WO2019210531A1 - 一种基于删除量化矩阵中0元素的点云属性压缩方法 - Google Patents

一种基于删除量化矩阵中0元素的点云属性压缩方法 Download PDF

Info

Publication number
WO2019210531A1
WO2019210531A1 PCT/CN2018/086793 CN2018086793W WO2019210531A1 WO 2019210531 A1 WO2019210531 A1 WO 2019210531A1 CN 2018086793 W CN2018086793 W CN 2018086793W WO 2019210531 A1 WO2019210531 A1 WO 2019210531A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
data stream
quantization matrix
elements
information
Prior art date
Application number
PCT/CN2018/086793
Other languages
English (en)
French (fr)
Inventor
李革
张琦
邵薏婷
高文
Original Assignee
北京大学深圳研究生院
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 北京大学深圳研究生院 filed Critical 北京大学深圳研究生院
Priority to US17/045,894 priority Critical patent/US11216985B2/en
Publication of WO2019210531A1 publication Critical patent/WO2019210531A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/597—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding specially adapted for multi-view video sequence encoding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
    • 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
    • G06T3/4023—Scaling of whole images or parts thereof, e.g. expanding or contracting based on decimating pixels or lines of pixels; based on inserting pixels or lines of pixels
    • 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
    • G06T3/4084—Scaling of whole images or parts thereof, e.g. expanding or contracting in the transform domain, e.g. fast Fourier transform [FFT] domain scaling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00—Image coding
    • G06T9/001—Model-based coding, e.g. wire frame
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00—Image coding
    • G06T9/40—Tree coding, e.g. quadtree, octree
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/124—Quantisation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/96—Tree coding, e.g. quad-tree coding

