Point cloud data preprocessing method and point cloud geometric coding and decoding method and device

Abstract

The invention discloses a point cloud data preprocessing method and a point cloud geometric coding and decoding method and device, and the method comprises the steps: carrying out the coordinate transformation of original point cloud data, and obtaining the representation of an original point cloud in a cylindrical coordinate system; unfolding the cylindrical coordinate system to obtain a two-dimensional structure; and performing regularization processing on the two-dimensional structure based on the geometric distortion measure to obtain a regularized structure. The coding method comprises the following steps: carrying out predictive coding on preprocessed original point cloud data to obtain a geometric information code stream; according to the point cloud geometric coding method provided by the invention, regularization preprocessing is performed on the original point cloud, so that the point cloud is in regularization distribution in both the horizontal direction and the vertical direction, the spatial correlation of the point cloud is better reflected, and the coding efficiency is improved.

Description

[0001] The present application is a divisional application of the patent application No. 202110180985.0 entitled "Point cloud data preprocessing method and point cloud geometry encoding and decoding method and device" filed on February 8, 2021. TECHNICAL FIELD

[0002] The present application belongs to the technical field of point cloud data processing, and particularly relates to a point cloud data preprocessing method and a point cloud geometry encoding and decoding method and device. BACKGROUND

[0003] In the point cloud G-PCC (Geometry-based Point Cloud Compression) encoder framework, the geometry information of the point cloud and the attribute information corresponding to each point are encoded separately. The current G-PCC geometry encoding and decoding can be divided into octree-based geometry encoding and prediction tree-based geometry encoding. The prediction tree-based geometry encoding first sorts the input point cloud, and establishes a prediction tree structure at the encoding end by using two different ways. Then, based on the prediction tree structure, each node in the prediction tree is traversed, the geometry position information of the node is predicted by selecting different prediction modes to obtain a prediction residual, and the geometry prediction residual is quantized by using a quantization parameter. Finally, the prediction residual of the prediction tree node position information, the prediction tree structure, and the quantization parameter are encoded by continuous iteration to generate a binary code stream.

[0004] The prediction tree encoding based on the laser radar calibration information is a commonly used geometry encoding method. For each laser scanner of the laser radar, the collected points belonging to the same laser scanner should be regularly distributed in the cylindrical coordinate system. However, due to factors such as noise, measurement error and device jitter, the actual data presents a non-uniform distribution, thereby causing poor correlation between the data and low prediction accuracy and encoding efficiency.

[0005] However, the prediction tree-based point cloud encoding and decoding technology only uses part of the parameters of the laser radar device to establish a tree structure, and the tree structure does not fully reflect the spatial correlation of the point cloud, which is not conducive to the prediction and entropy encoding of the point cloud, thereby affecting the encoding efficiency. The existing G-PCC method only determines the relationship between each point and the laser scanner through vertical direction correction, which causes the need to introduce other variables to assist the encoding of the horizontal direction information during encoding, thereby increasing the amount of information to be encoded and reducing the geometry encoding efficiency. SUMMARY

[0006] In order to solve the above problems existing in the prior art, the application provides a point cloud geometry prediction coding method and device based on a regularized structure.

[0007] A point cloud data preprocessing method comprises the following steps:

[0008] Coordinate conversion is performed on original point cloud data to obtain a representation of the original point cloud in a cylindrical coordinate system;

[0009] The cylindrical coordinate system is unfolded to obtain a two-dimensional structure;

[0010] The two-dimensional structure is regularized and preprocessed based on a geometric distortion measure to obtain a regularized structure.

[0011] In an embodiment of the application, the regularized preprocessing of the two-dimensional structure based on the geometric distortion measure comprises the following steps:

[0012] The two-dimensional structure is adjusted according to a point-to-plane geometric distortion measure to obtain a regularized structure.

[0013] In an embodiment of the application, the adjustment of the two-dimensional structure according to the point-to-plane geometric distortion measure comprises the following steps:

[0014] The nearest point to the current point in the azimuth and pitch angle direction of the two-dimensional structure is found;

[0015] A ray is constructed from the origin point according to the angle information of the nearest point to the current point;

[0016] A plane is constructed according to the current point and its normal line;

[0017] The intersection of the ray and the plane is calculated, and the distance from the origin point to the intersection position is recorded;

[0018] The distance is taken as the radius of the regularized current point from the center;

[0019] The above steps are repeated to complete the processing of all points to obtain the regularized structure of the original point cloud data.

