An Improved In-loop Deblocking Filter for Avs-P2

This paper presents an efficient in-loop deblocking filter algorithm which is based on human visual system to reduce blocking artifacts for AVS-P2 encoding. A theoretical analysis is performed to study the condition of boundary strength decision using human visual system characteristics for in-loop deblocking filter in AVS-P2. As a result, we map our boundary strength decision which is based on characteristics of the human visual system to the three filter levels in AVS-P2. Compared with original algorithms in reference software rm52j, the proposed algorithm provides a new and efficient method to reduce blocking artifacts. The experimental results demonstrate this new in-loop deblocking filter method improves the subjective video quality and provides better objective video quality in video sequences.