A novel minimum distance clip algorithm for video coding

In the process of video coding, distortion will inevitably exist in the encoded data. Some of the distorted color points may get out of the max range of color space, and a clip module is introduced in to compensate this error to some extent. The clip method in most video coding standards just confine every single component of the YUV color points by judging whether the component is out of the boundary of the color space cube in the aspect of binary pixel length, and this ignores the actual feature of YUV color space. Aiming at this issue, this paper proposes a novel clip algorithm which utilizes the YUV color space model and some geometry theories to compensate the out-of-range point by finding the minimum distance point on the surface of YUV color space. By applying the proposed algorithm to JM software, the experiment results show that video coding performance can be improved comparing to the conventional clip method in both subjective and objective aspects.