[0020] In an embodiment of the application, the regularized preprocessing of the two-dimensional structure based on the geometric distortion measure further comprises the following steps:

[0021] The two-dimensional structure is adjusted according to a point-to-point geometric distortion measure to obtain a regularized structure; or

[0022] The two-dimensional structure is adjusted according to a comprehensive distortion measure of point-to-point and point-to-plane to obtain a regularized structure; or

[0023] adjusting the two-dimensional structure according to a point-to-line geometric distortion measure to obtain a regularized structure.

[0024] Another embodiment of the present application also provides a point cloud geometry coding method, characterized in that comprising:

[0025] obtaining original point cloud data;

[0026] regularizing and preprocessing the original point cloud data by using the preprocessing method in the above embodiment to obtain a regularized structure;

[0027] determining a prediction mode of each point in the regularized structure, and performing geometry prediction on each point by using the selected prediction mode to obtain to-be-coded information;

[0028] sequentially coding the to-be-coded information to obtain a geometry information code stream.

[0029] In an embodiment of the present application, determining a prediction mode of each point in the regularized structure, and performing geometry prediction on each point by using the selected prediction mode to obtain to-be-coded information comprises:

[0030] establishing a prediction tree structure based on laser radar calibration information;

[0031] selecting a prediction mode of each point according to the prediction tree structure;

[0032] performing geometry prediction on each point in the prediction tree structure according to the selected prediction mode to obtain a geometry prediction residual of each point;

[0033] taking the geometry prediction residual as part of the to-be-coded information.

[0034] In an embodiment of the present application, performing geometry prediction on each point in the prediction tree structure according to the selected prediction mode to obtain a geometry prediction residual of each point comprises:

[0035] performing prediction on the cylindrical coordinates (r, j, i) of the current node according to the current node type and the selected prediction mode to obtain a prediction value (r', j', i') and a prediction residual (r r , j j , i i ) in the cylindrical coordinate system of the current node; wherein the prediction value j' of the azimuth angle of the current point is calculated according to the following formula:

[0036] j'=j prev +n;

[0037] wherein j prevdenotes the predicted azimuth angle of the current point; n denotes the number of points that need to be skipped between the parent node and the current point according to the scanning speed, and the prediction residual thereof is n' denotes the number of points that need to be skipped between the current point and the neighboring node that has been encoded;

[0038] According to the Cartesian coordinates (x, y, z) of the current point and the predicted Cartesian coordinates , the difference prediction is performed to obtain the prediction residual (r x , r y , r z ) in the Cartesian coordinate system.

[0039] Another embodiment of the present application also provides a point cloud geometry encoding device, comprising:

[0040] A first data acquisition module is configured to acquire original point cloud data;

[0041] A regularization module is configured to perform regularization preprocessing on the original point cloud data to obtain a regularized structure;

[0042] A first prediction module is configured to determine a prediction mode of each point in the regularized structure, and perform geometry prediction on each point by using the selected prediction mode to obtain to-be-encoded information;

[0043] An encoding module is configured to sequentially encode the to-be-encoded information to obtain a geometry information code stream.

[0044] Still another embodiment of the present application also provides a point cloud geometry decoding method, comprising:

[0045] Acquiring a geometry information code stream and decoding the geometry information code stream to obtain decoded data; wherein the decoded data comprises a prediction mode of a current node;

[0046] Performing geometry prediction on the current node according to the prediction mode to obtain a prediction residual; wherein the prediction residual comprises a prediction residual in a cylindrical coordinate system and a prediction residual in a Cartesian coordinate system;

[0047] Reconstructing a prediction tree structure according to the prediction residual in the cylindrical coordinate system, and performing coordinate conversion on points in the prediction tree structure to obtain a predicted Cartesian coordinate of the current point;

[0048] Performing point cloud reconstruction according to the prediction residual in the Cartesian coordinate system and the predicted Cartesian coordinate to obtain reconstructed point cloud data.

[0049] Still another embodiment of the present application also provides a point cloud geometry decoding device, comprising:

[0050] The second data acquisition module is configured to acquire a geometry information code stream and decode the geometry information code stream to obtain decoded data, wherein the decoded data comprises a prediction mode of a current node;

[0051] The second prediction module is configured to perform geometry prediction on the current node according to the prediction mode to obtain a prediction residual, wherein the prediction residual comprises a prediction residual in a cylindrical coordinate system and a prediction residual in a Cartesian coordinate system.

