Face Image Recognition in Embedded Imaging Sensor Nodes of WSN: Approaches and Identification Realization

摘要

This paper is concerned with approaches of face image recognition and the identification realization in the embedded imaging sensor nodes of wireless sensor network (WSN). Considering that the principal component analysis (PCA), the two-dimensional principal component analysis (2DPCA) and the linear discriminant analysis (LDA) have received significant attention in the human face image recognition, this paper first discusses these dominant image recognition approaches and 2DPCA plus LDA, a novel fast feature extraction technique. They are then both applied to the personal computer and to the embedded imaging sensor nodes for identification. Finally the ORL face image database is selected to evaluate the recognition accuracy and recognition time on the different hardware platforms. The experimental results demonstrate that the 2DPCA plus LDA approach improves the recognition accuracy by the small sample size (SSS) problem solving. Compared to the conventional PCA or 2DPCA approaches, the 2DPCA plus LDA approachadapts itself to the embedded imaging sensor nodes of WSN for identification by the innetwork computing strategy, with the higher recognition accuracy and the lower memory resource requirement.