An embedded wavelet coder algorithm which adopts different threshold values in different subbands is proposed in this work. The new image coder, called the multi-threshold wavelet coder (MTWC), achieves a better rate-distortion performance than existing embedded wavelet coders for applications requiring higher quality with higher bit rates (e.g. greater than 1 bpp). Another advantage of MTWC is its lower computational complexity. The superior performance of SITWC can be explained by examining the evolution of the distortion distribution in each subband as image data are compressed in the embedded coding process. Experimental results are given to demonstrate the performance of MTWC at various bit rates. threshold, wavelet coding.
H.-J. Wang, C.-C. Kuo