[0052] The prediction tree reconstruction module is configured to reconstruct a prediction tree structure according to the prediction residual in the cylindrical coordinate system, and perform coordinate conversion on points in the prediction tree structure to obtain predicted Cartesian coordinates of the current point.

[0053] The point cloud reconstruction module is configured to reconstruct a geometry point cloud according to the geometry prediction value and the prediction residual of the current node to obtain reconstructed point cloud data.

[0054] The present application has the following advantages:

[0055] 1. The point cloud data processing method provided by the present application performs regularized preprocessing on the original input point cloud, so that the point cloud is regularly distributed in the horizontal and vertical directions, better reflecting the spatial correlation of the point cloud, and facilitating further processing of the point cloud data.

[0056] 2. The present application uses a geometry distortion measure to adjust the two-dimensional structure during regularized processing, thereby ensuring the quality of the point cloud model.

[0057] 3. The point cloud geometry coding method provided by the present application performs regularized processing on the azimuth direction of the point cloud during preprocessing, so that no additional code stream is required to code the auxiliary information in the azimuth direction during coding, thereby saving code stream and improving coding efficiency.

[0058] 4. When the present application performs geometry coding on the point cloud after regularized processing, the horizontal and vertical directions are combined for prediction coding by effectively using the regularized structure, thereby improving the geometry coding efficiency.

[0059] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is a point cloud data preprocessing method provided by an embodiment of the present application;

[0061] Figure 2 is a laser radar structure schematic diagram provided by an embodiment of the present application;

[0062] Figure 3 is a schematic diagram of the distribution structure of the original collected data provided by an embodiment of the present application;

[0063] Figure 4 is a cylindrical coordinate system developed drawing provided by an embodiment of the present application;

[0064] Figure 5 is a point-to-plane interpolation processing schematic diagram provided by an embodiment of the present application;

[0065] Figure 6 is a before-and-after comparison diagram of regularization processing provided by an embodiment of the present application;

[0066] Figure 7 is a point cloud geometry encoding method schematic diagram provided by an embodiment of the present application;

[0067] Figure 8 is a point cloud geometry encoding device structure schematic diagram provided by an embodiment of the present application;

[0068] Figure 9 is a point cloud geometry decoding method schematic diagram provided by an embodiment of the present application;

[0069] Figure 10 is a point cloud geometry decoding device structure schematic diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0070] The present application will be further described in detail below with specific embodiments, but the embodiments of the present application are not limited thereto.

[0071] Embodiment one

[0072] Please refer to Figure 1 , Figure 1 is a point cloud data preprocessing method schematic diagram provided by an embodiment of the present application, comprising:

[0073] S1: performing coordinate conversion on the original point cloud data to obtain a representation of the original point cloud under a cylindrical coordinate system.

[0074] Specifically, the original point cloud data is usually composed of a group of three-dimensional space points, each of which records its own geometric position information, as well as additional attribute information such as color, reflectivity, normal, etc. Among them, the geometric position information of the point cloud is generally represented based on the Cartesian coordinate system, that is, represented by the x, y, z coordinates of the point. The original point cloud data can be obtained by 3D scanning devices such as laser radar, etc., or obtained by public data sets provided by various platforms. In this embodiment, the geometric position information of the obtained original point cloud data is represented as (x, y, z) based on the Cartesian coordinate system. It should be noted that the representation method of the geometric position information of the original point cloud data is not limited to the Cartesian coordinate.

[0075] Specifically, before the coordinate conversion of the original point cloud, the original point cloud can also be subjected to line quantization processing and reordering processing to facilitate subsequent prediction coding.

[0076] Please refer to Figure 2 , Figure 2 is a schematic diagram of a laser radar structure provided by an embodiment of the present application; the laser radar is composed of a plurality of laser scanners; when describing a laser radar, the so-called "line number" refers to the number of laser scanners contained therein; the laser scanners are distributed along the center axis of the laser radar on both sides and have different pitch angles theta i , so that the spatial information of different objects in the vertical direction in space can be obtained. Each can be regarded as a relatively independent acquisition system. When acquiring, all the laser scanners acquire the position information of the objects in the space at a certain sampling rate through the rotation of the base.

[0077] In the embodiment, the Cartesian coordinates (x, y, z) of the original point cloud data can be converted into cylindrical coordinates (r, phi, i) according to the existing conversion formula to obtain the representation of the original point cloud in the cylindrical coordinate system.

