Adaptive loop filter technology based on analytical design considering local image characteristics

Based on Wiener algorithm, the adaptive loop filter (ALF) technology is applied to minimize coding distortion over a frame. As the ALF is designed based on the Wiener algorithm, coding performance can be improved by spatially adaptive selection of filter coefficients dependent on a local image feature. However, since the conventional ALF approaches designed only a single set of frame basis filter coefficients, the filter coefficients are not necessarily optimal to achieve the significant coding gain. As the conventional work, the modified adaptive loop filter approach that employs the segmentation pattern has been proposed. This approach divides the current frame into a few segments based on the structure of image texture, and the filter coefficients are determined for each segment type. Although 14 candidates are defined as available segmentation modes, further improvement can be expected because such segmentation can not handle the complicated combination of various texture characteristics within a video frame. From this perspective, we propose an enhanced ALF approach based on spatially adaptive control of filter design dependent on the texture property of the local image region. Experimental results show that the bit reduction performance is improved by 1.6 points in the maximum case against the conventional scheme. Furthermore, the proposed scheme contributes to improving the subjective picture quality.