Transcoding Jpeg

Melvin Berena BTC640 February 1, 2012 Abstract There is a transcoding scheme that proposed to compresses existing JPEG files without losing the quality. It does it by using the H. 264-like block-adaptive intra prediction to use inter-block relationship of quantized DCT coefficients that are stored in the JPEG file. This prediction is performed in spatial domain of each block composed of 8? 8 pels, but the corresponding prediction residuals are calculated in DCT domain to ensure lossless reconstruction of the original coef? cients. Block-based classification is carried out to allow accurate modeling of PDF’s.
A multi-symbol arithmetic coder and the PDF model is used for entropy coding to predicting residual of each DCT coefficient located in the in JPEG files. For monochrome JPEG images end result usually shows the reduction of coding rates about 18 – 28 %. Transcoding [Scheme] Scheme is a digital-to-digital data conversion method of converting to another type/format, like movie or audio files. This is frequently done when a particular target device does not support the file format or has limited storage capacity that requires a reduced file size.
This is also done to convert incompatible or out dated data to a better and modern format. One good advantage of transcoding is to reduce storage cost. Example, Cineon and DPX are widely used as a standard movie format but it cost a lot of space (2hour movie = about 8 terabytes). It would be demand more storage and processing power to handle this this kind of file format. However, transcoding into a JPEG2000 lossless format has better compression performance than other lossless coding technologies. The Challenge The number one challenge I faced was when I approached this academic paper review and read the abstract.

I did not understand a lot of information printed in this paper but as I kept reading the whole article I started getting a basic idea of what the paper is talking about. I took me a lot of time to just read the Abstract as I was using the search engine (Google) to understand every other word. The second problem I faced was when I approached the Spatial Domain, I still don’t completely understand that concept but I get the basic understanding how the object ends up with the blurriness effect as the object enters a repetitive intensity variation. References
Academic Paper review: http://zenit. senecac. on. ca/wiki/imgs/Lossless_re-enconfig_of_jpeg_images. pdf Websites: http://en. wikipedia. org/wiki/JPEG#Block_splitting http://en. wikipedia. org/wiki/Transcoding http://en. wikipedia. org/wiki/Point_spread_function http://en. wikipedia. org/wiki/Discrete_cosine_transform http://www. eurasip. org/Proceedings/Eusipco/Eusipco2011/papers/1569428105. pdf (Coding rates) http://en. wikipedia. org/wiki/Entropy_encoding http://research. microsoft. com/en-us/people/fengwu/screen_icme_06. pdf (Block-based Classification)