Hand Segmentation from Complex Background for Gesture Recognition

Hand gesture recognition has become very popular in the field of human–computer interface in recent years. Hand region identification is a fundamental task in most vision-based gesture recognition systems, since the subsequent detection and then segmentation depends on the quality of segmentation. If the background is complex and the illumination is varying, then the segmentation can be a difficult task. Different physical controllers like data gloves, color bands, mouse, and joysticks are used for human–computer interaction, which hinders natural interface as there is a strong barrier between the user and the computer. In such environments, most hand detection techniques fail to obtain the exact region of the hand shape, especially in cases of dynamic gestures. Meeting these requirements becomes very difficult, due to real-life scenarios. To overcome these problems, in this paper, we propose a new method for static hand detection and contour extraction from a complex background. We employ a new technique, histogram thresholding which gives better result over depth thresholding to improve hand region extraction.