[0078] Please refer to Figure 3 , Figure 3 is a schematic diagram of the distribution structure of the original acquisition data in the cylindrical coordinate system. In an ideal case, the point cloud data obtained by the laser radar presents uniform distribution along the azimuth angle and the pitch angle in the cylindrical coordinate system. However, due to noise, mechanical jitter, and alignment of the internal coordinate system of the laser radar, the point cloud presents the characteristics of non-uniform distribution.

[0079] S2: unfolding the cylindrical coordinate system to obtain a two-dimensional structure.

[0080] In the embodiment, first, the pitch angle theta and the azimuth angle phi of each point after regularization are determined by the structure and acquisition parameters of the laser radar. Here, the pitch angle theta can be directly obtained from the vertical acquisition range of each laser scanner in the calibration file, and the azimuth angle phi needs to be determined through the sampling interval .

[0081] Specifically, please refer to Figure 4 , Figure 4 is a cylindrical coordinate system unfolding diagram provided by an embodiment of the present application, wherein the interval in the vertical direction and the interval in the horizontal direction can become similar to the concept of resolution in an image. Therefore, the vertical resolution theta and the horizontal resolution phi are respectively:

[0082] theta = laserNum;

[0083]

[0084] S3: regularizing the two-dimensional structure based on the geometric distortion measure to obtain a regularized structure.

[0085] After determining the vertical resolution and the horizontal resolution in step S2, it is also necessary to determine the radius r of each point from the center after regularizing.

[0086] The selection of the radius directly determines whether the point cloud after regularizing can maintain the same geometric structure as the input point cloud. For applications such as recognition and automatic driving, the distortion of the geometric structure directly determines the performance of such applications. Therefore, in order to reduce the geometric structure distortion of the model as much as possible, the embodiment calculates the r component of the corresponding point after regularizing by using the geometric distortion measure idea nearest neighbor interpolation, thereby ensuring that the distortion of the geometric reconstruction quality in D2 (point to plane) can be controlled within a certain range.

[0087] In the embodiment, the two-dimensional structure can be adjusted according to the point-to-plane geometric distortion measure (D2) to obtain a regularized structure.

[0088] See Figure 5 , Figure 5 is a schematic diagram of point-to-plane interpolation processing provided by the embodiment of the application, specifically,

[0089] 1. First, find the point closest to the current point in the azimuth angle and pitch angle direction in the two-dimensional structure;

[0090] 2. Construct a ray from the origin (origin) through the angle information of the point;

[0091] 3. Construct a plane according to the current point p i and its normal;

[0092] 4. Find the intersection point of the ray and the plane, and record the distance from the origin to the position of the intersection point.

[0093] 5. Take the distance information as the radius of the current point from the center after regularizing, that is, the r component in the cylindrical coordinate.

[0094] 6. Repeat the above steps to complete the processing of all points to obtain the regularized structure of the original point cloud data.

[0095] At this point, the regularizing preprocessing of the cylindrical coordinates (r, φ, i) of the point cloud is completed, and the coordinates (r, j, i) after regularizing are obtained. See Figure 6 , Figure 6 is a comparison diagram of regularizing processing provided by the embodiment of the application.

[0096] The regularized structure constructed through the above steps can guarantee that the distortion of each point to the plane is zero, maintains the geometric structure information of the original point cloud, has little influence on the performance of applications such as identification and automatic driving, and is extremely friendly to further processing of subsequent point clouds.

[0097] In another embodiment of the application, the two-dimensional structure can also be adjusted according to the point-to-point geometric distortion measure (D1) to obtain a regularized structure, so as to guarantee the geometric D1 distortion measure and the model quality of the point cloud.

[0098] In addition, the two-dimensional structure can also be adjusted according to the comprehensive distortion measure of point-to-point (D1) and point-to-plane (D2) to obtain a regularized structure; in this way, the overall distortion measure of geometric D1 and D2 and the model quality of the point cloud can be guaranteed.

[0099] Further, the two-dimensional structure can also be adjusted according to the point-to-line geometric distortion measure to obtain a regularized structure. The point-to-line geometric distortion measure is between the point-to-point and point-to-plane distortion measures based on statistics, so the overall distortion of geometric D1 and D2 and the model quality of the point cloud can also be guaranteed.

[0100] The point cloud data processing method provided by the application performs regularized preprocessing on the original input point cloud, so that the point cloud is regularly distributed in the horizontal and vertical directions, increases the correlation between data, facilitates further processing of the point cloud data, and guarantees the point cloud model quality by using the geometric distortion measure for horizontal regularized processing when performing the regularized processing.

