一种时空权重和边缘自适应的扫描格式转换

A Video Scan Format Conversation Algorithm Based on Spatio-temporal Weight and Edge Direction

为实现隔行扫描到逐行扫描的视频扫描格式转换,提出了一种时空权重和边缘自适应的去隔行算法,主要包括运动估计、小角度边缘搜索、时空权重计算、自适应插值等。该算法通过4场相邻像素的最大灰度变化与运动阈值比较实现对当前像素的运动估计,采用自适应搜索半径和并行搜索树的方法实现小角度边缘检测,并且增加对半像素边缘的考虑实现最小6°边缘的检测,最后通过时空权重自适应的插值算法实现去隔行处理,取得很好的处理效果。

Deinterlace is to convert interlaced images to progressive ones. In this paper, a deinterlacing algorithm based on spatio-temporal weight and edge direction is presented. It consists of motion detection, low- angle edge detection, and spatio-temporal weight adaptive interpolation. It uses total 4 fields information captured in field storage to detect the presence of motion. And which the 6°edge is obtained. The proposed method good display quality and satisfies real-time process of it detects low-angle edge by an adaptive searching radius in using spatio-temporal adaptive interpolation can obtain quite HDTV sequences.