Conditional Vertical Subsampling-A Technique to Assist in the Coding of Television Signals

In an interlaced scan television system, the vertical sampling rate of an image can be halved by sampling every other field. Picture elements in the missing fields are replaced in the display by both temporal and vertical interpolation, but the resulting pictures show some visible defects. This paper describes how these defects can be eliminated at the extra cost of fully sampling in selected areas of the picture. For a typical Picturephone® scene with active movement the selected areas make up about 6 percent of the picture elements in the unsampled field. The technique can be combined with a wide variety of interframe-coding techniques. In one particular example in which the television signal is specified as clusters of frame-to-frame differences, the cost of specifying “active” frames (14,000 significant frame differences per frame) is reduced from 68,000 bits to 42,500 bits. This corresponds to a reduction in bit rate from 2 Mbits sec −1 to 1.3 Mbits sec −1 .