Human expression recognition based on feature block 2DPCA and Manhattan distance classifier

In order to overcome slow speed of traditional PCA, the paper presents that feature vector can be obtained by feature block two dimensional principal component analysis, and the Manhattan distance classifier output recognition results. Calculation speed can be enhanced efficiently. Compared with Euclidean distance, recognition rate is improved by Manhattan distance. The experiments of training data includes test data and partly includes test data are tested respectively in the Japanese female facial expression database. The compared results show that the proposed approach appeared quicker calculation speed and higher recognition accuracy than other approaches.