ROI video coding based on H.263+ with robust skin-color detection technique

Abstract

The paper proposes an ROI (region of interest) video coding algorithm based on H.263+. The ROIs are automatically selected by a robust skin-color detection which utilizes the Cr and RGB variance instead of the traditional skin color models, such as YCbCr, HSI, etc. The skin color model defined by Cr and RGB variance can choose the skin color region more accurately than other methods. For the video coding system, a region-based video codec based on H.263+ with the option mode of modified quantization is set up. We adjust the distortion weight parameter and variance at the macroblock layer to control the qualities at different regions. Experimental results show that the proposed method can significantly improve the quality at the ROI. Our method is suitable for real time videoconferencing.