A Novel Multiple-Channel Video Mixing Scheme for Multiple-Point Processing Unit

Mixing multiple channel video streams is an essential function for the multiple-point processing unit of video conference system. The traditional mixing methods have the disadvantage of high computational complexity and severe image distortion. This paper proposed a mixing method based on the video syntax and semantics mapping in DCT-domain. The experimental results reveal that the PSNR has shown almost 2dB improvement compared with pixel-domain method. This method has been integrated into the multiple-point control unit (MCU) of the H.323 video conference system.