Motion compensation method and system based on occlusion detection

Abstract

The invention relates to a motion compensation method and system based on occlusion detection. The motion compensation method comprises the following steps: carrying out motion estimation to obtain a frontward motion vector and a backward motion vector, and detecting a revealed / shielded region; calculating revealing / shielding probability according to an overlapped area of a position where the motion vector reaches an interpolation frame and a block to be processed and calculating an interpolation weight ; calculating to obtain the motion vector of the interpolation frame according to respective corresponding frontward and backward motion vectors of the former frame and the current frame and a detection result of the revealed / shielded region; detecting a crossed part of the revealed / shielded region and a background through the consistency of the motion vector of the interpolation frame and the interpolation weight and correcting the interpolation weight; carrying out interpolation to obtain pixel-by-pixel interpolation weight; interpolating according to the motion vector of the interpolation frame and the pixel interpolation weight to finally obtain a video with the improved frame rate. The accuracy of detecting the revealed / shielded region is improved and a halo phenomenon in a motion compensation image is remarkably alleviated.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a motion compensation method and system based on occlusion detection. Background technique [0002] Due to the hold-type display characteristics of liquid crystal displays, motion blur and jitter will appear when playing moving pictures. The above-mentioned distortion can be alleviated by increasing the video frame rate through motion compensation technology. [0003] Due to the difference in motion of different areas in the image, exposed areas and occluded areas are formed. These areas only exist in one frame of two adjacent frames. Using the traditional motion estimation method based on two frames before and after, may cause exposed / occluded areas. The motion vector estimation error and the motion compensation error of , appear halo effect around the object, which is called halo. [0004] The existing occlusion detection methods mainly use two motion vector fields obtained by...

AI Technical Summary

Problems solved by technology

Although these methods can detect most of the exposed / occluded areas, since these methods are block-based methods, at the junction of the exposed / occluded area and the background, there may be a small proportion of the exposed / occluded area in the junction block , causing the block to be detected as a normal area (not an exposed / occluded area), resulting in a circle of halo at the junction of the exposed / occluded and background after motion compensation

Even if post-processing and filtering are used, the above-mentioned halo effect can only be reduced, but not completely removed.

Images