Abstract: A method and the system for bit-rate compression of digital data between a television transmitter and receiver employs at the transmitter a differential coding device comprising a code allocator coupled respectively to the outputs of predictors which deliver the spatial prediction errors of the contour areas or highly textured areas of the image, the spatial prediction errors of the uniform areas and the temporal prediction errors of the uniform and fixed areas. The code allocator is controlled by a selecting device which selects one output of the predictors for applying a prediction error value to the input of a coding structure of the code allocator in order to code each datum of a current point applied to the input of a predictor in three different fixed-length codes, depending on whether the current point is located in a contour area or highly textured area of the image or whether the current point is located in a fixed or moving uniform area.