Conditional-entropy-constrained trellis-searched vector quantization for image compression

摘要

This paper proposes conditional-entropy-constrained trellis-searched vector quantization for image compression. We introduce (i) the conditional-entropy index encoder to exploit interblock correlation and (ii) the trellis-searched encoding to get a long-term optimum. Simulations show that the proposed VQ provides higher PSNR by 3–5 dB, when compared to ECVQ.