Image Feature Extraction Based on Support Vector Machine and Multi-DSP Combined Structure

An image feature extraction method based on support vector machine (SVM) is presented in this paper, which first seeks the optimal separating hyperplane in small samples and then projects image data in the corresponding normal direction. In multiclass cases, the method has an optimal choose for selecting projecting axis by some sub-SVMs with simplified structure. A multi-DSP combined structure system has been designed to implement this method by TMS320DM648 and TMS320DM6446. The results show the proposed method is effective, and also meets the real-time requirement.

[1]  Bernhard Schölkopf,et al.  Kernel Principal Component Analysis , 1997, ICANN.

[2]  B. Scholkopf,et al.  Fisher discriminant analysis with kernels , 1999, Neural Networks for Signal Processing IX: Proceedings of the 1999 IEEE Signal Processing Society Workshop (Cat. No.98TH8468).

[3]  David G. Stork,et al.  Pattern Classification , 1973 .

[4]  S T Roweis,et al.  Nonlinear dimensionality reduction by locally linear embedding. , 2000, Science.

[5]  Gu Yan-feng.  The Research of Simplification of Structure of Multi-class Classifier of Support Vector Machine , 2005 .