JOURNAL OF COMPUTERS, VOL. 1, NO. 4, JULY 2006 1 Backward Coding of Wavelet Trees with Fine-grained Bitrate Control Jiangling Guo School of Information Science and Technology, Beijing Institute of Technology at Zhuhai, Zhuhai, P.R. China Email:
[email protected] Sunanda Mitra, Brian Nutter and Tanja Karp Department of Electrical and Computer Engineering, Texas Tech University, Lubbock, USA Email: {sunanda.mitra, brian.nutter, tanja.karp}@ttu.edu Abstract—Backward Coding of Wavelet Trees (BCWT) is Recently, several new wavelet-tree-based codecs have an extremely fast wavelet-tree-based image coding algo- been developed, such as LTW [7] and Progres [8], which rithm. Utilizing a unique backward coding algorithm, drastically improved the computational efficiency in BCWT also provides a rich set of features such as resolu- terms of coding speed. In [9] we have presented our tion-scalability, extremely low memory usage, and extremely newly developed BCWT codec, which is the fastest low complexity. However, BCWT in its original form inher- its one drawback also existing in most non-bitplane codecs, wavelet-tree-based codec we have studied to date with namely coarse bitrate control. In this paper, two solutions the same compression performance as SPIHT. With its for improving the bitrate controllability of BCWT are pre- unique backward coding, the wavelet coefficients from sented. The first solution is based on dual minimum quanti- high frequency subbands to low frequency subbands, zation levels, allowing BCWT to achieve fine-grained bi- BCWT also provides a rich set of features such as low trates with quality-index as a controlling parameter; the memory usage, low complexity and resolution scalability, second solution is based on both dual minimum quantiza- usually lacking in other wavelet-tree-based codecs.