Definitions

  • the invention belongs to the technical field of point cloud data processing, and relates to a point cloud data compression method, in particular to a point cloud attribute compression method based on deleting 0 elements in a quantization matrix.
  • the 3D point cloud is an important manifestation of digital realism in the real world. With the rapid development of 3D scanning devices (laser, radar, etc.), the accuracy and resolution of point clouds are higher. High-precision point clouds are widely used in the construction of digital maps in cities, and they play a supporting role in many popular studies such as smart cities, driverless, and cultural relics protection.
  • the point cloud is obtained by sampling the surface of the object by the 3D scanning device.
  • the number of points of a frame point cloud is generally one million, and each point contains geometric information and attribute information such as color and texture, and the amount of data is very large. The huge amount of data in the 3D point cloud brings great challenges to data storage and transmission, so point cloud compression is necessary.
  • Point cloud compression is mainly divided into geometric compression and attribute compression.
  • the existing point cloud attribute compression framework mainly includes the following:
  • This method first uses the octree to spatially decompose the point cloud to obtain the coded block, then performs depth-first traversal on the tree, and writes the traversed node color values in a serpentine order.
  • Into the two-dimensional JPEG table and then use the existing JPEG encoder to encode the obtained point cloud color table, wherein the JPEG encoder uses DCT.
  • the method utilizes the existing encoder, and the computational complexity is low, but the spatial correlation between points is not fully utilized, and the compression performance needs to be improved;
  • the method first uses the octree to spatially decompose the point cloud, and divides it into a specific level to obtain a transform block; forming a graph in each transform block, along any axis The two points whose distance is less than 1 are connected by one edge, and the weight of the edge is inversely proportional to the Euclidean distance; then the attribute information of the node in the figure is transformed.
  • the compression performance of the method is good; but the computational complexity is high, and the patterning method may bring subgraph problems, affecting the efficiency of graph transformation, and there is still room for improvement;
  • the method first uses the KD tree to spatially decompose the point cloud, and divides it to a certain depth to obtain a transform block, wherein each transform block contains approximately the same number of points: within the transform block
  • Each two points are connected by one edge, and the weight of the edge is related to the Euclidean distance.
  • the set edge distance Euclidean distance threshold determines the sparseness of the graph; then the graph transformation of the attribute information of the point in the graph is performed.
  • the method completely solves the subgraph problem, and at the same time, the compression performance is improved compared with the former two methods, but the operation complexity is high and the performance still needs to be improved.
  • the present invention uses seven traversal sequences for the quantization matrix at the encoding end of the point cloud attribute compression, so that the distribution of the 0 elements in the data stream is more concentrated at the end; Delete the 0 element at the end of the data stream, reduce the amount of data that needs to be entropy encoded; combine the point cloud geometry information with the deleted 0 element at the decoding end, restore the quantization matrix according to the traversal order, and improve without introducing new errors. Compression performance.
  • a point cloud attribute compression method based on deleting 0 elements in a quantization matrix, for the quantization matrix in the point cloud attribute compression process, the optimal traversal order is used at the encoding end to make the 0 elements concentrated in the generated data stream at the end, After these 0s are deleted, the entropy coding is performed, the data volume of the data stream is reduced, and the code stream generated after the coding is reduced; the deleted 0 elements are restored at the decoding end by combining the point cloud geometric information, so as to ensure that the method does not introduce additional errors. Including the following steps:
  • the point cloud data to be compressed is first divided into KD trees according to the geometric information.
  • the block obtained by the last layer of the KD tree is the coding block of the point cloud, and the attribute information in each coding block is subjected to intra prediction and residual. After the transformation and quantization, a quantization matrix is obtained.
  • each quantization matrix 7 different traversal sequences are used to convert the two-dimensional matrix into a one-dimensional data stream, and the number of consecutive occurrences of the 0 element at the end of the data stream is compared.
  • the data stream is the optimal data stream, and the corresponding traversal mode is recorded.
  • the optimal data stream After the optimal data stream is obtained, all consecutive zeros at the end of the data stream are deleted, and the clipped data stream is obtained. After performing the same operation on all the coded blocks, the entropy coding is uniformly performed to obtain the code stream compressed by the point cloud attribute.
  • the decoding side restores the quantization matrix with reference geometry information:
  • Entropy decoding the code stream to obtain the clipped data stream combining the geometric information of the point cloud, finding the number of deleted 0 elements and filling up the original data stream, and traversing the one-dimensional data stream according to the traversal order Restore to a two-dimensional quantization matrix.
  • the restored quantization matrix is inversely quantized, inverse transformed, and predicted and compensated in order, and the attribute information of the point cloud is decoded.
  • midpoint KD tree division method is a binary division method; the point cloud to be processed has N points, and the KD tree sets the division depth to d, and after dividing the point cloud d times, 2 d are obtained.
  • Code block the number of points in each block is close, there are n or n+1, and the calculation method of n is as shown in Equation 1. Number all coded blocks in order of breadth-first traversal This number will be used as the order in which the code blocks are post processed.
  • the quantization matrix size obtained in the above step 1) is related to the number of points in the coding block, that is, n ⁇ 3 or (n + 1) ⁇ 3.
  • step 2) for each quantization matrix, 7 different traversal sequences are used, namely: YUV progressive scan, YUV column-by-column scan, YVU column-by-column scan, UYV column-by-column scan, UVY column-by-column scan, VYU column by column Scan, VUY column by column scan.
  • the n ⁇ 3 two-dimensional matrix is transformed into a one-dimensional data stream of length 3n, and the data stream with the largest number of consecutive occurrences of the 0 element is selected as the optimal data stream, and the corresponding traversal mode m i is recorded.
  • the optimal data stream length selected in the above step 3) is 3n. Assuming that the number of 0 elements continuously appearing at the end of the data stream is l i , the length of the clipped data stream is 3n-l i .
  • the entropy decoding obtains the clipped data stream, and the length is l c , and it is necessary to know the number of deleted 0 elements to restore the original data stream.
  • the geometric information of the point cloud is divided into the same KD tree as the encoding end, and 2 d coding blocks are obtained, and there are n (or n+1) points in each block, and then according to the breadth of all the coding blocks.
  • the priority traversal order is numbered, and the obtained result is in one-to-one correspondence with the encoding end.
  • Equation 2 the number of deleted 0 elements l 0 can be obtained.
  • the data stream length after the completion of the 0 element is 3n, and the data stream is converted into an n ⁇ 3 quantization matrix according to the saved traversal pattern m i .
  • step 5 The specific details in step 5) above are as follows:
  • step 4 After step 4), a quantization matrix is obtained, and the quantization matrix is inverse quantized, inverse transformed, predicted and compensated, and the attribute information of the point cloud is decoded;
  • the code stream based on the point cloud attribute compression method for deleting the 0 element in the quantization matrix is mainly composed of two parts of the compression header information and the coding block information.
  • the header information mainly includes quantization step size, prediction mode information, traversal mode information of the quantization matrix, and the like;
  • the coding block information is arranged in units of coding blocks in the order of coding blocks, and each block mainly includes color residual information of the coding block.
  • the performance of point cloud attribute compression is measured by the code rate and Peak Signal to Noise Ratio (PSNR), where the unit of code rate is bpp (bits per point), and the unit of PSNR is decibel dB; The smaller the code rate, the larger the PSNR and the better the compression performance of the point cloud attribute.
