An empirical study of selection of the appropriate color space for skin detection: A case of face detection in color images

摘要

In most of the studies related to human computer interaction like face detection, human detection, face recognition and emotion detection in the color image or videos, skin color was used as a cue or a feature. Because it handles some well defined problems such as pose, orientation and partial occlusion and varying illumination conditions. In spite of advantages of using skin as cue for various human computer interaction applications, there is another challenge of selection of suitable color space among various color spaces available. Therefore, in this paper an empirical/experimental study was done by keeping same objects with same illumination or lighting conditions for selection of a suitable color space which one would be capable of skin color detection from color image or videos. Due to the existence of huge imprecision, this study also used fuzzy logic concepts for human skin modeling.