| 摘要 |
A composite video signal is separated into luminance and chrominance components by forming at least two estimates of the value of each component, based upon at least two sets of samples of the signal, and by selecting the best estimate based upon local picture characteristics. In one embodiment, the first estimate is derived from samples associated with picture elements which lie along the same scan line as the element then being processed, while the second estimate is derived from samples associated with picture elements in different lines above and below the present line. The best estimate selection is based on the relationship between each estimate and the component values of neighboring picture elements. If desired, delayed separation can be used to determine the effect of each estimate on future separations, before a selection is made. |