  • PSNR Peak Signal to Noise Ratio
  • the invention provides a point cloud attribute compression method based on deleting 0 elements in a quantization matrix, which has the following technical advantages:
  • the optimal traversal order is used at the encoding end so that the 0 elements are concentrated in the generated data stream at the end, and these 0s are deleted and then entropy encoded to reduce the data stream.
  • the amount of data reduces the code stream generated after encoding.
  • 1a is a flow chart of the encoding end of the method provided by the present invention.
  • 1b is a block diagram showing the decoding end of the method provided by the present invention.
  • FIG. 2 is a diagram showing an example of seven traversal sequences adopted by the encoding end for the quantization matrix
  • 3 is a diagram showing an example of a code stream structure after compression of point cloud attribute information
  • Figure 4a is a comparison of compression performance of the method of the present invention and the conventional method of testing Longdress-vox10-1300.ply;
  • Figure 4b is a comparison of the compression performance of the method of the present invention with the existing conventional method of testing Queen-frame-0200.ply;
  • Figure 4c is a graph comparing the compression performance of the method of the present invention with the existing conventional method of Soldier-vox 10-0690.ply.
  • a point cloud attribute compression method based on deleting 0 elements in a quantization matrix according to the present invention for the quantization matrix in the point cloud attribute compression process, adopting an optimal traversal order at the encoding end to make the 0 elements are concentrated in the generated data stream. At the end, these 0s are deleted and entropy coded, the data volume of the data stream is reduced, the code stream generated after encoding is reduced, and the deleted 0 element is restored at the decoding end by combining the point cloud geometry information, so as to ensure that the method does not introduce additional error.
  • Figure 1a is a block flow diagram of the encoding side of the method of the present invention.
  • the first step is to input the geometric information and attribute information of the point cloud data to be compressed at the encoding end;
  • the second step is to perform KD tree division according to the geometric information of the point cloud, and obtain the coding blocks whose numbers are close to each other, and number them sequentially; Steps corresponding to the content of the invention (1)
  • the attribute information is subjected to intra prediction, transform, and quantization 1.
  • the attribute information in each coding block is subjected to intra prediction, transform, and quantization to obtain a corresponding quantization matrix; the fourth step corresponds to the invention content.
  • Step (2) selects 0 element distribution optimal data stream 2 by using multiple traversal sequences, converts the quantization matrix into a data stream by using multiple traversal sequences, and selects the optimal data stream with the most concentrated distribution of 0 elements at the end;
  • Step (3) delete the 0 element 3 at the end of the data, delete the 0 element at the end of the data stream obtained in the previous step; the sixth step uniformly entropy encodes the data stream of all the coding blocks; finally, the point cloud attribute information is obtained.
  • Figure 1b is a block flow diagram of the decoding side of the method of the present invention.
  • the first step is to input the code stream of the point cloud attribute information to be processed at the decoding end;
  • the second step is to entropy encode the code stream to obtain the data stream of all the coding blocks;
  • the third step is to complete the step (4) corresponding to the invention content.
  • the 0 element and restore the quantization matrix 4 refer to the geometric information, fill the 0 element at the end of the data stream in the order of the coding block and restore it to the quantization matrix; the fourth step corresponds to the steps of the invention (5) inverse quantization, inverse transformation, The prediction compensation 5 performs inverse quantization, inverse transformation, and prediction compensation on the quantized mean, and obtains attribute information corresponding to each block; finally, the decoded point cloud attribute information is output.
  • FIG. 2 is a diagram showing an example of seven traversal sequences used by the encoding end for the quantization matrix, and (a) to (g) respectively show YUV progressive scanning, YUV column-by-column scanning, YVU column-by-column scanning, and the n ⁇ 3 quantization matrix, UYV column by column scanning, UVY column by column scanning, VYU column by column scanning, VUY column by column scanning, each point corresponds to each element in the quantization matrix, where the white point is the starting point of the traversal.
  • FIG. 3 is a diagram showing an example of a code stream structure after compression of point cloud attribute information.
  • the code stream is mainly composed of header information and coded block information of each code block.
  • the header information mainly includes information such as a quantization step size, a prediction mode, and a traversal mode of the quantization matrix.
  • the coded block information is arranged in units of coding blocks according to the traversal order of the coding blocks, and each code block information is a color residual of the coding block. .
  • Figures 4a, b and c are graphs comparing the compression performance of the inventive method of Longdress-vox10-1300.ply, Queen-frame-0200.ply and Soldier-vox10-0690.ply with existing conventional methods.
  • the horizontal axis is the code rate
  • the unit is bpp (bits per point)
  • the vertical axis is the peak signal to noise ratio PSNR (Peak Signal to Noise Ratio) of the luminance Y.
  • PSNR Peak Signal to Noise Ratio
  • the unit is decibel db.
  • Point cloud Longdress-vox10-1300.ply has a total of 857966 points, the KD tree partition depth is set to 13, after dividing, there are a total of 8192 code blocks, the number of points in the block is 104 or 105, taking the first code block as an example. There are 104 points, and the attribute information in the block is subjected to intra prediction, residual transformation, and quantization (quantization step size is 4) to obtain a 104 ⁇ 3 quantization matrix Q 1 .
  • the data streams are respectively written into the respective code block information. Then, the information such as the quantization step size, the traversal mode and the prediction mode is written into the compression header information, and the entropy coding is unified, and the structure of the output final stream file is as shown in FIG. 3.
  • the decoding end inputs the code stream file to perform entropy decoding, and obtains the header information and the coding block information of 8192 blocks.
  • the length of the clipped data stream is 46, and the number of deleted 0 elements needs to be known.
  • the geometric information of the point cloud is divided into the same KD tree as the encoding end, and 8192 coding blocks are obtained.
  • 266 elements of 0 elements can be obtained.
  • the restored quantization matrix is combined with quantization step size, prediction mode and other information, and inverse quantization, inverse transformation, and prediction compensation are sequentially performed to obtain attribute information of the point cloud.
  • the performance of point cloud attribute compression is measured by the code rate and Peak Signal to Noise Ratio (PSNR), where the unit of code rate is bpp (bits per point) and the unit of PSNR is decibel dB.
  • PSNR Peak Signal to Noise Ratio
  • the method of deleting the 0 element in the quantization matrix of the present invention is introduced, and then tested.
  • Three types of typical point cloud data yield stable and significant performance gains.
  • the method adopts an optimal traversal order for the quantization matrix at the encoding end of the point cloud attribute compression, so that the 0 elements are concentratedly distributed at the end of the data stream and deleted, thereby reducing redundant information; and the point cloud geometric information is deleted at the decoding end and deleted.
  • the 0 element restores the quantization matrix in traversal order, ensuring that no new errors are introduced.
  • the experimental results show that the performance of the point cloud attribute compression is improved at each code rate point, and the gain generated by the present invention is stable, and the advantages are outstanding.
  • the present invention provides a point cloud attribute compression method based on deleting 0 elements in a quantization matrix.
  • the optimal traversal order is used at the encoding end to make the 0 elements are concentrated in the generated data stream.
  • these 0s are deleted and entropy coded, the data volume of the data stream is reduced, the code stream generated after encoding is reduced, and the deleted 0 element is restored at the decoding end by combining the point cloud geometry information, so as to ensure that the method does not introduce additional Error; with the rapid development of 3D scanning equipment (laser, radar, etc.), the accuracy and resolution of point clouds are higher.
  • High-precision point clouds are widely used in the construction of digital maps in cities, and they play a supporting role in many popular studies such as smart cities, driverless, and cultural relics protection.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

