Mapped middle view’s inter-view prediction to multiview pre-diction in MVC

Based on the study of several methods involved in getting the benefit of extracting the inter-view redundancy, a new technique is suggested in this paper. The complexity of choosing the reference frames for different views is simplified. Starting with the camera placed in the center of the scene and predicting its frames from the left, right and the middle view, a map table is created per MB (Macro Block). This map table is simply extended to all the other views for respective inter-view prediction. This method of inter-view prediction applies the best for low to medium motion pictures. The complexity is way reduced as only one of the view’s prediction is derived and the same is comprehended to all other views. The method of a variable reference frame buffer where one frame from each of the buffers is stored also contributes to the reduction in complexity of the approach to inter-view prediction. Using the standard multi-view test sequences, few experimental results had proved that this approach can be opted for low motion sequences with a lesser complexity and improved compression with a coding gain of ~30%. The quality loss in PSNR (Peak Signal-to-Noise Ratio) is negligible with a loss of 0.2dB.