Spatial correlation and texture-based fast Adaptive Loop Filter for HEVC

Many techniques have been developed to improve the coding performance of the emerging next generation High Efficient Video Coding (HEVC) standard. Among these techniques, Adaptive Loop Filter (ALF) is proposed and effectively enhances the coding efficiency, which is however at the cost of much higher computational complexity, thus increases the difficulty of practical real-time video encoding. Therefore, we address this problem and propose a fast algorithm of adaptive loop filter. We pre-judge the filter design mode for every largest coding unit (LCU) based on the spatial correlation and texture information of video content. By skipping some unnecessary mode, the time complexity of adaptive loop filter can be significantly reduced. The experimental results show that the proposed fast ALF algorithm can save on average 71.7% of the time for the adaptive loop filter in comparison with the original algorithm with only negligible video quality degradation.