Skin color segmentation using adaptive PCA and modified elliptic boundary model

Skin color segmentation is a promising research field in computer vision. It has a wide variety of applications such as: development of more effective interfaces for human-computer interaction (HCI), face detection, content-based image retrieval (CBIR) systems, gesture analysis, and medical applications. However, the skin color detection is a challenging task as the skin color in an image is sensitive to various factors like illumination, camera characteristics, and ethnicity. This paper introduces a newly developed method of human skin color detecting and segmentation under different challenges. The skin colors are first fully decorellated using the adaptive color PCA (ACPCA) and then segmented using the modified elliptic boundary model (MEBM). It can be observed from the experimental results that the proposed method successes in skin color detecting and segmentation for different ethnic groups and imaging conditions. In comparison with other famous color models: the dedicated detection rate is about 98% and the false positive rate is 4% on an average. The advantages of the new method for skin color segmentation are the higher accuracy, relatively low computational complexity. Also, it does not require a priori information about the number of objects in the image or applying any post-processing morphological operations for the segmented areas.