Facial Expression Recognition Using Gabor-Mean-DWT Feature Extraction Technique

Facial Expression Recognition is key vital topic for research in current scenario and has wide application as activity recognition, human-computer interaction, and person satisfaction index evaluation. This paper presents a novel feature extraction technique: Gabor-mean-DWT for automatic facial expression recognition from video image sequences and proposed technique is illumination invariant also. Facial Emotion can be recognized using unique edge and texture pattern on face. Gabor filter is able to extract edges and texture pattern of faces but have a problem of huge dimension and high redundancy. The problem of huge dimension and high redundancy is reduced by proposed Average-DWT feature reduction technique in order to increase accuracy of system. Proposed Gabor-Average-DWT provides a compact feature vector compared to existing Gabor based expression classification. Detailed quantitative analysis is done and results shows that the accuracy of proposed technique is better than existing state of art results.