To solve the problem of inefficiency of implementing embedded block coding in JPEG 2000,a novel fast algorithm to achieve word-level sequential and parallel bit plane coding (BPC) was proposed.The proposed BPC algorithm adopted a technique of performing the coding pass prediction and context formation as parallel and pipelined fashion.By using this BPC algorithm,the coefficients bit modeling for all bit planes in a code block were completed as parallel and sequential mode by one scan.Based on this new coding algorithm,an efficient high-speed architecture (HSA) was presented,which could perform context formation for 4 coefficients belonging to the same stripe column at one intra- clock cycle,and achieve context formation for an N×N code-block in approximate N2/4 intra- clock cycles.Theoretical analysis and experimental results demonstrate that,the proposed HSA could significantly increase throughput rate with good performance in terms of speedup to cost,compared with the up-to-date design.