当前位置:网站首页>【点云压缩】Sparse Tensor-based Point Cloud Attribute Compression
【点云压缩】Sparse Tensor-based Point Cloud Attribute Compression
2022-06-12 02:44:00 【Jonathan_Paul 10】
Sparse Tensor-based Point Cloud Attribute Compression
该文运用了稀疏卷积,将多个稀疏卷积堆积成VAE的基本框架;用这个基本框架来做learning-based的方法。值得一提的是,本文是用在属性编码上的,而learning-based用在属性编码上很少。
Ideas
Background
Related work
End-to-End (E2E) Learning
对一个E2E的架构,一般会分为两个编码器:一个为主要的编码器(Main codec)和超先验的编码器(Hyper codec)。
在超先验编码器中,它会被Hyper Encoder提取先验信息,之后用Hyper Decoder来解码得到main codec的均值与方差 ( μ i , σ i ) (\mu_i,\sigma_i) (μi,σi)。
Sparse convolution
要解决的是对稀疏数据进行更有效的特征提取问题。我们知道,点云是具有稀疏性的,如果运用传统的卷积计算,花销势必会很大。因此,人们使用稀疏卷积来有效地计算稀疏数据,而并非对图片像素点(2D)和空间的体素点一一扫描。[1]
Quantization and Rate-Distortion
量化的粗糙程度决定了最终模型表现的好坏(即失真 D D D的多少)。而码率 R R R直接决定了量化的粗糙程度。
Model
Definition
首先需要定义一个Sparse Tensor: { C → , F → } \{\overrightarrow{\mathbf{C}}, \overrightarrow{\mathbf{F}}\} { C,F}。其中, C → = { ( x i , y i , z i ) ∣ i ∈ [ 0 , N − 1 ] } \overrightarrow{\mathbf{C}}=\left\{\left(x_{i}, y_{i}, z_{i}\right) \mid i \in[0, N-1]\right\} C={ (xi,yi,zi)∣i∈[0,N−1]}、 F → = { ( R i , G i , B i ) ∣ i ∈ [ 0 , N − 1 ] } \overrightarrow{\mathbf{F}}=\left\{\left(R_{i}, G_{i}, B_{i}\right) \mid i \in[0, N-1]\right\} F={ (Ri,Gi,Bi)∣i∈[0,N−1]}

在该模型框架中,熵编码模型将会被使用。而针对熵编码模型,作者沿用了[2]中的先验知识,以提升熵编码模型的表现。
Reference
[1] https://towardsdatascience.com/how-does-sparse-convolution-work-3257a0a8fd1
[2] VARIATIONAL IMAGE COMPRESSION WITH A SCALE HYPERPRIOR
边栏推荐
- WPS表格 学习笔记 - 高亮显示重复值
- Proxy and reflection (II)
- Introduction to architecture - who moved my cake
- Navicat for MySQL 11 Linux cracking method
- errno: -4091, syscall: ‘listen‘, address: ‘::‘, port: 8000
- Force deduction solution summary interview question 01.05 Edit once
- Maya Front Office Rendering plug - in Mel script Tool
- Force deduction solution summary 883- projected area of 3D shape
- Start ticwatch2
- El upload upload file
猜你喜欢

【高代码文件格式API】道宁为您提供文件格式API集——Aspose,只需几行代码即可创建转换和操作100多种文件格式

Maya Front Office Rendering plug - in Mel script Tool

架构入门讲解 - 谁动了我的蛋糕

In 2022, don't you know the difference between arrow function and ordinary function?

Application of ard3m motor protector in coal industry

Ue4\ue5 touch screen touch event: single finger and double finger

Getting started with RPC

Drawcall, batches, setpasscall in unity3d

About 100 to realize the query table? Really? Let's experience the charm of amiya.

Abaqus中批量对节点施加集中力荷载
随机推荐
Calculus review 2
How to build urban smart bus travel? Quick code to answer
[high code file format API] downing provides you with the file format API set Aspose, which can create, convert and operate more than 100 file formats in just a few lines of code
Cupp dictionary generation tool (similar tools include crunch)
Force deduction solution summary 388- longest absolute path of file
Pydub MP3 to WAV
Ue4\ue5 touch screen touch event: single finger and double finger
Force deduction solution summary 449 serialization and deserialization binary search tree
DbNull if statement - DbNull if statement
如何防止商场电气火灾的发生?
Research Report on Chinese psoriasis drug market evaluation and investment direction (2022 Edition)
errno: -4078, code: ‘ECONNREFUSED‘, syscall: ‘connect‘, address: ‘127.0.0.1‘, port: 3306;postman报错
el-upload上传文件
函数模板 Function Templates
Xcall cluster script (view JPS command)
One article to show you how to understand the harmonyos application on the shelves
I2C protocol overview
Force deduction solution summary 398 random number index
Force deduction solution summary 699- dropped blocks
微积分复习2