An Efficient Reference Image Sharing Method for the Parallel Video Encoding Architecture

Abstract

Video coding hardware is essential for various video applications, and to meet demands for higher resolution, the parallel video encoding architecture is often adopted, in which reference images are shared among encoder modules to maintain coding efficiency. However, conventional sharing methods are not always efficient in utilization of transferred image data. In this paper, we propose a method to reduce the amount of data transfer efficiently by using both pre-transfer and on-demand transfer. Experimental results show that the data transfer can be reduced to 19.8-35.3% of the conventional method on average although the access error rate increases to 0.5-2%. This makes it possible to reduce the required bandwidth of the inter-chip interface by saving the amount of data transfer.