本发明公布了一种基于删除量化矩阵中0元素的点云属性压缩方法，针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，将这些0删除后进行熵编码，降低数据流的数据量，减小编码后生成的码流，在解码端结合点云几何信息复原被删除的0元素，保证本方法不引入额外的误差；包括：编码端优化对量化矩阵的遍历顺序；删除数据流末端的0元素；解码端参考几何信息复原量化矩阵；点云属性压缩编码过程和解码过程。本发明在点云属性压缩的编码端对量化矩阵采用7种遍历顺序，使0元素在数据流中的分布更集中在末端；删除数据流末端的0元素，去除冗余信息，减小需要进行熵编码的数据量；在解码端结合点云几何信息补齐被删除的0元素，按照遍历顺序复原量化矩阵，在不引入新误差的前提下提高压缩性能。

Description

一种基于删除量化矩阵中0元素的点云属性压缩方法 技术领域
本发明属于点云数据处理技术领域，涉及点云数据压缩方法，尤其涉及一种基于删除量化矩阵中0元素的点云属性压缩方法。
背景技术
三维点云是现实世界数字化的重要表现形式。随着三维扫描设备(激光、雷达等)的快速发展，点云的精度、分辨率更高。高精度点云广泛应用于城市数字化地图的构建，在如智慧城市、无人驾驶、文物保护等众多热门研究中起技术支撑作用。点云是三维扫描设备对物体表面采样所获取的，一帧点云的点数一般是百万级别，其中每个点包含几何信息和颜色、纹理等属性信息，数据量十分庞大。三维点云庞大的数据量给数据存储、传输等带来巨大挑战，所以点云压缩十分必要。
点云压缩主要分为几何压缩和属性压缩，现有的点云属性压缩框架主要包括以下几种：
一、基于八叉树分解和DCT的方法：该方法先用八叉树对点云进行空间分解得到编码块，然后对树进行深度优先遍历，将遍历的结点颜色值按照蛇形的顺序写入二维的JPEG表，再利用现有的JPEG编码器对获得的点云颜色表进行编码，其中JPEG编码器使用的是DCT。该方法利用现有的编码器，计算复杂度低，但并未充分利用点与点之间的空间相关性，在压缩性能上有待提高；
二、基于八叉树分解和图变换的方法：该方法先用八叉树对点云进行空间分解，划分到特定层次得到变换块；在每个变换块内形成图，将沿着任意坐标轴的距离不超过1的两点用一条边连接起来，边的权重与欧氏距离成反比；再对图中结点的属性信息进行图变换。该方法的压缩性能佳；但运算复杂度较高，其构图方式可能会带来子图问题，影响图变换的效率，仍有改善的空间；
三、基于KD树分解和图变换的方法：该方法先用KD树对点云进行空间分解， 划分到一定深度得到变换块，其中每个变换块内所包含的点数大致相同：在变换块内，每两点用一条边连接起来，边的权重与欧氏距离有关，设定的边欧式距离阈值决定图的稀疏度；然后再对图内点的属性信息进行图变换。该方法彻底解决了子图问题，同时在压缩性能方面较前两种方法都有较大的提升，但运算复杂度较高，性能仍待改善。
发明内容
为了进一步提升上述技术的性能，在考虑计算复杂度的条件下，本发明在点云属性压缩的编码端对量化矩阵采用7种遍历顺序，使0元素在数据流中的分布更集中在末端；删除数据流末端的0元素，减小需要进行熵编码的数据量；在解码端结合点云几何信息补齐被删除的0元素，按照遍历顺序复原量化矩阵，在不引入新误差的前提下提高压缩性能。
本发明提供的技术方案是：
一种基于删除量化矩阵中0元素的点云属性压缩方法，针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，并将这些0删除后进行熵编码，降低数据流的数据量，减小编码后生成的码流；在解码端结合点云几何信息复原被删除的0元素，保证本方法不引入额外的误差。包括如下步骤：
1)点云属性压缩的编码过程
待压缩的点云数据首先会根据几何信息进行KD树划分，KD树划分的最后一层所得到的块即为点云的编码块，每一个编码块内的属性信息经过帧内预测、残差变换、量化之后会得到一个量化矩阵。
2)编码端优化对量化矩阵的遍历顺序：
对于每一个量化矩阵，采用7种不同的遍历顺序，将二维的矩阵转化为一维的数据流，比较0元素在数据流末端连续出现的个数，选择0元素在末端连续出现个数最多的数据流为最优数据流，同时记录对应的遍历模式。
3)删除数据流末尾的0元素：
得到了最优数据流之后，将数据流末端所有连续出现的0都删除，得到裁剪后的数据流，对所有编码块进行相同操作之后，统一进行熵编码，得到点云属性压缩的码流。
4)解码端参考几何信息复原量化矩阵：
在解码对码流进行熵解码得到裁剪后的数据流，结合点云的几何信息，求出被删除的0元素的个数并补齐得到原始的数据流，依照遍历顺序将一维的数据流复原为二维的量化矩阵。
5)点云属性压缩的解码过程：
按顺序对复原出的量化矩阵进行反量化、逆变换、预测补偿，解码出点云的属性信息。
上述步骤1)中点KD树划分方法为二元划分方法；设待处理的点云共有N个点，KD树设定的划分深度为d，经过对点云d次划分后，得到2 d个编码块，每个块内点的个数接近，有n个或n+1个，n的计算方法如式1。对所有的编码块按照广度优先遍历的顺序进行编号
Figure PCTCN2018086793-appb-000001
该编号将作为编码块后期处理的顺序。
Figure PCTCN2018086793-appb-000002
上述步骤1)中得到的量化矩阵大小与编码块中点的个数有关，即n×3或(n+1)×3。
上述步骤2)中对于每一个量化矩阵，采用7种不同的遍历顺序，分别为：YUV逐行扫描、YUV逐列扫描、YVU逐列扫描、UYV逐列扫描、UVY逐列扫描、VYU逐列扫描、VUY逐列扫描。扫描后n×3的二维矩阵被转化为长度为3n的一维数据流，选择0元素在末端连续出现个数最多的数据流为最优数据流，同时记录对应的遍历模式m i。
上述步骤3)中选出的最优数据流长度为3n，假设数据流末端连续出现的0元素个数为l i，则裁剪后的数据流长度为3n-l i。
上述步骤4)中熵解码得到裁剪后的数据流，长度为l c，需要知道被删除的0元素的个数来恢复原始的数据流。在解码端对点云的几何信息进行与编码端相同的KD树划分，得到2 d个编码块，每个块内有n个(或n+1个)点，再对所有的编码块按照广度优先的遍历顺序进行编号，得到的结果与编码端一一对应，则根据式2可以求出被删除的0元素的个数l 0。补齐0元素之后的数据流长度为3n，根据保存的遍历模式m i将数据流转化为n×3的量化矩阵。
l 0＝3n-l c                (式2)
上述步骤5)中具体细节如下：
(5-1)经过步骤4)得到了量化矩阵，对量化矩阵进行反量化、逆变换、预测补偿，解码出点云的属性信息；
(5-2)基于删除量化矩阵中0元素的点云属性压缩方法的码流主要由压缩头信息和编码块信息两大部分组成。头信息主要包括量化步长、预测模式信息、量化矩阵的遍历模式信息等；编码块信息以编码块为单位，按照编码块的顺序排列，每个块内主要包括编码块的颜色残差信息。
(5-3)点云属性压缩的性能由码率和峰值信噪比PSNR(Peak Signal to Noise Ratio)来衡量，其中码率的单位是bpp(bits per point)，PSNR的单位是分贝dB；码率越小，PSNR越大，点云属性压缩性能越好。
与现有技术相比，本发明的有益效果是：
本发明提供一种基于删除量化矩阵中0元素的点云属性压缩方法，具有以下技术优势：
(一)针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，将这些0删除后进行熵编码，降低数据流的数据量，减小编码后生成的码流。
