| 摘要 |
A motion estimator/compensator determines whether to use half-pixel motion vector encoding by comparing the relative benefit of using half-pixel encoding, represented in terms of distortion between the macroblock as encoded and prior to encoding, to the processing burden imposed upon the client computer system in decoding the half-pixel encoded macroblock. Specifically, the motion estimator/compensator quantifies distortions in encoding a subject macroblock as a motion vector to (i) a whole pixel pseudo-macroblock, (ii) a half-column pixel pseudo-macroblock, (iii) a half-row pixel pseudo-macroblock, or (iv) a half-column/half-row pixel pseudo-macroblock. The type of motion vector encoding having the smallest combination of distortion and decoder processing burden is selected and used to encode the subject macroblock. Thus, each type of half-pixel motion estimation is evaluated individually and used for encoding the subject macroblock only if the benefit of the type of half-pixel motion estimation justifies the corresponding additional processing burden imposed upon a client computer system.
|
