With the continuous progress of human computer interaction, face detection as well as facial expression recognition is gaining the attention of researchers from the fields of security, psychology, image processing, and computer vision. In this area the most challenging thing is to recognize accurate facial expression with minimum time requirement. In this work, our main focus is to minimize the time using fusion based Independent Component Analysis (ICA). Research studies show ICA has significant success on face image analysis. Among several architectures of ICA we mainly used here Gaussian kernel based FastICA algorithm due to time efficiency. We apply FastICA on whole faces to recognize facial expressions. Also we apply FastICA on different facial parts, by proposing two algorithms namely WAPA-FastICA and OEPA-FastICA, to analyze the influence of different parts for several basic emotions. Our experiment shows OEPA-FastICA and WAPA-FastICA outperforms existing predominant FastICA algorithm. We also compared these proposed algorithms with our previous PCA based facial expression recognition work.
With the increasing applications of human computer interactive systems, recognizing accurate and application oriented human expressions is becoming a challenging topic. The face is highly attractive biometric trait for expression recognition because of its physiological structure, its robustness and location. In this paper we proposed modified subspace projection method that is an extension of our previous work [11]. Our previous work was FER analysis on full face and half faces by using principal component analysis (PCA) for feature extraction. This is obviously an extension of existing PCA algorithm. In this paper PCA is applied on facial parts like left eye, right eye, nose and mouth for feature extraction. A Flow chart for the whole system is depicted in section 3. The objective of this research is to develop a more effective approach to distinguish between seven prototypic facial expressions, such as neutral, smile, anger, surprise, fear, disgust, and sadness.These techniques clearly outperform our previous paper [11]. The whole procedure is applied on Cohnkanade FEA dataset and we achieved higher accuracy than our previous method.
Abstract-Face and Facial Expression Recognition is a broad research area for its diversified applicability in different applications from security, surveillance to medical diagnosis. The main challenge in this area is to decrease the recognition time as well as to increase the accuracy rate. In this paper, we propose face identification system and facial expression recognition system based on non-negative matrix factorization (NMF). As facial parts are more prominent to express a particular facial expression rather than whole faces and NMF performs part based analysis, so we get a significant result for face recognition. We test on CK+ and JAFFE dataset and we find the face identification accuracy is nearly 99% and 96.24%. But the facial expression recognition (FER) rate is not as good as it required to be. We propose fusion based NMF method and we name it as OEPA-NMF, where OEPA means Optimal Expression specific Parts Accumulation. Our experimental result shows that OEPA-NMF outperforms the predominant NMF.Index Terms-Non-negative matrix factorization (NMF), facial expression recognition (FER), optimal expressionspecific parts accumulation (OEPA), face recognition (FR).
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.