期刊名称：International Journal on Computer Science and Engineering
印刷版ISSN：2229-5631
电子版ISSN：0975-3397
出版年度：2010
卷号：2
期号：5
页码：1552-1557
出版社：Engg Journals Publications
摘要：Facial expression analysis is rapidly becoming an area of intense interest in computer science and human-computer interaction design communities. The most expressive way humans display emotions is through facial expressions. In this paper a method is implemented using 2D appearance-based local approach for the extraction of intransient facial features and recognition of four facial expressions. The algorithm implements Radial Symmetry Transform and further uses edge projection analysis for feature extraction and creates a dynamic spatiotemporal representation of the face, followed by classification into one of the expression classes. The algorithm achieves an accuracy of 81.0% for facial expression recognition from grayscale image.
