| 摘要 |
An interlaced image processing module receives interlaced first and second fields (355a) and produces a reference field and error field (360a). The reference field corresponds to the still image content, whereas the error field corresponds to the motion content between fields. Motion between fields is thus represented in the error field (E), without redundant representation of the still image content provided by the first field. Where there is little motion between fields, the error terms are small so the predictor (315a) preserves the coding efficiency provided by any auto-correlation in the image. The interlaced image processing method does not rely on pixel group classification, thus avoiding classification errors and the loss of coding efficiency from still image content in motion classified blocks. Problems of relative motion between fields and local artifacts are avoided. Another embodiment transforms the interlaced fields into frame data having high frequency and low frequency fields.
|
