PREDICTIVE CODING SYSTEM FOR PICTURE SIGNAL

摘要

PURPOSE:To improve accuracy for a signal by estimating a correlative coefficient from a real forecast error signal to which inverse quantization is executed by an encoder or a decoder, updating a forecast coefficient by the estimated correlative coefficient and adaptively controlling the forecast coefficient by the statistical character of an input picture signal. CONSTITUTION:A normalizing forecast error signal YN is quantized encoded, outputted by a quantizer 44 and applied to a local inverse quantizer 46 and the inverse processing of the quantizer 44 is executed to a quantizing and encoding signal I. An inverse normalizer 48 executes inverse normalization by multiplying standard deviation to a forecast error signal Y'N. Normalizing forecast error power is inputted to a forecasting error power estimating device 61 and a corrected forecasting error power ratio eta' is inputted to a correlative coefficient estimating device 62 so as to calculate deviation amount rho from an estimated correlative coefficient (rho). Afterwards, while being adaptive to the real signal, the correlative coefficient rhon calculated by a correlative coefficient estimating device 62 is converted to a forecasting coefficient alphai by a forecasting coefficient estimating device 63 and applied to a forecast computing element 51.