[0101] Embodiment Two

[0102] See Figure 7 , Figure 7 is a point cloud geometric coding method provided by an embodiment of the application, including the following steps:

[0103] Step 1: Obtain original point cloud data.

[0104] In this embodiment, the original point cloud data is expressed as (x, y, z) in Cartesian coordinates.

[0105] Step 2: Perform regularized preprocessing on the original point cloud data to obtain a regularized structure.

[0106] Specifically, the original point cloud data can be preprocessed by the preprocessing method provided in Embodiment One to obtain a regularized structure.

[0107] More specifically, the cylindrical coordinates (x, y, z) are converted into a regular (r, j, i) structure after the regularized processing.

[0108] Step 3: Determine the prediction mode of each point in the regularized structure, and perform geometric prediction on each point using the selected prediction mode to obtain the to-be-encoded information.

[0109] First, a prediction tree structure is established based on the laser radar calibration information.

[0110] Then, the prediction mode of the current point is selected according to the prediction tree structure.

[0111] In this embodiment, the established prediction tree is traversed in a depth-first order, and each node in the tree can only be predicted by its ancestor.

[0112] Further, the following four prediction modes are provided in this embodiment:

[0113] Mode0: No prediction

[0114] Mode1: Delta prediction (i.e. p0)

[0115] Mode2: Linear prediction (i.e. 2p0-p1)

[0116] Mode3: Parallelogram prediction (i.e. p0+p1-p2)

[0117] Wherein, p0, p1, p2 are the positions of the parent node, grandparent node, and great-grandparent node of the current node, respectively. The best prediction mode can be selected according to the reconstruction quality to perform prediction.

[0118] Next, each point in the prediction tree structure is geometrically predicted according to the selected prediction mode to obtain the geometric prediction residual of each point. In this embodiment, the geometric prediction includes cylindrical coordinate prediction and Cartesian coordinate prediction, and the specific process is as follows:

[0119] According to the current node type and the selected prediction mode, the cylindrical coordinates (r, j, i) of the current node are predicted to obtain the prediction value (r', j', i') and the prediction residual (r r ,r j ,r i ) in the cylindrical coordinate system of the current node;

[0120] If the current node is the root node of the prediction tree, the prediction mode is selected as Mode0, i.e. no prediction of the cylindrical coordinates of the current node, and the corresponding cylindrical coordinate prediction value is (r min ,j prev ,i prev ), wherein r min is the minimum value of the r component obtained after coordinate conversion of the entire point cloud, and if the current node has no parent node, j prev ,iprev is set to 0, otherwise it is the cylindrical coordinate component of the parent node. The Cartesian coordinate prediction value is obtained by inverse conversion of the cylindrical coordinate prediction value (r', j', i') of the current point

[0121] If the current node is not the root node and the prediction mode is selected as Mode 1, the cylindrical coordinate of the current point is predicted by the cylindrical coordinate (r min ,j prev ,i prev ) of its parent node, and the cylindrical coordinate prediction value (r', j', i') of the current point is obtained, and the Cartesian coordinate prediction value is obtained by inverse conversion of the original cylindrical coordinate (r, j, i)

[0122] If the current node is not the root node and the prediction mode is selected as Mode 2 or Mode 3, the cylindrical coordinate of the current point is predicted by the corresponding prediction mode, and the cylindrical coordinate prediction value (r', j', i') of the current point is obtained, and the Cartesian coordinate prediction value is obtained by inverse conversion of the original cylindrical coordinate (r, j, i)

[0123] The cylindrical coordinate (r, j, i) of the current point is predicted by the best prediction mode of the current node, and the corresponding prediction residual (r r ,r j ,r i ) in the cylindrical coordinate system is obtained.

[0124] It should be noted that the prediction value j' of the azimuth angle of the current point is calculated according to the following formula:

[0125] j' = j prev + n;

[0126] wherein j prev represents the predicted azimuth angle of the current point; n represents the number of points that need to be skipped between the parent node and the current point according to the scanning speed, and if the Laser of the current node is i and the adjacent Laser of the current node is i+1 has been coded, the parameter n can be further predicted by difference prediction using the corresponding position node n' of the Laser i+1, to obtain the prediction residual of the number of points that need to be skipped, i.e.

