data in this embodiment may specifically include:

[0116] A decomposition unit, configured to decompose the lidar point cloud data to be dimensionally reduced according to a preset decomposition criterion to obtain a number of local decomposition data;

[0117] A first calculation unit, configured to calculate local coordinates corresponding to each local decomposition data by using a principal component analysis algorithm;

[0118] A second calculation unit, configured to calculate corresponding global coordinates according to the local coordinates of each local decomposition data;

[0119] A third calculation unit, configured to calculate corresponding low-dimensional data according to the global coordinates of each local decomposition data.

[0120] Optionally, the decomposition criterion includes: there is an overlap between data full coverage of the point cloud data to be dimensionally reduced and local decomposition data;

[0121] Decompose the point cloud data to be dimensionally reduced of the lidar according to a preset decomposition criterion to obtain a plurality of local decomposition data, specifically including:

[0122] Obtain the point cloud data to be dimensionally reduced of the lidar;

[0123] Decompose all data in the point cloud data to be dimensionally reduced to obtain a plurality of local decomposition data, wherein one local decomposition data overlaps with at least one local decomposition data.

[0124] Optionally, calculating local coordinates corresponding to each local decomposition data by using a principal component analysis algorithm specifically includes:

[0125] Calculate the centered matrix corresponding to each local decomposition data;

[0126] Perform singular value decomposition on each centered matrix to obtain a corresponding decomposition matrix;

[0127] Calculate corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data.

[0128] Optionally, calculating corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data specifically includes:

[0129] Based on a first preset formula, calculate corresponding local coordinates according to the decomposition matrix corresponding to each local decomposition data, and the first preset formula is:

[0130]

[0131] wherein, Θ m is the local coordinate of the local decomposition data X m is, is the local decomposition data X m is the centered matrix, U m,d is a matrix composed of the first d column vectors of U m is, U m is the local decomposition data Xm Decomposition matrix

[0132] Optionally, calculate the corresponding global coordinates according to the local coordinates of each local decomposition data, specifically including:

[0133] Based on the second preset formula, calculate the corresponding global coordinates according to the local coordinates of each local decomposition data. The second preset formula is:

[0134]

[0135] In the formula, Θ m is the local coordinate of the local decomposition data X m , is the global coordinate of the local decomposition data X m , A m is the rotation and scaling matrix

[0136] Optionally, calculate the corresponding low-dimensional data according to the global coordinates of each local decomposition data, specifically including:

[0137] Obtain the conversion coefficient between the low-dimensional data and the global coordinates;

[0138] Based on the third preset formula, calculate the corresponding low-dimensional data according to the global coordinates of each local decomposition data. The third preset formula is:

[0139]

[0140] In the formula, is the low-dimensional data of the local decomposition data X m , is the global coordinate of the local decomposition data X m , w m,i is the conversion coefficient between the low-dimensional data and the global coordinate .

[0141] Optionally, the expression of the conversion coefficient w m,i is:

[0142]

[0143] In the formula, N m is the number of high-dimensional data included in the local decomposition data X m , is the centering matrix is a vector of all 1s, s m,i is the selection matrix, where the element in the Nth m row and the ith column is 1, and the others are 0, Θ m is the local coordinate of the local decomposition data X m , is Θ m right pseudoinverse of

[0144] In this embodiment, the laser point cloud data dimensionality reduction device first decomposes the point cloud data to be dimensionally reduced of the lidar according to a preset decomposition criterion to obtain a number of local decomposition data, then uses the principal component analysis algorithm to calculate the local coordinates corresponding to each local decomposition data, then calculates the corresponding global coordinates according to the local coordinates of each local decomposition data, and finally calculates the corresponding low-dimensional data according to the global coordinates of each local decomposition data. This application can effectively reduce the dimension of laser point cloud data, solving the technical problem that although the existing dimensionality reduction of laser point cloud data can achieve a certain dimensionality reduction effect, the dimensionality reduction effect is not ideal, resulting in high-dimensional data still existing in the data after dimensionality reduction.

[0145] This application embodiment also provides an embodiment of a laser point cloud data dimensionality reduction device. The device includes a processor and a memory; the memory is used to store program code and transmit the program code to the processor; the processor is used to execute the laser point cloud data dimensionality reduction method of the foregoing embodiment according to the instructions in the program code.

[0146] This application embodiment also provides an embodiment of a storage medium. The storage medium is used to store program code, and the program code is used to execute the laser point cloud data dimensionality reduction method of the foregoing embodiment.

[0147] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0148] Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same and similar parts among the embodiments can be referred to each other.

[0149] Those skilled in the art should understand that the embodiments of this application can be provided as a method, a device, or a computer program product. Therefore, the embodiments of this application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0150] Embodiments of the present application are described with reference to the flowcharts and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing terminal devices generate means for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or in one or more blocks.

[0151] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or in one or more blocks.

[0152] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, such that a series of operation steps are executed on the computer or other programmable terminal device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide steps for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or in one or more blocks.

