Face expression recognition using integrated approach of Local Directional Number and Local Tetra Pattern

Recognizing the human facial expression and emotion by computer is an interesting and challenging problem. This project proposes a method for facial expression recognition (FER) using Local Directional Number Pattern (LDN) and Local Tetra Pattern (LTrP). The proposed FER method removes the noise using Modified Decision Based Unsymmetrical Median Filter (MDBUTMF) method. The Local Directional Number (LDN) pattern descriptor is found based on the eight Gaussian edge descriptors. The proposed method encodes the relationship between the referenced pixel and its neighbors, based on the directions that are calculated using the first-order derivatives in vertical and horizontal directions. The training poses such as disgust, sad, smile and surprise are undergone the Gaussian mask generation, Gaussian edge detection, LDN process, LTrP process, MLDN histogram and LTrP histogram process. These training histograms are used to train up the facial expressions using Support Vector Machine (SVM). The SVM classifier classifies the facial expressions into disgust, sad, smile and surprise. The proposed method increases the FER accuracy at a significant level. The proposed method is a suitable one for any facial expression recognition needs.