(二)在解码端结合点云几何信息补齐被删除的0元素，按照遍历顺序复原量化矩阵，在不引入新误差的前提下提高压缩性能。
附图说明
图1a是本发明提供方法的编码端的流程框图；
图1b是本发明提供方法的解码端的流程框图；
图2是编码端对量化矩阵采用的7种遍历顺序的示例图；
图3是点云属性信息压缩后的码流结构示例图；
图4a是测试Longdress-vox10-1300.ply本发明方法与现有传统方法的压缩性能对比图；
图4b是测试Queen-frame-0200.ply本发明方法与现有传统方法的压缩性能对比图；
图4c是测试Soldier-vox10-0690.ply本发明方法与现有传统方法的压缩性能对比图。
实施本发明的最佳方式
下面结合附图，通过实施例进一步描述本发明，但不以任何方式限制本发明的范围。
本发明的一种基于删除量化矩阵中0元素的点云属性压缩方法，针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，将这些0删除后进行熵编码，降低数据流的数据量，减小编码后生成的码流，在解码端结合点云几何信息复原被删除的0元素，保证本方法不引入额外的误差。
图1a是本发明方法的编码端的流程框图,。第一步在编码端输入待压缩点云数据的几何信息和属性信息；第二步根据点云的几何信息进行KD树划分，得到点的个数接近的编码块，并按顺序编号；第三步对应发明内容的步骤(1)属性信息经过帧内预测、变换、量化1，每个编码块内的属性信息经过帧内预测、变换、量化，得到对应的量化矩阵；第四步对应发明内容的步骤(2)采用多种遍 历顺序选择0元素分布最优数据流2，采用多种遍历顺序将量化矩阵转化为数据流，选择0元素在末端分布最集中的最优数据流；第五步对应发明内容的步骤(3)删除数据末端的0元素3，删除上一步得到的数据流末端的0元素；第六步对所有编码块的数据流统一进行熵编码；最后得到点云属性信息的码流。
图1b是本发明方法的解码端的流程框图。第一步在解码端输入需要待处理的点云属性信息的码流；第二步对码流进行熵编码，得到所有编码块的数据流；第三步对应发明内容的步骤(4)补齐0元素并复原量化矩阵4，参照几何信息，按编码块的顺序补齐数据流末端的0元素并将其复原为量化矩阵；第四步对应发明内容的步骤(5)反量化、逆变换、预测补偿5，对量化均值进行反量化、逆变换、预测补偿，得到每个块对应的属性信息；最后输出解码的点云属性信息。
图2是编码端对量化矩阵采用的7种遍历顺序的示例图，(a)到(g)分别表示对n×3的量化矩阵进行YUV逐行扫描、YUV逐列扫描、YVU逐列扫描、UYV逐列扫描、UVY逐列扫描、VYU逐列扫描、VUY逐列扫描，每个点对应量化矩阵中的每个元素，其中白色的点为遍历的起始点。
图3是点云属性信息压缩后的码流结构示例图。码流主要由头信息和各个编码块的编码块信息组成。头信息主要包括量化步长、预测模式、量化矩阵的遍历模式等信息；编码块信息以编码块为单位，按照编码块的遍历顺序排列，每个编码块信息内是该编码块的颜色残差。
图4a、b和c是测试Longdress-vox10-1300.ply、Queen-frame-0200.ply和Soldier-vox10-0690.ply的本发明方法与现有传统方法的压缩性能对比图。横轴是码率，单位是bpp(bits per point)，纵轴是亮度Y的峰值信噪比PSNR(Peak Signal to Noise Ratio)，单位是分贝db，码率越小，峰值信噪比越大，点云属性压缩性能越好；
以下针对MPEG点云压缩工作组中的官方点云数据集Longdress-vox10-1300.ply，Queen_frame_0200.ply，Soldier-vox10-0690.ply采用本发明方法进行点云属性压缩，以Longdress-vox10-1300.ply为例，如图1所示，具体实施步骤为：
(1)点云属性压缩的编码过程：
点云Longdress-vox10-1300.ply共有857966个点，KD树划分深度设为13，经过划分后共有8192个编码块，块内点的数量为104或105，以第1个编码块为例，有104个点，块内的属性信息经过帧内预测、残差变换、量化(量化步长为4)之后得到一个104×3的量化矩阵Q 1。
(2)编码端优化对量化矩阵的遍历顺序：
对得到的量化矩阵Q 1，采用7种不同的遍历顺序，比较发现YUV逐列扫描得到的数据流中0元素在末端连续出现个数最多，因此选择该模式将二维的量化矩阵转化为一维的数据流S 1，长度为312，同时记录第1个块对应的遍历模式m 1＝1。
(3)删除数据流末尾的0元素：
对得到的数据流S 1，删除其末端连续出现的266个0元素，得到新的数据流长度为46。8192个编码块都进行相同的操作之后，数据流分别写入各自的编码块信息，再将量化步长、遍历模式和预测模式等信息写入压缩头信息中，统一进行熵编码，输出的最终码流文件的结构如图3所示。
(4)解码端参考几何信息复原量化矩阵：
解码端输入码流文件进行熵解码，获取头信息和8192个块的编码块信息，以第1个块为例，得到裁剪后的数据流长度为46，需要知道被删除的0元素的个数来恢复原始的数据流。在解码端对点云的几何信息进行与编码端相同的KD树划分，得到8192个编码块，第1个块内有104个点，则可以求出被删除的0元素有266个。补齐0元素之后的数据流长度为312，根据头信息中第1个块的遍历模式m 1＝1，采用YUV逐列扫描的遍历顺序，将数据流转化为104×3的量化矩阵。，
(5)点云属性压缩的解码过程：
对复原的量化矩阵结合量化步长、预测模式等信息，依次进行反量化、逆变换、预测补偿，求出点云的属性信息。点云属性压缩的性能由码率和峰值信噪比PSNR(Peak Signal to Noise Ratio)来衡量，其中码率的单位是bpp(bits per point)，PSNR的单位是分贝dB。
为了验证本发明的一种基于删除量化矩阵中0元素的点云属性压缩方法的效 果，我们使用上述3个数据集Longdress-vox10-1300.ply，Queen-frame-0200.ply，Soldier-vox10-0690.ply进行实验，在压缩性能上与现有的方法对比结果如图4所示。
从图4可以看出，在使用相同的基于帧内预测、图变换、量化、熵编码的点云属性压缩编码器情况下，引入本发明的删除量化矩阵中0元素的方法后，在测试的三类典型的点云数据上，产生了稳定且显著的性能增益。本方法在点云属性压缩的编码端对量化矩阵采用最优遍历顺序，使0元素集中分布在数据流末端并删除，减少冗余的信息；在解码端结合点云几何信息补齐被删除的0元素，按照遍历顺序复原量化矩阵，保证不引入新的误差。实验结果表明在各个码率点下，点云属性压缩的性能均得到改善，本发明产生的增益稳定，优点突出。
需要注意的是，公布实施例的目的在于帮助进一步理解本发明，但是本领域的技术人员可以理解：在不脱离本发明及所附权利要求的精神和范围内，各种替换和修改都是可能的。因此，本发明不应局限于实施例所公开的内容，本发明要求保护的范围以权利要求书界定的范围为准。
工业实用性
本发明提供一种基于删除量化矩阵中0元素的点云属性压缩方法，针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，将这些0删除后进行熵编码，降低数据流的数据量，减小编码后生成的码流，在解码端结合点云几何信息复原被删除的0元素，保证本方法不引入额外的误差；随着三维扫描设备(激光、雷达等)的快速发展，点云的精度、分辨率更高。高精度点云广泛应用于城市数字化地图的构建，在如智慧城市、无人驾驶、文物保护等众多热门研究中起技术支撑作用。

