Symmetry description and face recognition using face symmetry based on local binary pattern feature

Considering the face image is approximate symmetry, we proposed a measurement factor to quantify the symmetrical characteristic of face images. Firstly we transform the face image into even-odd face images using parity decomposition method. LBP features extracted in even images to construct training sets. Then we apply Adaboost training algorithm to structure a strong classifier. Experiment proved that this method can overcome environment disturbance, and effectively improve the recognition rate.