[0127] In addition, since the geometry prediction coding is based on the regularized structure in the present embodiment, the Laser is coded in sequence according to each Laser, and therefore the node j component of the Laser i that has been coded needs to be temporarily stored for the prediction of the node j component of the corresponding position when the Laser i+1 ~ N is coded.

[0128] with the predicted Cartesian coordinates x y z

[0129] Finally, the predicted residual (r r j i ) in the cylindrical coordinate system, the predicted residual (r x y z ) in the Cartesian coordinate system, and the predicted residual together with the rest of the parameters to be encoded, such as the number of child nodes of the current node, the prediction mode of the current node, and the like, are taken as the information to be encoded.

[0130] Step 4: sequentially encode the information to be encoded to obtain the geometric information code stream.

[0131] Specifically, for the information to be encoded of each node, first, the number of child nodes of the current node is encoded, then the prediction mode of the current node is encoded, and the predicted residuals (r r j i ) and (r x y z ) corresponding to the current node and the predicted residual

[0132] Thus far, the geometric prediction encoding of the point cloud is completed.

[0133] The point cloud geometry encoding method provided in the embodiment has regularized the original input point cloud, so that the point cloud is regularly distributed in the horizontal and vertical directions, the correlation between data is increased, and the encoding efficiency is improved. Meanwhile, due to the regularization in the azimuth direction, auxiliary information in the azimuth direction does not need to be encoded using additional code streams during encoding, code streams are saved, and the encoding efficiency is improved.

[0134] When the point cloud is geometrically encoded after being regularized, the regularized structure is effectively utilized, the horizontal and vertical directions are combined for prediction encoding, and the geometric encoding efficiency is improved.

[0135] In another embodiment of the present application, step 3 can also use an existing geometry encoding mode based on the prediction number to convert the Cartesian coordinates (x, y, z) into cylindrical coordinates and then perform prediction to obtain the predicted value​​​​​​​​​​​​ and and prediction residual and (r x , r y , r z ), and the number of points n corresponding to the skip of the component is encoded, and the specific process is not described in detail.

[0136] Embodiment three

[0137] On the basis of the above-mentioned embodiment two, a mode switch can also be set to guide whether the original point cloud data is initialized in the entire encoding process.

[0138] Specifically, the geometry parameter set syntax (gps) is introduced: geom_enable_regular_flag, which is used to guide whether the geometry of the present application is enabled in the entire encoding. When geom_enable_regular_flag is 1, it means that it is enabled; otherwise, it is closed; see Table 1 for details.

[0139] When the regularized preprocessing scheme is enabled, the point cloud data preprocessing method provided in the above-mentioned embodiment one can be used to preprocess the point cloud data, and then the encoding scheme provided in the embodiment two or the existing geometry prediction encoding method is used to predict and encode the point cloud data, so as to improve the encoding efficiency.

[0140] When the regularized preprocessing scheme is not enabled, the prediction method provided in the above-mentioned embodiment two can also be directly used to predict and encode the original point cloud data.

[0141] Embodiment four

[0142] On the basis of the above-mentioned embodiment two, the present embodiment provides a point cloud geometry encoding device. Please refer to Figure 8 , Figure 8 is a structural schematic diagram of a point cloud geometry encoding device provided by the embodiment of the present application, which comprises:

[0143] The first data acquisition module 11 is used to acquire the original point cloud data.

[0144] The regularized module 12 is used to perform regularized preprocessing on the original point cloud data to obtain a regularized structure.

[0145] The first prediction module 13 is used to determine the prediction mode of each point in the regularized structure, and perform geometry prediction on each point by using the selected prediction mode to obtain to-be-encoded information.

[0146] The encoding module 14 is configured to sequentially encode the information to be encoded to obtain the geometry information code stream.

[0147] The device provided in the embodiment can implement the encoding method provided in the second embodiment, and the specific implementation process is not described here.

[0148] Embodiment five

[0149] See Figure 9 , Figure 9 is a point cloud geometry decoding method provided in the embodiment of the application, comprising:

[0150] Step one: obtaining the geometry information code stream and decoding to obtain decoded data; wherein the decoded data comprises the prediction mode of the current node.

[0151] Step two: performing geometry prediction on the current node according to the prediction mode to obtain a prediction residual; wherein the prediction residual comprises a prediction residual in a cylindrical coordinate system and a prediction residual in a Cartesian coordinate system.