Claims (7)

  1. 一种基于删除量化矩阵中0元素的点云属性压缩方法，针对点云属性压缩过程中的量化矩阵，在编码端采用最优的遍历顺序使0元素在生成的数据流中集中分布在末端，将这些0删除后进行熵编码，包括如下步骤：
    1)点云属性压缩的编码过程
    针对待压缩的点云数据，首先根据几何信息进行KD树划分，KD树划分的最后一层所得到的块即为点云的编码块，每一个编码块内的属性信息经过帧内预测、残差变换、量化之后会得到一个量化矩阵；
    2)编码端优化对量化矩阵的遍历顺序：
    对于每一个量化矩阵，采用7种不同的遍历顺序，将二维的矩阵转化为一维的数据流，比较0元素在数据流末端连续出现的个数，选择0元素在末端连续出现个数最多的数据流为最优数据流，同时记录对应的遍历模式；
    3)删除数据流末端的0元素：
    得到了最优数据流之后，将数据流末端所有连续出现的0都删除，得到裁剪后的数据流，对所有编码块进行相同操作之后，统一进行熵编码，得到点云属性压缩的码流；
    4)解码端参考几何信息复原量化矩阵：
    在解码对码流进行熵解码得到裁剪后的数据流，结合点云的几何信息，求出被删除的0元素的个数并补齐得到原始的数据流，依照遍历顺序将一维的数据流复原为二维的量化矩阵；
    5)点云属性压缩的解码过程：
    按顺序对复原出的量化矩阵进行反量化、逆变换、预测补偿，解码出点云的属 性信息。
  2. 如权利要求1所述点云属性压缩方法，其特征是，步骤1)中点KD树划分方法为二元划分方法；设待处理的点云共有N个点，KD树设定的划分深度为d，经过对点云d次划分后，得到2 d个编码块，每个块内点的个数接近，有n个或n+1个，n的计算方法如式1。对所有的编码块按照广度优先的遍历顺序进行编号
    Figure PCTCN2018086793-appb-100001
    该编号将作为编码块后期处理的顺序。
    Figure PCTCN2018086793-appb-100002
  3. 如权利要求1所述点云属性压缩方法，其特征是，步骤1)中得到的量化矩阵大小与编码块中的点个数有关，为n×3或(n+1)×3。
  4. 如权利要求1所述点云属性压缩方法，其特征是，步骤2)中对于每一个量化矩阵，采用7种不同的遍历顺序，分别为：YUV逐行扫描、YUV逐列扫描、YVU逐列扫描、UYV逐列扫描、UVY逐列扫描、VYU逐列扫描、VUY逐列扫描。扫描后n×3的二维矩阵被转化为长度为3n的一维数据流，选择0元素在末端连续出现个数最多的数据流为最优数据流，同时记录对应的遍历模式m i。
  5. 如权利要求1所述点云属性压缩方法，其特征是，步骤3)中选出的最优数据流长度为3n，假设数据流末端连续出现的0元素个数为l i，则裁剪后的数据流长度为3n-l i。
  6. 如权利要求1所述点云属性压缩方法，其特征是，步骤4)中熵解码得到裁剪后的数据流，长度为l c，需要知道被删除的0元素的个数来恢复原始的数据流。在解码端对点云的几何信息进行与编码端相同的KD树划分，得到2 d个编码块，每个块内有n个(或n+1个)点，再对所有的编码块按照广度优先的遍历顺序进行编号，得到的结果与编码端一一对应，则根据式2可以求出被删除的的0元素的个数l 0。 补齐0元素之后的数据流长度为3n，根据保存的遍历模式m i将数据流转化为n×3的量化矩阵。
    l 0＝3n-l c                 (式2)
  7. 如权利要求1所述点云属性压缩方法，其特征是，步骤5)中具体细节如下：
    (7-1)经过步骤4)得到了量化矩阵，依次对量化矩阵进行反量化、逆变换、预测补偿，求出点云的属性信息；
    (7-2)基于删除量化矩阵中0元素的点云属性压缩方法的码流主要由压缩头信息和编码块信息两大部分组成；头信息主要包括量化步长、预测模式信息、量化矩阵的遍历模式信息等；编码块信息以编码块为单位，按照编码块的顺序排列，每个块内主要包括编码块的颜色残差信息；
    (7-3)点云属性压缩的性能由码率和峰值信噪比PSNR(Peak Signal to Noise Ratio)来衡量，其中码率的单位是bpp(bits per point)，PSNR的单位是分贝dB；码率越小，PSNR越大，点云属性压缩性能越好。
PCT/CN2018/086793 2018-05-03 2018-05-15 一种基于删除量化矩阵中0元素的点云属性压缩方法 WO2019210531A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/045,894 US11216985B2 (en) 2018-05-03 2018-05-15 Point cloud attribute compression method based on deleting 0 elements in quantisation matrix

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810412818.2 2018-05-03
CN201810412818.2A CN108833927B (zh) 2018-05-03 2018-05-03 一种基于删除量化矩阵中0元素的点云属性压缩方法

Publications (1)

Publication Number Publication Date
WO2019210531A1 true WO2019210531A1 (zh) 2019-11-07

Family

ID=64148037

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/086793 WO2019210531A1 (zh) 2018-05-03 2018-05-15 一种基于删除量化矩阵中0元素的点云属性压缩方法

Country Status (3)

Country Link
US (1) US11216985B2 (zh)
CN (1) CN108833927B (zh)
WO (1) WO2019210531A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114332259A (zh) * 2021-12-29 2022-04-12 福州大学 一种基于车载激光雷达的点云编解码方法

