Research On High-speed Binary Arithmetic Coder Based On JPEG2000

JPEG2000compression standard adopts context-adaptive binary arithmetic codingfor entropy coding.Arithmetic coding has excellent coding performance.In mostcases,its average code length can be close to the source’ entropy. However, it has highcomputational complexity and includes a large number of bit-level operations andcomplex control logics.Therefore, how to improve the arithmetic coder’s throughput isthe bottleneck of design and implementation.Firstly,this paper studies the principle of binary arithmetic coding depthly.Then, weanalyze the coding process of MQ coder of JPEG2000, and describe key technologiesapplied to MQ coder.These technologies include probability estimates based on statetransition,bit padding and spacing mechanism,and the interval update withoutmultiplication.Based on the above, we design high-speed dual-symbolic processing MQ encoder.First,the double-symbol processing has two kinds of context case,as the same contextsand the different contexts.For each case, the encoding process is broken down into fourbranches: handling double most probability symbols (MPSMPS), dealing with two leastprobability symbols (LPSLPS), then a most probability symbol followed by a leastprobability symbol (MPSLPS) and a least probability symbol connected with a mostprobability Symbol (LPSMPS).Therefore,data dependency between symbols areseparated.Concurrent encoding is possible.Then,we predict that the division of theinterval range have25species.The interval update process is described in detail.Finally,we summarize the stream output situation in the case of two-symbol processing.In this paper,single and double symbol MQ coder is implemented by program Crespectively.We use sample sentence context for coding and decoding,then check theresults of two kinds coder and compare the data before encoding with the data afterdecoding.Coder design is verified.Then,this paper presents a four lines dual symbolMQ encoder’s VLSI structure.Finally,we optimize the probability estimation table,andits size is significantly reduced.