Facial expression recognition based on local directional number pattern and ANFIS classifier

In this work, an efficient algorithm for facial expression recognition using a local feature descriptor, Local Binary Pattern (LBP), Local Directional Number Pattern (LDN) and Soft Computing Technique, Adaptive Neuro-Fuzzy Inference Systems (ANFIS) is presented. In the first experiment local binary pattern is computed using the input image.In the second experiment, the face image is subjected to a Kirsch compass mask that gives the directional information of the image and with the help of masked output Local Directional Number Pattern (LDN) code is computed. The obtained LBP and LDN image is divided into several regions and the distribution of the LBP and LDN features are extracted from them. These features are then concatenated into a feature vector, which is used for ANFIS training and classification. The experimental evaluation of the presented method is carried out using Japanese Female Facial Expression Database (JAFFE) and Indian Face Database (IFD). The results obtained from the experiments prove that the presented method successfully recognize the facial expression variations.