Adaptive quantizer, adaptive quantization method and adaptive quantization program

摘要

Activities of a video signal are obtained for subblocks of each macroblock of each picture. The smallest activity is detected as an input-picture activity per macroblock. Obtained are activities of a motion-compensated predictive signal per first picture and activites of the video signal per second picture, for the subblocks of each macroblock. A mean value of the activities is obtained per macroblock for each picture, as an error activity per macroblock. A mean error activity is obtained for error activities per picture. A quantization step size for quantization of the video signal is adjusted per macroblock, according to the input-picture activity, to obtain a smaller step size when the error activity is equal to or larger than the mean error activity than when the former is smaller than the latter.