Research On Modified Two-Dimensional DCT Coding And Decoding Technology In JPEG Area

JPEG is the most widely seen image storage format,due to its high compression ratio,JPEG is seen everywhere on the internet,the core of its standard is DCT algorithm,dividing the image into nonoverlapping 8×8 module,processing each independently and combine them in the last,thus JPEG will bring the boundary effect which is also called block effect.Modified Discrete Cosine Transform(MDCT)plays an important role in the area of signal processing.Compared with DCT algorithm,MDCT algorithm comes with higher computation complexity,due to its overlapping quality,it brings better effect in the area of signal processing.In recent years,one-dimensional MDCT algorithm is extensively used in MP3/AAC and other audio compression area.Multi-dimensional MDCT has better effect than Multi-dimensional DCT,and it can effectively reduce Artifacts stemming from the block boundaries.To reduce the boundary effect,we come up with the idea that replace the DCT algorithm in the JPEG standard with MDCT,along with producing new standards,here is the specific content:(1)Analyse the flow of JPEG image compression standard in detail,particularly quantization and entropy coding part.Decompose the process of how to realize image data compression in JPEG standard in detail.Test and explain the difficulty Hoffman coding steps experimentally.At the same time,test the compression performance of JPEG standard by actual experiments,and discuss the limitations of JPEG standard.(2)Analyse and compare the characteristic of MDCT and DCT algorithm.Then we put forward two different fast algorithm of MDCT:fast two-dimensional MDCT algorithm based on2×2 module?MDCT/IMDCT transformed into DCT-? to reduce computation complexity and increase software and hardware implementation possibilities.we'll explain these algorithms in detail and compare the algorithm complexity with traditional determinant algorithm.(3)Because of the overlapping output of MDCT algorithm is only 1/4 of the input,it is necessary to improve the quantization process and design a new quantization table,and propose an improved JPEG image compression and reconstruction method.Two kinds of experiments are carried out to test whether the modified image reconstruction method is correct in theory and experiment.Firstly,DCT algorithm and MDCT algorithm are used to compress and reconstruct gray images with different resolutions,and then the peak signal-to-noise ratio(PSNR)of the compressed reconstructed images and the running time of the compressed reconstructed programs are compared.Secondly,two kinds of compression and reconstruction experiments are carried out for gray and color images with different characteristics.Two experimental resultsshow that with the improvement of image resolution,the actual cost of MDCT algorithm is increasing,but the effect of compression and reconstruction is not improved.The more complex the image composition,the more details,regardless of color or gray image,using MDCT algorithm instead of DCT algorithm can improve the repetition effect.From the analysis of JPEG standard we come to know that block effect limits the development of JPEG and using MDCT algorithm instead of DCT algorithm requires corresponding improvements in the quantization and image reconstruction process of the standard.At the same time,fast algorithm must be used instead of determinant algorithm to reduce the complexity of the algorithm,verify the modified standard theoretically and practically.Finally,based on the actual image compression and reconstruction experiment,it is concluded that:MDCT replaces the DCT algorithm of JPEG standard improves the compression and reconstruction effect with different time cost at different resolutions.The more complex and detailed the composition,the higher the compression and reconstruction effect.