Background Subtraction Based on Robust Principle Component A
发布时间：2020-12-31 02:23
本文中,我基于现有的鲁棒性主成分分析算法提出了一个基于鲁棒性主成分分析的在线算法并描述了它的GPU实现和优化。全文一共有四个部分:绪论、理论基础、新算法的提出和新算法的GPU实现方法。第一部分中文章主要介绍了背景提取的概念、研究意义和研究现状。第二部分从背景提取的基本步骤和方法讲起,然后到基于主成分分析的背景提取算法到基于鲁棒主成分分析的背景提取算法的发展进化历程。新算法的提出部分我主要介绍了新算法的整体框架和一些具体的实现细节。在最后一部分GPU实现方法部分中,我介绍了如何将新的在线鲁棒性主成分分析算法通过CUDA技术来实现在GPU上的运行和优化。随着人工智能的不断发展,对机器视觉的需求也越来越大。而对于任何机器人来说,获取视觉信息的最便宜最直接的方式就是使用视频。目前,视频和摄像机已经被使用在视频游戏,安全监控,人体健康检测等诸多领域。对智能机器人来说,最重要的不是对视频的提取,而是通过机器的处理去理解视频中的各种元素。这就需要视频中运动对象的识别和分离,也就是对视频的前后景的分离:对视频中的图像序列进行分析和处理,检测或者识别视频中的运动的对象然后提取出视频中的完整的背景信息。因...
【文章来源】：华中师范大学湖北省 211工程院校 教育部直属院校
【文章页数】：69 页
【学位级别】：硕士
【文章目录】：
acknowledgement
abstract
Nomenclature
Chapter 1 Introduction
1.1 Background
1.2 Research significance
1.3 State of the arts
1.4 Thesis structure
Chapter 2 Background subtraction
2.1 Introduction to background subtraction
2.2 Assumptions in background subtraction
2.3 Steps in background subtraction
2.4 Traditional approaches in background subtraction
Chapter 3 From PCA to RPCA
3.1 Introduction
3.2 Principle Component Analysis
3.3 Robust Principle Component Analysis
3.4 Traditional algorithm for solving RPCA
3.4.1 IT algorithm
3.4.2 APG algorithm
3.4.3 ALM algorithm
Chapter 4 Background modelling by RPCA
4.1 Main point of work in this chapter
4.2 RPCA for background subtraction
4.3 General model of Online Robust PCA with Initialization
4.4 Batch step
4.4.1 ADMM algorithm
4.4.2 Noise reduction
4.5 Online RPCA with Initialization
4.6 Experiments and evaluation
Chapter 5 Implementation and optimisation using CUDA
5.1 Main point of work in this chapter
5.2 Introduce to CUDA
5.3 GPU programming basics
5.4 Algorithm steps
5.4.1 Parallel computing
5.4.2 Pre-processing
5.4.3 Generate the background model
5.5 Implementation details
5.5.1 Alternating Direction Method of Multipliers algorithm
5.5.2 Online Robust PCA with Initialization algorithm
5.5.3 Other processing
Chapter 6 Summary
Appendix A
References
本文编号：2948761
【文章来源】：华中师范大学湖北省 211工程院校 教育部直属院校
【文章页数】：69 页
【学位级别】：硕士
【文章目录】：
acknowledgement
abstract
Nomenclature
Chapter 1 Introduction
1.1 Background
1.2 Research significance
1.3 State of the arts
1.4 Thesis structure
Chapter 2 Background subtraction
2.1 Introduction to background subtraction
2.2 Assumptions in background subtraction
2.3 Steps in background subtraction
2.4 Traditional approaches in background subtraction
Chapter 3 From PCA to RPCA
3.1 Introduction
3.2 Principle Component Analysis
3.3 Robust Principle Component Analysis
3.4 Traditional algorithm for solving RPCA
3.4.1 IT algorithm
3.4.2 APG algorithm
3.4.3 ALM algorithm
Chapter 4 Background modelling by RPCA
4.1 Main point of work in this chapter
4.2 RPCA for background subtraction
4.3 General model of Online Robust PCA with Initialization
4.4 Batch step
4.4.1 ADMM algorithm
4.4.2 Noise reduction
4.5 Online RPCA with Initialization
4.6 Experiments and evaluation
Chapter 5 Implementation and optimisation using CUDA
5.1 Main point of work in this chapter
5.2 Introduce to CUDA
5.3 GPU programming basics
5.4 Algorithm steps
5.4.1 Parallel computing
5.4.2 Pre-processing
5.4.3 Generate the background model
5.5 Implementation details
5.5.1 Alternating Direction Method of Multipliers algorithm
5.5.2 Online Robust PCA with Initialization algorithm
5.5.3 Other processing
Chapter 6 Summary
Appendix A
References
本文编号：2948761
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2948761.html
