Signal Processing: Image Communication 2 (1990) 397-407 397 Elsevier MOTION VECTOR REPLENSIHMENT FOR LOW BIT-RATE VIDEO M. GHANBARI Department of Electronic Systems Engineering, University of Essex, Colchester C04 3SQ, England Received 15 March 1990 CODING Abstract. A low bit-rate video codec based on motion vector replenishment is described. Motion vectors are used to update pictures at full frame rate. In addition, part of each frame is conditionally updated with a strip of interframe video data. The video data fill the remaining channel capacity not used for motion vectors. Thus under most conditions, each frame is fully updated by motion vectors and partially with interframe video data. This method has a comparable compression efficiency with that of the frame dropping method, but does not introduce any picture 'jerkiness'. Finally the application of the proposed method to packet video networks is examined. Keywords. Motion compensation, low bit-rate coding, conditional replenishment, packet video. 1. Introduction Conditional replenishment (CR) is a key-tech- nique for low bit-rate video coding [1]. In this method only moving parts of successive television frames are coded and transmitted. At the receiver the non-transmitted parts are substituted from pre- viously received ones. The compression efficiency of this coding technique is improved if the motion between the successive frames (interframe) can be compensated. Here by estimating the interframe motion, only the differences between the current frame and the motion compensated previous one are coded. Various motion compensation methods have been introduced in the past decade [8]. The most common is the block matching technique. In this method, a moving block of picture elements (pels) from the current frame is compared against blocks of pels in the previous frame inside a search win- dow. An estimated motion (displacement) vector is the relative position of a block in the previous frame with the best match to the current one. The motion vectors, together with the video errors, are transmitted. The video errors are the difference signals between the current moving blocks and those in the previous frame displaced by the motion vectors. It has been reported that efficient motion com- pensation can result in a data reduction factor of 1.5 to 2 [7]. However, there are occasions when the interframe coding technique, even with motion compensation, is not sufficient to bring the data- rate down the channel capacity. Hence, since the data-rate should not exceed the channel-rate, the coder will have to skip several frames. During a skipped frame the receiver repeats the previously received one. Depending on the picture activity and channel-rate, the transmitted frame rate can be very low. For example for 64 kbit/s, the frame- rate can be as low as 1-2 per second. This intro- duces the undesirable motion-related artifact known as picture 'jerkiness'. This paper introduces a coding method that has a comparable compression efficiency to frame skip- ping but without introducing any jerkiness. 0923-5965/90/$03.50 © 1990 - Elsevier Science Publishers B.V.