Modular 2DPCA Face Recognition Algorithm Based on Image Segmentation

In order to further improve the recognition rate and computing efficiency of modular 2DPCA in face recognition, an improved modular 2DPCA method based on image segmentation is proposed. Firstly, segmentation of threshold value optimization is utilized to segment face image of training samples into several non-overlapping sub-image spaces so that the pixel number has uniform distribution in each sub-image space. Then, sub-images are synthesized, and the composite image has only few gray levels. Finally, modular 2DPCA is utilized to extract feature of composite image, and the nearest distance classification is used to distinguish each face. The experimental results on ORL face database show that the proposed method on the recognition rate is better than ordinary modules 2DPCA methods.