压缩感知与动态纹理模型的视频信号分类  

Video signal classification of compressed sensing and dynamic texture model

作　　者：蒋行国[1] 李志丰[1] 张龙[1] 

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004

出　　处：《电子技术应用》2015年第5期159-162,166,共5页Application of Electronic Technique

摘　　要：针对视频数据的动态纹理特性,提出结合视频压缩感知技术,首先通过压缩采样技术对视频数据进行采样,得到少量的采样数据;然后建立线性动态系统模型,通过少量的压缩采样数据直接估计出模型参数;最后通过计算模型间的马氏距离实现动态纹理视频数据的分类。实验结果表明,提出的压缩感知参数估计方法在20%的低采样率情况下,对交通视频数据的分类正确率达到87%以上。Based on the feature of dynamic texture in video data, this paper presents a method, which combine with the tech- nology of video compressive sensing. Firstly,a few video sampling data is gotten using the technology of compressive sampling. Then linear dynamical system is set up and the model parameters from the little sampling data also are estimated. Finally, the martin distance between models is calculated to classify texture video data. Results show that at the sampling rate of 20%, the rate of ac- curacy about classifying video data is over 87% using the proposed method.

关 键 词：动态纹理模型 压缩感知 马氏距离 视频信号分类 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]