Near-aperiodic dot-diffused block truncation coding
Yun-Fu Liu, Jing-Ming Guo, Zong-Jhe Wu, Hua Lee · 2015
In this study, an improved Block Truncation Coding (BTC) image compression scheme, namely Near-Aperiodic Dot-Diffused BTC (NADDBTC), is described. Firstly, the existing regular structures for the generation of bitmap are completely modified for aperiodic compressed results. Moreover, an adaptive quantization levels selection strategy and two parameters Class Matrix (CM) and Diffused Matrix (DM) for image compression are developed and co-optimized. The improvements produce results of superior image quality. Furthermore, the adaptive quantization levels are introduced for balanced false contour, impulsive noise, and blocking artifact. Experimental results demonstrate that the proposed NADDBTC is capable of providing excellent image quality and visual perception, as well as processing efficiency, similar to DDBTC by exploiting the innate parallelism advantage of dot diffusion.