[0153] Although the preferred embodiments of the embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present application.

[0154] Finally, it should also be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising said element.

[0155] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for dimensionality reduction of lidar point cloud data, characterized in that, it includes: Decompose the lidar point cloud data to be dimensionally reduced according to a preset decomposition criterion to obtain several locally decomposed data, specifically including: obtaining the lidar point cloud data to be dimensionally reduced; decomposing all the data in the point cloud data to be dimensionally reduced to obtain several locally decomposed data, where one locally decomposed data overlaps with at least one other locally decomposed data; the decomposition criterion includes: there is data full coverage of the point cloud data to be dimensionally reduced and overlap between the locally decomposed data; Use the principal component analysis algorithm to calculate the local coordinates corresponding to each of the locally decomposed data; Calculate the corresponding global coordinates according to the local coordinates of each of the locally decomposed data; Calculate the corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data, specifically including: obtaining the conversion coefficient between the low-dimensional data and the global coordinates; based on the third preset formula, calculate the corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data, and the third preset formula is: ； Wherein, is the low-dimensional data of the local decomposition data , is the global coordinate of the local decomposition data , is the conversion coefficient between the low-dimensional data and the global coordinate ; The conversion coefficient has the following expression: ； Wherein, is the number of high-dimensional data contained in the local decomposition data , is the centering matrix , is a vector of all 1s is the element in the th row and th column of the selection matrix is 1 and others are 0 is the local coordinate of the local decomposition data , is 's right pseudo-inverse 2. The method for dimensionality reduction of lidar point cloud data according to claim 1, characterized in that, The step of using the principal component analysis algorithm to calculate the local coordinates corresponding to each of the locally decomposed data specifically includes: Calculate the centralized matrix corresponding to each of the locally decomposed data; Perform singular value decomposition on each of the centralized matrices to obtain the corresponding decomposition matrix; Calculate the corresponding local coordinates according to the decomposition matrix corresponding to each of the locally decomposed data.

3. The method for dimensionality reduction of lidar point cloud data according to claim 2, characterized in that, The step of calculating the corresponding local coordinates according to the decomposition matrix corresponding to each of the locally decomposed data specifically includes: Based on the first preset formula, calculate the corresponding local coordinates according to the decomposition matrix corresponding to each of the locally decomposed data, and the first preset formula is: ； In the formula, is the local decomposition data of the local coordinates, is the local decomposition data of the centralized matrix, is composed of of the first matrix formed by the column vectors, is the local decomposition data of the decomposition matrix.

4. The method for dimensionality reduction of lidar point cloud data according to claim 1, characterized in that, The step of calculating the corresponding global coordinates according to the local coordinates of each of the locally decomposed data specifically includes: Based on the second preset formula, calculate the corresponding global coordinates according to the local coordinates of each of the locally decomposed data, and the second preset formula is: ； In the formula, is the local coordinate of the local decomposition data , is the global coordinate of the local decomposition data , is the rotation and scaling matrix.

5. A device for dimensionality reduction of lidar point cloud data, characterized in that, it includes: A decomposition unit for decomposing the lidar point cloud data to be dimensionally reduced according to a preset decomposition criterion to obtain several locally decomposed data, specifically including: obtaining the lidar point cloud data to be dimensionally reduced; decomposing all the data in the point cloud data to be dimensionally reduced to obtain several locally decomposed data, where one locally decomposed data overlaps with at least one other locally decomposed data; the decomposition criterion includes: there is data full coverage of the point cloud data to be dimensionally reduced and overlap between the locally decomposed data; A first calculation unit for using the principal component analysis algorithm to calculate the local coordinates corresponding to each of the locally decomposed data; A second calculation unit for calculating the corresponding global coordinates according to the local coordinates of each of the locally decomposed data; A third computing unit, configured to calculate corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data, specifically including: obtaining a conversion coefficient between the low-dimensional data and the global coordinates; calculating corresponding low-dimensional data according to the global coordinates of each of the locally decomposed data based on a third preset formula, where the third preset formula is: ； In the formula, is the low-dimensional data of the local decomposition data , is the global coordinate of the local decomposition data , is the low-dimensional data and the global coordinate is the conversion coefficient; The conversion coefficient has the following expression: ； In the formula, is the number of high-dimensional data contained in the local decomposition data , is the centering matrix, , is a vector of all 1s, is the th row and th column element of the selection matrix is 1 and the others are 0, is the local coordinates of the local decomposition data , is 's right pseudo-inverse.

6. A dimensionality reduction device for laser point cloud data, characterized in that the device includes a processor and a memory; the memory is configured to store program code and transmit the program code to the processor; the processor is configured to execute the dimensionality reduction method for laser point cloud data according to any one of claims 1 to 4 based on the instructions in the program code.

7. A storage medium, characterized in that the storage medium is configured to store program code, and the program code is used to execute the dimensionality reduction method for laser point cloud data according to any one of claims 1 to 4.

Images