Patent US 5,485,210

Television images are decoded by forming a predicted current image frame, receiving and decoding a quantized subband-represented residual frame, and reconstructing a current frame by combining the decoded residual frame and a fraction of the predicted current image frame in a plurality of frequency subbands. A television signal also may be decoded to determine the state of an indicator that specifies for an entire received frame whether the received frame is an actual image frame. If the indicator specifies that the frame is an actual image frame, the frame is decoded to reconstruct the actual image and the decoded frame is used without motion compensation. If the indicator specifies that the frame is not an actual image frame, motion compensation is used to reconstruct the entire frame.