Some Adaptive Quantizers for HDTV Image Compression

This paper proposes scene adaptive quantizers for image sequence compression. The proposed quantizers incorporate properties of the human visual system as well as channel buffer status in the decision of quantizer step size. Both spatial domain and DCT domain analyses are performed to determine the activity of the macroblock. The quantizer step size is adjusted according to the activity of the macroblock. Performance of the proposed scene adaptive quantizers is demonstrated by computer simulation using an HDTV test sequence. Subjective tests show that picture quality is substantially improved with the proposed quantizers. In particular, block artifacts are considerably reduced with the proposed methods, compared to those with the conventional, buffer-controlled quantizer.