Abstract: Input MPEG coded image data is input through a buffer to a variable length decoding unit where it is subject to variable length decoding. The quantized DCT coefficients are output to an inverse quantization unit, while the motion vectors are output to a motion compensation prediction unit. The quantized DCT coefficients are inversely quantized at the inverse quantization unit to generate DCT coefficients. The DCT coefficients have values of the items of frequencies higher than the desired frequency limit converted to 0 at the DCT coefficient conversion unit whereby the motion data found at the format conversion unit is added at the adder and the result input as the original image data to the format conversion unit. The pixels are thinned at the format conversion unit to generate a compressed image which is then displayed on a display unit.