Rate-distortion analysis of variable block size VQ-based motion compensated video codecs

Low bit-rate video compression systems using motion compensation and vector quantization (VQ) have been widely studied. This coding scheme first reduce the temporal correlation through motion prediction then the motion compensated difference images are compressed using vector quantization (VQ). To improve the coding performance, variable block size VQ schemes have been developed. In this approach, the subimage blocks are segmented into homogeneous subregions and vector quantized independently. This form of VQ can be perceived as a component vector quantizer where a signal is decomposed into subsignal or component and individually compressed. The superior performance of this VQ scheme over fixed block VQ has been observed empirically. In this paper, rate-distortion analysis is carried out to provide some insights to the underlying coding gain achieved by this class of VQ for motion compensated video compression systems