[0152] According to the prediction mode of the current point and the reconstructed cylindrical coordinates of the ancestors of the current point, the cylindrical coordinates of the current point are predicted and the corresponding prediction value is generated, and the specific process is as follows:

[0153] If the current node is the root node of the prediction tree, the prediction mode is selected as Mode0, that is, the cylindrical coordinates of the current node have no prediction, and the corresponding cylindrical coordinate prediction value is (r min ,j prev ,i prev ), wherein r min is the minimum value of the r component obtained after coordinate conversion of the entire point cloud, if the current node has no parent node, then j prev ,i prev is 0, otherwise it is the cylindrical coordinate component of the parent node. The Cartesian coordinate prediction value is

[0154] If the current node is not the root node, and the prediction mode is selected as Mode1, the cylindrical coordinates of the current point are predicted by the cylindrical coordinates (r min ,j prev ,i prev ) of the parent node to obtain the cylindrical coordinate prediction value (r',j',i') of the current point, and the Cartesian coordinate prediction value is

[0155] If the current node is not the root node and the prediction mode is selected as Mode 2 or Mode 3, the cylindrical coordinates of the current point are predicted by the corresponding prediction mode, and the predicted value of the cylindrical coordinates of the current point is (r', j', i'), and similarly, the predicted value of the Cartesian coordinates is obtained by inverse conversion of the original cylindrical coordinates (r, j, i)

[0156] The cylindrical coordinates (r, j, i) of the current point are predicted by using the best prediction mode of the current node, and the corresponding prediction residual (r r ,r j ,r i ) in the cylindrical coordinate system is obtained.

[0157] As at the encoding end, it should be noted that the predicted value j' of the azimuth angle of the current point is calculated according to the following formula:

[0158] j' = j prev + n.

[0159] wherein j prev represents the predicted azimuth angle of the current point, and n represents the number of points that need to be skipped between the parent node and the current point according to the scanning speed. It should be noted that if the laser of the current node is i, and the adjacent laser of the current node is i-1 and has been encoded and decoded, the parameter n is obtained by using the corresponding position node n' of the laser i-1, that is:

[0160] At this point, the prediction residual in the cylindrical coordinate system, the number of points that need to be skipped, and the prediction residual in the Cartesian coordinate system are obtained.

[0161] Step three: reconstruct the prediction tree structure according to the prediction residual in the cylindrical coordinate system, and perform coordinate conversion on the points in the prediction tree structure to obtain the predicted Cartesian coordinates of the current point.

[0162] Specifically, the reconstructed cylindrical coordinates (r, j, i) of the current point are calculated by using the decoded cylindrical coordinate residual (r r ,r j ,r i ) and the predicted cylindrical coordinates (r', j', i') of the current point.

[0163] (r, j, i) = (r', j', i') + (r r ,r j ,r i ).

[0164] After obtaining the reconstructed cylindrical coordinates of the current point, the position of the current point in the prediction tree can be further determined according to the reconstructed cylindrical coordinates (r, j, i), so as to reconstruct the prediction tree.

[0165] The reconstructed cylindrical coordinates (r, j, i) of the current point are converted into Cartesian coordinates according to the following formula That is, the predicted Cartesian coordinates of the current point.

[0166]

[0167]

[0168]

[0169]

[0170] Wherein, i is the LaserID corresponding to the point, the prior information of each Laser is different, that is, the elevation angle θ and the height zLaser in the vertical direction are different, so the elevation angle corresponding to the i-th Laser is θ(i), and the height in the vertical direction is zLaser(i).

[0171] Step four: according to the predicted residual error in the Cartesian coordinate system and the predicted Cartesian coordinates, the point cloud is reconstructed to obtain the reconstructed point cloud data.

[0172] According to the following formula, the Cartesian coordinate residual error (r x , y , z ) obtained by decoding and the predicted Cartesian coordinates of the current point The reconstructed Cartesian coordinates (x, y, z) of the current point are calculated.

[0173]

[0174] At this point, the decoding of the point cloud is completed, and the reconstructed voxelized point cloud is obtained.

[0175] Embodiment six

[0176] On the basis of the above-mentioned embodiment five, the embodiment provides a point cloud geometry decoding device. Please see Figure 10 , Figure 10 is a structure diagram of a point cloud geometry decoding device provided by an embodiment of the present application, comprising:

[0177] The second data acquisition module 21 is configured to acquire the geometry information code stream and decode to obtain decoded data; wherein the decoded data comprises a prediction mode of the current node;

[0178] The second prediction module 22 is configured to perform geometry prediction on the current node according to the prediction mode to obtain a prediction residual error; wherein the prediction residual error comprises a prediction residual error in a cylindrical coordinate system and a prediction residual error in a Cartesian coordinate system;

[0179] The prediction tree reconstruction module 23 is configured to reconstruct the prediction tree structure according to the prediction residual in the cylindrical coordinate system, and perform coordinate conversion on the points in the prediction tree structure to obtain the predicted Cartesian coordinates of the current point.

