Recursive Estimation of Local Characteristics of Edges in TV Pictures as Applied to ADPCM Coding

In this paper, an algorithm is presented for extracting edges and estimating their local characteristics in TV pictures; the algorithm is designed for application to adaptive DPCM coding of TV pictures and it has the following properties: 1) it takes into account the causality constraint encountered in DPCM coding; 2) it is fitted to the line-by-line scanning of TV pictures; 3) it is fast, involving only a small amount of computation, and thus allowing real-time implementation. The whole processing is split into a local preprocessing and a global processing where the characteristics of the edges are renewed recursively line by line. This algorithm was used for implementing an adaptive choice of the prediction in DPCM coding of TV pictures.