Families Citing this family (56)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11818401B2 (en) 2017-09-14 2023-11-14 Apple Inc. Point cloud geometry compression using octrees and binary arithmetic encoding with adaptive look-up tables
US10897269B2 (en) 2017-09-14 2021-01-19 Apple Inc. Hierarchical point cloud compression
US10861196B2 (en) 2017-09-14 2020-12-08 Apple Inc. Point cloud compression
US10909725B2 (en) 2017-09-18 2021-02-02 Apple Inc. Point cloud compression
US11113845B2 (en) 2017-09-18 2021-09-07 Apple Inc. Point cloud compression using non-cubic projections and masks
US10607373B2 (en) 2017-11-22 2020-03-31 Apple Inc. Point cloud compression with closed-loop color conversion
WO2019183113A1 (en) 2018-03-20 2019-09-26 Pcms Holdings, Inc. System and method for optimizing dynamic point clouds based on prioritized transformations
US10867414B2 (en) 2018-04-10 2020-12-15 Apple Inc. Point cloud attribute transfer algorithm
US10909727B2 (en) 2018-04-10 2021-02-02 Apple Inc. Hierarchical point cloud compression with smoothing
US11010928B2 (en) 2018-04-10 2021-05-18 Apple Inc. Adaptive distance based point cloud compression
US10909726B2 (en) 2018-04-10 2021-02-02 Apple Inc. Point cloud compression
US10939129B2 (en) 2018-04-10 2021-03-02 Apple Inc. Point cloud compression
WO2019240215A1 (ja) * 2018-06-13 2019-12-19 パナソニック インテレクチュアル プロパティ コーポレーション オブ アメリカ 三次元データ符号化方法、三次元データ復号方法、三次元データ符号化装置、及び三次元データ復号装置
US11017566B1 (en) 2018-07-02 2021-05-25 Apple Inc. Point cloud compression with adaptive filtering
US11202098B2 (en) 2018-07-05 2021-12-14 Apple Inc. Point cloud compression with multi-resolution video encoding
US11012713B2 (en) 2018-07-12 2021-05-18 Apple Inc. Bit stream structure for compressed point cloud data
JP7415937B2 (ja) * 2018-10-02 2024-01-17 ソニーグループ株式会社 画像処理装置および方法
US11367224B2 (en) 2018-10-02 2022-06-21 Apple Inc. Occupancy map block-to-patch information compression
US10853973B2 (en) * 2018-10-03 2020-12-01 Apple Inc. Point cloud compression using fixed-point numbers
US11430155B2 (en) 2018-10-05 2022-08-30 Apple Inc. Quantized depths for projection point cloud compression
JPWO2020116563A1 (ja) * 2018-12-06 2021-10-28 パナソニック インテレクチュアル プロパティ コーポレーション オブ アメリカＰａｎａｓｏｎｉｃ Ｉｎｔｅｌｌｅｃｔｕａｌ Ｐｒｏｐｅｒｔｙ Ｃｏｒｐｏｒａｔｉｏｎ ｏｆ Ａｍｅｒｉｃａ 三次元データ符号化方法、三次元データ復号方法、三次元データ符号化装置、及び三次元データ復号装置
CN118781209A (zh) * 2018-12-14 2024-10-15 交互数字Vc控股公司 用于对空间数据进行程序化着色的方法和装置
CN111479114B (zh) * 2019-01-23 2022-07-22 华为技术有限公司 点云的编解码方法及装置
EP3935857A1 (en) 2019-03-08 2022-01-12 Vid Scale, Inc. Methods and apparatus for adaptive streaming of point clouds
WO2020187283A1 (zh) * 2019-03-19 2020-09-24 华为技术有限公司 点云编码方法、点云解码方法、装置及存储介质
CN118042135A (zh) * 2019-03-19 2024-05-14 华为技术有限公司 点云编码方法、点云解码方法、装置及存储介质
WO2020190093A1 (ko) * 2019-03-20 2020-09-24 엘지전자 주식회사 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
WO2020197228A1 (ko) * 2019-03-22 2020-10-01 엘지전자 주식회사 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
US11057564B2 (en) 2019-03-28 2021-07-06 Apple Inc. Multiple layer flexure for supporting a moving image sensor
CN112384950A (zh) * 2019-06-12 2021-02-19 浙江大学 点云的编解码方法及装置
WO2020248243A1 (zh) * 2019-06-14 2020-12-17 深圳市大疆创新科技有限公司 三维数据点的编解码方法和装置
CN111699697B (zh) * 2019-06-14 2023-07-11 深圳市大疆创新科技有限公司 一种用于点云处理、解码的方法、设备及存储介质
WO2020248562A1 (zh) * 2019-06-14 2020-12-17 深圳市大疆创新科技有限公司 一种用于点云处理、解码的方法、设备及存储介质
US11711544B2 (en) 2019-07-02 2023-07-25 Apple Inc. Point cloud compression with supplemental information messages
WO2021000658A1 (zh) 2019-07-03 2021-01-07 Oppo广东移动通信有限公司 点云编解码方法、编码器、解码器及计算机存储介质
CN112449754B (zh) * 2019-07-04 2024-03-08 深圳市大疆创新科技有限公司 一种数据编码、数据解码方法、设备及存储介质
CN112385238B (zh) * 2019-07-10 2023-03-24 深圳市大疆创新科技有限公司 一种数据编码、数据解码方法、设备及存储介质
US11627314B2 (en) 2019-09-27 2023-04-11 Apple Inc. Video-based point cloud compression with non-normative smoothing
US11562507B2 (en) 2019-09-27 2023-01-24 Apple Inc. Point cloud compression using video encoding with time consistent patches
EP4040793A4 (en) * 2019-09-30 2023-01-11 Guangdong Oppo Mobile Telecommunications Corp., Ltd. TRANSFORMATION METHOD, INVERSE TRANSFORMATION METHOD, ENCODER, DECODER AND STORAGE MEDIUM
US11538196B2 (en) 2019-10-02 2022-12-27 Apple Inc. Predictive coding for point cloud compression
JP2022550877A (ja) * 2019-10-03 2022-12-05 エルジー エレクトロニクス インコーポレイティド ポイントクラウドデータ送信装置、ポイントクラウドデータ送信方法、ポイントクラウドデータ受信装置及びポイントクラウドデータ受信方法。
US11895307B2 (en) 2019-10-04 2024-02-06 Apple Inc. Block-based predictive coding for point cloud compression
CN113632492B (zh) * 2019-12-06 2024-07-12 深圳市大疆创新科技有限公司 一种用于点云处理、解码的方法、设备及存储介质
US11798196B2 (en) 2020-01-08 2023-10-24 Apple Inc. Video-based point cloud compression with predicted patches
US11625866B2 (en) 2020-01-09 2023-04-11 Apple Inc. Geometry encoding using octrees and predictive trees
US20230154052A1 (en) * 2020-04-24 2023-05-18 Lg Electronics Inc. Point cloud data transmission device, point cloud data transmission method, point cloud data reception device and point cloud data reception method
CN116132671A (zh) * 2020-06-05 2023-05-16 Oppo广东移动通信有限公司 点云压缩方法、编码器、解码器及存储介质
CN112689997B (zh) * 2020-06-10 2023-06-20 深圳市大疆创新科技有限公司 点云的排序方法和装置
US11620768B2 (en) 2020-06-24 2023-04-04 Apple Inc. Point cloud geometry compression using octrees with multiple scan orders
US11615557B2 (en) 2020-06-24 2023-03-28 Apple Inc. Point cloud compression using octrees with slicing
US11948338B1 (en) 2021-03-29 2024-04-02 Apple Inc. 3D volumetric content encoding using 2D videos and simplified 3D meshes
US12101160B2 (en) * 2022-04-12 2024-09-24 King Fahd University Of Petroleum And Minerals Blind two-dimensional-SIMO channel identification using helix transform and cross relation technique
WO2024014902A1 (ko) * 2022-07-13 2024-01-18 엘지전자 주식회사 포인트 클라우드 데이터 송신 장치, 포인트 클라우드 데이터 송신 방법, 포인트 클라우드 데이터 수신 장치 및 포인트 클라우드 데이터 수신 방법
CN116756383B (zh) * 2023-08-18 2023-11-21 凯尔菱电（山东）电梯有限公司 基于物联网的电梯数据管理系统及方法
CN118656537A (zh) * 2024-08-22 2024-09-17 山东盛德智能科技股份有限公司 基于物联网的网站数据自动化爬取方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106952297A (zh) * 2017-03-22 2017-07-14 电子科技大学 一种三维激光扫描数据点云度压缩方法
US20170214943A1 (en) * 2016-01-22 2017-07-27 Mitsubishi Electric Research Laboratories, Inc. Point Cloud Compression using Prediction and Shape-Adaptive Transforms
CN107403456A (zh) * 2017-07-28 2017-11-28 北京大学深圳研究生院 一种基于kd树和优化图变换的点云属性压缩方法

