A Novel De-noising Model Based on Independent Component Analysis and Beamlet Transform
A Novel De-noising Model Based on Independent Component Analysis and Beamlet Transform
复制标题
DOI：
10.4304/jmm.7.3.247-253
复制
发表时间：
2012-01
期刊：
影响因子：
--
通讯作者：
Guangming Zhang;Zhiming Cui;Pengpeng Zhao;Jian Wu
中科院分区：
文献类型：
--
作者：
Guangming Zhang;Zhiming Cui;Pengpeng Zhao;Jian Wu
车载视频关键帧处理作为智能交通系统的重要组成部分，起着重要的作用。传统的车辆视频关键帧提取往往存在大量噪声，不能满足识别和跟踪的要求。本文提出了一种将独立分量分析与子束变换相结合的方法。首先，产生一个随机矩阵，将关键帧分离成一个独立的图像进行估计。然后应用子束变换对系数进行优化。最后通过小波变换的逆变换选择系数进行图像重建。相比之下，该方法可以去除更多的噪声和保留更多的细节，并且该方法的效率优于其他传统的去噪方法。
Vehicle video key frame processing as an important part of intelligent transportation systems plays a significant role. Traditional vehicle video key frame extraction often has lots of noises, it can't meet the requirements of the recognition and tracking. In this paper, a novel method which is combined independent component analysis with beamlet transform is proposed. Firstly, a random matrix was produce to separate the key frame into a separated image for estimate. Then beamlet transform was applied to optimize the coefficients. At last, the coefficients were selected for image reconstruction by inverse of the beamlet transform. By contrast, this approach could remove more noises and reserve more details, and the efficiency of our approach is better than other traditional de-noising approaches.