Sharpening filter for interlayer prediction

This paper describes method of predicting enhancement layer pictures from base layer pictures for Scalable extension of High Efficiency Video Coding (SHVC). SHVC stream contains multiple layers of different types of scalability: temporal, spatial or quality. The interlayer prediction requires to increase the amount of high frequency details which is absent in a base layer reference picture. It is realized by the proposed sharpening filter that applied to base layer reference picture. Applying this filter allows to reduce residual between source and the predicted pictures. The simulations under SHVC common test conditions defined by JCT-VC show 1.8% BD-rate reduction on average for all intra-2× [AI] configurations. For Class A test sequences 3.1% BD-rate reduction on average for AI-2× configurations was achieved.