Grid-Based Parallel Elastic Graph Matching Face Recognition Method

This paper presents a grid-based parallel elastic graph matching face recognition method. We firstly divide the face into several sub-regions by geometric features decomposing algorithm, and match the sub-regions of probe face image with the corresponding sub-regions of the sample faces by sub-region elastic graph matching algorithm, and these matching processes can be assigned onto a lot of nodes in the multimedia service grid and scheduled by the optimal costs algorithm. Eventually, we obtain the match degrees between the probe face and the sample faces by integrating the weighted matched results of sub-regions. We carried out extensive experiments on the special face database and HMMC standard face recognition database. Compared with the previous algorithms, our algorithm offers better recognition accuracy and more rapid recognition speed.