A jpeg2000 encoding method for real-time truncated code stream and its implementation system

Abstract

The invention discloses a JPEG2000 encoding method for real-time truncated code streams. According to the importance of the wavelet sub-band and the number of channels of the wavelet sub-band, each wavelet sub-band is assigned a certain number of bytes; inside the wavelet sub-band, according to each wavelet sub-band The number of channels of a code block, and allocate the corresponding number of bytes to each code block; during the encoding process, when the accumulated encoded bytes of each code block is greater than the number of pre-allocated bytes of the code block or the encoding of the code block is completed, the current code block is stopped. Block coding, calculate the rate-distortion slope of each channel, and accumulate the total compressed code stream; after each wavelet sub-band is encoded, update the number of allocated bytes of the wavelet sub-band, when the accumulated total compressed code stream is greater than the target bit stream , if there are uncoded wavelet subbands, change the allocation strategy, re-allocate a certain number of bytes for uncoded wavelet subbands, continue coding until all code blocks are coded, use the minimum slope discarding method to find the best threshold, and complete the coding. The real-time performance of the encoding system is improved, and unnecessary storage and calculation consumption are reduced.

Description

technical field [0001] The invention relates to the technical field of VLSI design, in particular to a JPEG2000 encoding method for real-time truncated code streams, and also relates to a system for implementing the method. Background technique [0002] JPEG2000 first preprocesses the original image, then performs wavelet transformation and quantization on the image, and the quantized wavelet coefficients enter the EBCOT encoding stage. EBCOT coding includes Tier1 coding and Tier2 coding. Tier1 encoding performs bit-plane encoding and MQ arithmetic encoding on the quantized wavelet coefficients according to the code block, and then uses the post-compression rate-distortion optimization algorithm PCRD to obtain the optimal truncation threshold, discards the compressed codewords that are smaller than the threshold, and discards the compressed codewords that are larger than the threshold. The compressed codeword is encoded by Tier2 to form a compressed code stream. [0003] T...

AI Technical Summary

Problems solved by technology

[0004] The PCRD algorithm can optimally allocate the code rate while meeting the target code rate, so as to minimize the total distortion. It realizes the theoretically optimal rate-distortion optimization allocation, but the optimal layering can only be performed after the entire Tier1 encoding is completed. cut off

The Tier1 encoding part accounts for 45% to 60% of the entire encoding time, and when the bit rate is low, most of the encoded bit streams will be discarded, which greatly increases the unnecessary calculation and storage loss of the system, making the system real-time decreased sex

Images