Adaptive video encoding based on skin tone region detection

It is often desirable in today's highly congested inter networks that the data rate may be adaptively controlled in order to prevent congestion in the networks. Nowadays, streaming applications are finding their importance in applications like video conferencing. These applications are responsible for pumping larger amounts of data in a network, as they follow UDP (User Datagram Protocol), so a means that can adapt to a congested network by controlling the rate, at which the data is produced, would be useful. We propose a technique, which can control the data rate, if needed, by encoding the skin regions separately. By varying the quantizer scale factor in a video encoder. the data rate at which a video is encoded may be controlled. This method is termed as Content Based Reduction. In a congested network, a higher scale factor for the background and a lower one for the skin would prove helpful as data rate produced would be lower, and video quality would be less deteriorated as skin regions are encoded in the same (or better) fashion. Our work is a step towards this proposal. We present an algorithm that can detect skin tone regions in the video sequences.