Facial Image Processing

1188 Words5 Pages
Abstract Various applications have been proposed in the field of facial image processing throughout the last forty years or so, among which are face recognition, face detection, gender/age classification, facial expression, etc. However, no application related to the identification of similarity between family members has been introduced yet. This paper is the first experience, which considers this phenomenon and proposes a framework for clustering similar family members. Three features include: “The Whole Face,” “The Facial Feature Perimeters” and “The Ratio between Facial Features” have been used. A color based method is utilized for face localization, an anthropometric based method is used to extract the region of interest and finally a geometric feature-based algorithm is applied to locate the exact facial feature points. After extracting all features, the standard Eigenfeature algorithm is performed in a hierarchical manner and four similar feature classes are selected. First, a set of subjective test on the Facial Family Image Database (FFIDB) is get done to find out which feature is the most effective feature in clustering this kind of similarity. Then, the efficiency of the proposed method is compared with three state of the art clustering algorithms. The results indicate that the analysis power of proposed framework is better than human in many cases. 1. Introduction Various applications have been proposed in the field of facial image processing throughout the last forty years or so, among which are face recognition[], face detection[], gender/age classification[], facial expression[], etc. Face recognition is considered as the most attractive field by researchers and many state of the art algorithms have been proposed... ... middle of paper ... ...ussed in section 3. Section 4 reports an experimental setup and finally the paper is concluded in Section 5. 2. Background No previous work has been reported on any aspects of similarity recognition in images of family faces. However, it is appropriate to review research on face recognition, as many of the issues encountered in our problem are similar to those encountered in related problems. Face recognition systems has been carried out in two distinct paradigms. In the first paradigm researchers first extract facial features such as the eyes, nose, etc., then they utilize clustering/classification algorithms to recognition. The second paradigm treats the complete face image as an input vector and bases analysis and recognition on algebraic transformations of the input space. The current research has adopted these two paradigms for family similarity recognition.
Open Document