Learning a Similarity Constrained Discriminative Kernel Dictionary from Concatenated Low-Rank Features for Action Recognition

Shijian HUANG  Junyong YE  Tongqing WANG  Li JIANG  Changyuan XING  Yang LI  

IEICE TRANSACTIONS on Information and Systems   Vol.E99-D   No.2   pp.541-544
Publication Date: 2016/02/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2015EDL8148
Type of Manuscript: LETTER
Category: Pattern Recognition
low-rank feature,  similarity constraint,  kernel method,  human action recognition,  

Full Text: PDF(776.8KB)
>>Buy this Article

Traditional low-rank feature lose the temporal information among action sequence. To obtain the temporal information, we split an action video into multiple action subsequences and concatenate all the low-rank features of subsequences according to their time order. Then we recognize actions by learning a novel dictionary model from concatenated low-rank features. However, traditional dictionary learning models usually neglect the similarity among the coding coefficients and have bad performance in dealing with non-linearly separable data. To overcome these shortcomings, we present a novel similarity constrained discriminative kernel dictionary learning for action recognition. The effectiveness of the proposed method is verified on three benchmarks, and the experimental results show the promising results of our method for action recognition.