[0180] The point cloud reconstruction module 24 is configured to reconstruct the geometric point cloud according to the geometric prediction value of the current node and the prediction residual to obtain the reconstructed point cloud data.

[0181] The apparatus provided in the embodiment can implement the decoding method provided in the fourth embodiment, and the specific implementation process will not be described here.

[0182] The above is a further detailed description of the present application in combination with specific preferred embodiments, and the specific implementation of the present application should not be limited to these descriptions. For ordinary skilled persons in the technical field of the present application, several simple deductions or substitutions can be made without departing from the concept of the present application, and all of them should be considered as falling within the protection scope of the present application.

[0183] Table 1: Geometry parameter set syntax

[0184]

[0185]

Claims

1. A method of point cloud geometry decoding, the method comprising: The method comprises the following steps: obtaining a geometry information code stream and decoding the geometry information code stream to obtain decoded data, wherein the decoded data comprises a prediction mode of a current node; performing geometry prediction on the current node according to the prediction mode to obtain a prediction value and a prediction residual, wherein the prediction value comprises a cylindrical coordinate prediction value, and the prediction residual comprises a prediction residual in a cylindrical coordinate system and a prediction residual in a Cartesian coordinate system; obtaining reconstructed geometry coordinates in the cylindrical coordinate system according to the cylindrical coordinate prediction value and the prediction residual in the cylindrical coordinate system; obtaining reconstructed geometry coordinates in the Cartesian coordinate system according to the reconstructed geometry coordinates in the cylindrical coordinate system and the prediction residual in the Cartesian coordinate system; wherein the step of obtaining the reconstructed geometry coordinates in the Cartesian coordinate system according to the reconstructed geometry coordinates in the cylindrical coordinate system and the prediction residual in the Cartesian coordinate system comprises the following steps: obtaining reconstructed azimuth angle information according to a j component of the reconstructed geometry coordinates in the cylindrical coordinate system, and then obtaining a predicted Cartesian coordinate according to the reconstructed azimuth angle information, distance information and an i component of the reconstructed geometry coordinates in the cylindrical coordinate system; performing point cloud reconstruction according to the prediction residual in the Cartesian coordinate system and the predicted Cartesian coordinate to obtain reconstructed point cloud data.

2. The method of Claim 1, wherein, The distance information is a radius of the current node from a center after normalization.

3. An apparatus for point cloud geometry decoding, the apparatus comprising: The method comprises the following steps: a second data obtaining module (21) is configured to obtain a geometry information code stream and decode the geometry information code stream to obtain decoded data, wherein the decoded data comprises a prediction mode of a current node; a second prediction module (22) is configured to perform geometry prediction on the current node according to the prediction mode to obtain a prediction value and a prediction residual, wherein the prediction value comprises a cylindrical coordinate prediction value, and the prediction residual comprises a prediction residual in a cylindrical coordinate system and a prediction residual in a Cartesian coordinate system; a prediction tree reconstruction module (23) is configured to obtain reconstructed geometry coordinates in the cylindrical coordinate system according to the cylindrical coordinate prediction value and the prediction residual in the cylindrical coordinate system; obtaining reconstructed geometry coordinates in the Cartesian coordinate system according to the reconstructed geometry coordinates in the cylindrical coordinate system and the prediction residual in the Cartesian coordinate system; wherein the step of obtaining the reconstructed geometry coordinates in the Cartesian coordinate system according to the reconstructed geometry coordinates in the cylindrical coordinate system and the prediction residual in the Cartesian coordinate system comprises the following steps: obtaining reconstructed azimuth angle information according to a j component of the reconstructed geometry coordinates in the cylindrical coordinate system, and then obtaining a predicted Cartesian coordinate according to the reconstructed azimuth angle information, distance information and an i component of the reconstructed geometry coordinates in the cylindrical coordinate system; a point cloud reconstruction module (24) is configured to perform point cloud reconstruction according to the prediction residual in the Cartesian coordinate system and the predicted Cartesian coordinate to obtain reconstructed point cloud data.

4. The apparatus of claim 3, wherein, The distance information is a radius of the current node from a center after normalization.

Images