CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleThe Research and Application of CUDA in Intelligence Visual Analysis Algorithms
Abstract智能视频分析算法已经越来越广泛的应用于各个领域,比如交通控制、安全监控和娱乐等领域,并且取得了一定的效果。但是随着数据量的增加,智能视频分析算法在面对海量数据时无法保证实时的计算,这使系统无法承载更多的功能,限制了其应用场景。 虽然CUDA(Compute Unified Device Architecture)技术在最近几年才兴起,但是它已经有了快速的发展,并且已经有很多的应用实例,特别是在娱乐和医学领域。由于GPU价格低廉,已经有越来越多的个人电脑安装有独立GPU。 基于上述原因,本文围绕如何利用CUDA技术对智能视频分析算法进行加速展开研究,并将研究结果应用在实际项目中,主要的工作和贡献有:  详细分析了CUDA技术特点和智能视频分析算法特点,并在此基础上归纳得到CUDA技术加速框架和若干条通用性方法。  运用CUDA技术通用性方法给混合高斯模型加速,加速后的混合高斯模型是加速前的30倍以上,效果明显,证明了通用性加速方法的有效性。  运用CUDA技术通用性方法给HLSIFT算法加速,取得了20倍以上的加速比,具有良好的效果,验证了通用性加速方法的有效性。  将加速后的HLSIFT算法应用于银行虚拟自动取款机项目中。它不仅证明了CUDA技术在实际项目中的有效性,还扩展了智能视频分析算法的应用领域,并且为银行卡密码保护问题提供了新的思路和一种解决途径。 本文不仅在研究上取得了一定的进展,而且将研究结果应用在实际项目中,具有一定的应用价值。
Other AbstractIntelligent visual analysis algorithms are used more and more widely in different areas, such as traffic control, security monitoring, entertainment areas and so on. With the increase in the amount of data, intelligent visual analysis algorithms cannot run in real time in the face of huge amounts of data. This makes systems unable to be integrated with more functions, limiting applications to few scenarios. Although CUDA (Compute Unified Device Architecture) only rises in the recent years, it has gained rapid development and lots of applications, especially in the entertainment and medicine areas. For the low price of GPUs, many desktop personal computers have installed GPU. For the reasons above, the paper focuses on how to use CUDA to accelerate intelligent visual analysis algorithms. And the results are applied in practical projects. The main work and contributions are listed below:  The characteristics of the CUDA technology and intelligent visual analysis algorithms are analyzed in detail. On the basis, the framework of using CUDA technology and several general accelerating approaches are summarized.  The summarized general approaches are used to accelerate Gaussian mixture model. The accelerated algorithm is as 30 times faster as the original one. The effect has proved the validity of the summarized general approaches.  The summarized general approaches are used to accelerate HLSIFT, achieving the speedup of more than 20 times. The good results verify the validity of the summarized general approaches.  The accelerated HLSIFT algorithm is used in the bank virtual automatic teller machine project. It does not only prove the validity of CUDA technology in practical, but also extends the application field of intelligent visual analysis algorithms. In addition, it provides a new way of thinking and a solution for the bank card password protection. The paper does not only make some progress in the research, but also applies research results in practical projects.
KeywordCuda 智能视频分析算法 混合高斯模型 Hlsift 虚拟自动取款机 Cuda Intelligent Visual Analysis Algorithms Gaussian Mixture Model Hlsift Virtual Automatic Teller Machine
Document Type学位论文
Recommended Citation
GB/T 7714
饶超. CUDA技术在智能视频分析算法中的研究和应用[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20082800902906(2173KB) 暂不开放CC BY-NC-SA
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[饶超]'s Articles
Baidu academic
Similar articles in Baidu academic
[饶超]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[饶超]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.