An adaptive motion-compensated approach for video deinterlacing

摘要

Deinterlacing, defined as the process of converting a stream of interlaced frames into a sequence of progressive frames, represents a key feature in video processing. The interlaced video format, introduced by the old analog television transmission systems as a trade-off between framerate and bandwidth capacity, has become obsolete nowadays, when all transmissions are digital. Moreover, almost all recent displays-whether LCD or plasma-require progressive video input, whereas much of the available video content is in interlaced format. In this paper an adaptive, edge-preserving motion-compensated approach for video deinterlacing is proposed. The algorithm preserves strong edges and interpolates the missing pixels along the contours depending on the motion-degree of the region to which they belong. Our proposal is optimized to lower heavy computation, which is the main drawback of motion-compensated deinterlacing algorithms. Therefore it provides complexity scalability as a trade-off tool between performance and computation time. Experiments demonstrate a significant gain in reconstruction quality as compared to other deinterlacing implementations.