Kernel-based sparse representation for gesture recognition

• Published in: Pattern recognition Vol. 46; no. 12; pp. 3208 - 3222
• Main Authors: Zhou, Yin, Liu, Kai, Carrillo, Rafael E., Barner, Kenneth E., Kiamilev, Fouad
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.12.2013; Elsevier
• Subjects: Applied sciences; Artificial intelligence; Classification; Coding, codes; Compressive sensing; Computer science; control theory; systems; Computer vision; Dictionaries; Dictionary learning; Exact sciences and technology; Feature extraction; Gesture recognition; Human; Information, signal and communications theory; Pattern recognition; Pattern recognition. Digital image processing. Computational geometry; Representations; Sampling, quantization; Signal and communications theory; Signal processing; Sparse representation; Strategy; Telecommunications and information theory; Time series; Computer vision; Sparse representation; Compressive sensing; Dictionary learning; Gesture recognition; Discriminant analysis; Dictionaries; Image recognition; Time series; Pattern recognition; Algorithm; Kernel method; Learning; Statistical method; Coding; Gesture; Signal processing; Feature extraction; Compressed sensing; Principal component analysis
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2013.06.007

In this paper, we propose a novel sparse representation based framework for classifying complicated human gestures captured as multi-variate time series (MTS). The novel feature extraction strategy, CovSVDK, can overcome the problem of inconsistent lengths among MTS data and is robust to the large variability within human gestures. Compared with PCA and LDA, the CovSVDK features are more effective in preserving discriminative information and are more efficient to compute over large-scale MTS datasets. In addition, we propose a new approach to kernelize sparse representation. Through kernelization, realized dictionary atoms are more separable for sparse coding algorithms and nonlinear relationships among data are conveniently transformed into linear relationships in the kernel space, which leads to more effective classification. Finally, the superiority of the proposed framework is demonstrated through extensive experiments. •An effective feature extraction method is proposed for gesture recognition.•Kernel trick is applied to learning the dictionary for sparse representation.•Comparisons with state-of-the-art methods are performed.•Superior recognition rate is achieved.