Comparison of Quantisation Techniques for DCT-based Image Coding
Stephen So, Kuldip K. Paliwal · 2001
In this paper, preliminary results of applying different types of scalar quantisers and vector quantisers to lossy image coding are pre- sented. The advantages of these different methods become evident when used in conjunction with a standard DCT-based image coder. It is shown that there is negligible improvement in applying vector quantisation to- gether with the DCT. This comparative study between different quantisa- tion techniques constitutes the first six months of research work and will provide the basis for further research into improving the efficiency of the DCT-based image coder. I. INTRODUCTION HE DCT or discrete cosine transform is the predominant transform used in transform image coding. It is used in im- age and video coding standards such as JPEG and MPEG. The information loss in the DCT-based coder mostly occurs in the quantisation stage and improvements in rate-distortion perfor- mance can be realised by designing more efficient quantisers. It is the aim of this paper to present a comparison of differ- ent quantisation techniques used in conjunction of a DCT-based coding system.