ANUVAB: SOFTWARE FOR FACIAL EXPRESSION RECOGNITION ON REAL TIME VIDEO SEQUENCE
Barshan Das, Bikash Santra, Swapna Agarwal
Electronics & Communication Sciences Unit, Indian Statistical Institute, Kolkata, India
Abstract: Face expression analysis and recognition has been one of the fast developing areas. It covers wide range of areas such as emotion analysis, biometrics, and image retrieval. Thus, our objective primarily concentrates on the recognition of facial emotional expressions. In this context we have developed a software "Anuvab". Anuvab can recognize the facial expressions on image sequence or real time video sequences. To do that, Anuvab requires different methods consisting pre-processing and processing parts in solving facial expression analysis problems. Pre-processing part possesses detection and extraction of face images together with the normalization process. Normalization process is to align these extracted images independent of varying environmental conditions such as illumination and orientation. Whereas the processing part aims to extract specific features from the already pre-processed images. Also, processing part inculcate the recognition of facial action units/facial expressions depending on these features. The steps of the resulting systems mainly consist of normalization, feature extraction, feature selection and binary & multi-class classifications of the frontal face images. Finally "Anuvab" is designed after going through aforesaid steps to fulfill our objective. (pdf, CSI annual report, CSI YITPA album)