Hand Gesture Segmentation from Complex Color-Texture Background Image

Gestures provide a rich, intuitive and natural form of interaction between human and other devices. In this paper an automatic hand gesture segmentation technique from the complex color-texture Image is developed for segmentation of hand gesture with less false positive rate(FPR). In this approach we propose a model for Skin Color Characterization and define a Potential of a Pixel (PoP) which are then used to segment the hand gesture. This new skin segmentation technique takes into account both the color-texture features for efficient segmentation. It is observed that the classifier is robust with respect to usage of hand and mode of hands like front or back side of hand. To evaluate the system the hand gesture images have been acquired from set of students under various complex background. The gesture segmentation technique has false positive rate of nearly 5.7% and true positive rate near to 98.93%.