Low GPU occupancy approach to fast arithmetic coding in JPEG2000
MEMICS'11 Proceedings of the 7th international conference on Mathematical and Engineering Methods in Computer Science
Two-Symbol FPGA Architecture for Fast Arithmetic Encoding in JPEG 2000
Journal of Signal Processing Systems
| Hi-index | 0.00 |
JPEG2000 is the latest still-image coding standard. It was designed to overcome the limitation of the original JPEG standard and provide high quality images at low bit rates. The JPEG2000 algorithm is fundamentally based on the Discrete Wavelet Transform (DWT) and Embedded Block Coding with Optimal Truncation (EBCOT). Both algorithms are computation- and memory- intensive. JPEG2000 uses an adapted version of Q-coder, called MQ-coder. Pervious researches mainly focused on speeding up the DWT and BPC other than renormalization part of MQ- coder. Renormalization procedure in MQ-coder contributes to the computation in relatively high percentage. In this paper, we propose an enhanced renormalization algorithm for arithmetic coding in EBCOT Tier-1 coding procedure. By analyzing and simplifying the existing algorithm, the proposed algorithm can reduce the computation with a significant percentage. Experimental results show that the computation is reduced approximately 25% comparing to the renormalization procedure of the regular arithmetic coding.