Abstract: A video coding apparatus ensuring a high coding efficiency even at a low bit rate includes a moving object analyzer which extracts a moving part from an input picture signal, analyzes its motion, and outputs a residual signal relative to a reconstruction image and motion parameters. The apparatus further includes a residual coding device for coding the residual signal, a reconstruction device for reconstructing a picture using the motion parameters, and a device that performs a variable length coding of the motion parameters and residual coded information.

Abstract: An input image signal is divided into blocks each of 8 x 8 pixels, is subjected to cosine transform and is then converted to a signal of a spatial frequency region. After the cosine transform, the data is subjected to subtraction with respect to first to fourth types of data. Data of the same block of the one previous frame is used as the first data to provide a first difference between frames. Data of different blocks in the same frame is used as the second data to provide a second difference within the same frame. The third data is background image data and the fourth data is specific data. The minimum one of the first to fourth difference data is selected, and is then quantized and transmitted. Accordingly, a difference data involving a small amount of information to be transmitted can be selected for each block, thus always realizing a high transmission efficiency irrespective of the degree of image change.