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6839468B1 (en) * 2000-06-28 2005-01-04 International Business Machines Corporation Scaling of multi-dimensional data in a hybrid domain
US8421804B2 (en) * 2005-02-16 2013-04-16 At&T Intellectual Property Ii, L.P. System and method of streaming 3-D wireframe animations
CN102750449B (zh) * 2012-06-20 2015-05-20 北京航空航天大学 基于分步三维空间-特征域映射的点云直线特征提取方法
CN103489169A (zh) * 2013-08-14 2014-01-01 西北工业大学 一种改进的基于最小二乘法的深度数据拼接方法
PT3111560T (pt) * 2014-02-27 2021-07-08 Ericsson Telefon Ab L M Método e aparelho para indexação e desindexação de quantificação vetorial em pirâmide de vetores de amostra de áudio/vídeo
US9734595B2 (en) * 2014-09-24 2017-08-15 University of Maribor Method and apparatus for near-lossless compression and decompression of 3D meshes and point clouds
US10430975B2 (en) * 2016-11-17 2019-10-01 Google Llc Advanced k-D tree encoding for point clouds by most significant axis selection
US10496336B2 (en) * 2016-11-17 2019-12-03 Google Llc K-D tree encoding for point clouds using deviations
US10861196B2 (en) * 2017-09-14 2020-12-08 Apple Inc. Point cloud compression
US10909725B2 (en) * 2017-09-18 2021-02-02 Apple Inc. Point cloud compression
US11010928B2 (en) * 2018-04-10 2021-05-18 Apple Inc. Adaptive distance based point cloud compression
US10853973B2 (en) * 2018-10-03 2020-12-01 Apple Inc. Point cloud compression using fixed-point numbers

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170214943A1 (en) * 2016-01-22 2017-07-27 Mitsubishi Electric Research Laboratories, Inc. Point Cloud Compression using Prediction and Shape-Adaptive Transforms
CN106952297A (zh) * 2017-03-22 2017-07-14 电子科技大学 一种三维激光扫描数据点云度压缩方法
CN107403456A (zh) * 2017-07-28 2017-11-28 北京大学深圳研究生院 一种基于kd树和优化图变换的点云属性压缩方法

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114332259A (zh) * 2021-12-29 2022-04-12 福州大学 一种基于车载激光雷达的点云编解码方法

Also Published As

Publication number Publication date
CN108833927B (zh) 2019-08-16
US20210142522A1 (en) 2021-05-13
US11216985B2 (en) 2022-01-04
CN108833927A (zh) 2018-11-16

Similar Documents

Publication Publication Date Title
WO2019210531A1 (zh) 一种基于删除量化矩阵中0元素的点云属性压缩方法
US10939123B2 (en) Multi-angle adaptive intra-frame prediction-based point cloud attribute compression method
US11122293B2 (en) Intra-frame prediction-based point cloud attribute compression method
WO2021022621A1 (zh) 一种基于邻居的权重优化的点云帧内预测方法及设备
CN114339215B (zh) 点云编解码方法、编码器、解码器及计算机存储介质
WO2019153342A1 (zh) 一种基于增强图变换的点云属性压缩方法
CN112218079A (zh) 一种基于空间顺序的点云分层方法、点云预测方法及设备
US9819964B2 (en) Limited error raster compression
WO2022057091A1 (zh) 一种点云属性编码方法、解码方法、编码设备及解码设备
CN108028945A (zh) 通过使用单例系数更新执行变换的设备和方法
EP2723071A1 (en) Encoder, decoder and method
CN113518226A (zh) 一种基于地面分割的g-pcc点云编码改进方法
US20240040101A1 (en) Method and device for compressing data representative of a volumetric three-dimensional scene with a view to real-time decompression for online viewing
CN101860753A (zh) 一种基于分形的视频压缩与解压缩方法
CN106559668B (zh) 一种基于智能量化技术的低码率图像压缩方法
CN107277521A (zh) 一种基于hevc视频编码的tu树状结构抉择算法
CN114025146B (zh) 基于场景流网络与时间熵模型的动态点云几何压缩方法
Zhu et al. An improved SPIHT algorithm based on wavelet coefficient blocks for image coding
CN107770537B (zh) 基于线性重建的光场图像压缩方法
WO2023168712A1 (zh) 零游程值编解码方法及视频编解码方法、装置和系统
CN105812803A (zh) 一种针对变换单元tu的残差舍弃方法及装置
Krishna et al. Incorporation of DCT and MSVQ to Enhance Image Compression Ratio of an image
WO2024007253A1 (zh) 点云率失真优化方法及属性压缩方法、装置和存储介质
WO2023184196A1 (zh) 零游程值编解码方法及视频编解码方法、装置和系统
CN116458158A (zh) 帧内预测方法及装置、编解码器、设备、存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18917122

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18917122

Country of ref document: EP

Kind code of ref document: A1