| 摘要 |
A digital video signal is conducted to an edge detection block 37, which, operating in a plurality of directions, including the horizontal direction, recognizes for each sample of the field the direction of any edge present and records the same as a coded edge information value in the memory. In a quantification block 34, the sample values of the non-jump-line-scanned video signal are quantified. At the same time and separately from each other, two block matchings are used. In the edge information matching 36, instead of the actual samples of successive fields, the associated edge information value is used, whereupon a first motion vector candidate is obtained for each examined block. In the figure matching 35, instead of actual samples of successive non-jump-line-scanned fields, samples obtained therefrom by quantification 34 are used, whereupon a second motion vector candidate is obtained for each examined block. In a zone comprising the examined block and at least its directly adjoining block, for the vector candidates a reduction is carried out in the number of vector candidates, for example with the use of an MSE-criterion. The final motion vector is selected in the course of the further processing. <IMAGE>
|
