Dynamic sub-range search methods for image colour quantisation

Abstract

This paper presents two image coding methods based on colour image quanitsation (CIQ). Multiple colour palttes of different sizes are required to provide different numbers of bit rates in CIQ. To provide the adjustable number of bit rates using a colour paltte, the high degree of similarity among neighbouring colour pixels is exploited in the first method. Besides, the similarity among the neighbouring encoded indices is expoited in the second method. The search range of each colour pixel to be compressed is dynamiclly determined in both two methods. Experimental results show that the proposed methods provide better image qualities at different numbers of bit rates compared to other CIQ when suitable threshold values are set. Meanwhile, the proposed methods consume little computational cost.

Keywords:

• colour image quantisation
• palette design
• bit rate reduction

This research was supported by Providence University, Taichung, Taiwan, under contract PU 97-B21 and the National Science Council, Taipei, Taiwan, under contract NSC 98-2221-E-126-008.