Implementation of ROI Based Baseline Sequential Adaptive Quantization

The JPEG[1] (Joint Photographic Experts Group) introduced a still image graphics format in 1987, known as JPEG. The process flow in the baseline mode, converts color images to 8x8 block based DCT mostly applied for gray images, then we applied quantization for the DCT in zigzag ordering, and obtained entropy coding using Huffman tables in the encoding process is obtained and vice versa for decoding process, provided flexibility of different quantization matrices are used for luminance and chrominance components. However, in the JPEG standard, the Q matrix must remain the same across a color plane, i.e., spatially-variable quantization is not allowed. This prevents the encoder from, for example, varying the quantization within an image to achieve a desired output file size, or quantizing subject and background separately to permit region-of-interest (ROI) coding. New versions of JPEG including JPEG-LS, JPEG-2000 poised to continue the success that JPEG has achieved along with the features of ROI. However they not yet widely adopted. Hence, we implement the feature in baseline JPEG, so that it can have significant impact, even though JPEG is an old standard. Keywords—Entropy coding, Huffman coding, JPEG, JPEG-2000, Quantization, ROI.