Discrete Walsh-Hadamard transform vector quantization for motion-compensated frame difference signal coding

Theoretical performance analysis and test simulations showed that the DWHT (discrete Walsh-Hadamard transform) is as good as the DCT (discrete cosine transform) for the encoding of low-correlation MCFD (motion-compensated frame difference) signals. However, the straightforward algorithm for the DWHT makes it a better choice than the DCT. The transform vector quantization (TRVQ) technique is employed for low bit rates. It is also observed that the robustness of the VQ (vector quantization) technique for the MCFD signals is much better than that of the VQ of the still frame images.<
>