计算机技术与发展2017，Vol.27Issue(2)：37-41,5.DOI:10.3969/j.issn.1673-629X.2017.02.009
基于新型鲁棒字典学习的视频帧稀疏表示
Sparse Representation of Video Frame Based on Novel Robust Dictionary Learning
摘要
Abstract
Dictionary learning is a very effective signal sparse representation method which has been widely used in the field of sparse signal processing.However,in practice,both training and testing samples may be corrupted and contain noises and a few outlier data,which may heavily affect the learning performance of it.Hence,in contrast to the conventional dictionary learning methods that learn the dictionary from clean data,a novel robust dictionary learning algorithm is proposed to handle the outliers in training data.In the proposed algorithm,the alternating proximal linearized method is used for solving the non-convex l0 norm based dictionary learning problem.Thus,the robust dictionary can be learned and outliers can be isolated in the training samples simultaneously.The simulation experimental results demonstrate that the method has the promising robusness and can provide significant performance improvement.关键词
字典学习/稀疏表示/异常数据/鲁棒性Key words
dictionary learning/sparse representation/outlier data/robustness分类
信息技术与安全科学引用本文复制引用
钱阳,李雷..基于新型鲁棒字典学习的视频帧稀疏表示[J].计算机技术与发展,2017,27(2):37-41,5.基金项目
国家自然科学基金资助项目(61070234,61071167,61373137,61501251) （61070234,61071167,61373137,61501251）
江苏省2015年度普通高校研究生科研创新计划项目(KYZZI5_0235) （KYZZI5_0235）
南京邮电大学引进人才科研启动基金资助项目(NY214191) （NY214191）
