An Improved Motion Adaptive Deinterlacing Method Using Variable Block-Size Motion Detection

This paper describes an improved motion adaptive deinterlacing method which uses five consecutive fields for motion detection. The motion detection algorithm, searches for motion activity between the fields by a hierarchical block based structure so it reduces the computation amount considerably. This reduction depends on the level of motion activity between the fields and could be of the order of two for slow motion scenes. The motion detection algorithm applies the same motion detection structure to all three color components of the video sequence and the motion detection results of different colors are then compared to obtain more reliable motion detection results. Searching between five fields and also using color data in motion detection makes the motion detection step more accurate and robust to noise and consequently it improves the overall performance of the deinterlacing algorithm. Missing lines of interlaced video are then interpolated using the motion activity data of the interlaced video fields; by giving more weight to temporal (or interfield) interpolation in static parts of the video scene and more weight to spatial (or intrafield) interpolation in dynamic parts of the video scene. Experimental results show that the proposed deinterlacing method gives less interpolation errors than most available deinterlacing methods.