Perceptual pre-processing filter for adaptive video on demand content delivery

We describe the use of perceptual pre-processing to reduce the bitrate needed for delivery of Video on Demand (VOD) content. The proposed system exploits estimated viewing conditions to remove image oscillations which are not visible to the user under the specific viewing conditions. The pre-processing uses parameters such as: viewing distance, pixel density, ambient illumination, etc. A model of human visual system contrast sensitivity is used to remove oscillations in the image data which cannot be seen and need not be encoded. Experiments demonstrate significant bitrate savings compared to conventional encoding methods which do not exploit viewing conditions.