Image coding for monochrome and colour images.
Date of Award
Electrical and Computer Engineering
Sid-Ahmed, M. A.,
Engineering, Electronics and Electrical.
CC BY-NC-ND 4.0
This work is an investigation of different algorithms to implement a lossy compression scheme. Special emphasis was focused on the quantization techniques. A CODEC (coder/decoder) based on a scheme proposed for standardization by a group known as JPEG (Joint Photographic Experts Group) was developed. Finally, a new decoding approach was developed, based on modifying the concepts of transition table used in compilers to break a binary string into variable length codes. The JPEG algorithm works in sequential mode by dividing the image into small blocks of 8 x 8 pixels. Each block is compressed separately by processing it through an 8 x 8 Discrete Cosine Transform, Quantization, Run length and Huffman coding. The two dimensional DCT was implemented by a fast 1-D DCT expanded into a 2-D DCT, using the row-column method. Quantization is carried out by dividing the transformed block by the "JPEG scaling matrix" and rounding the results to the nearest integer. It was found to work well for a large number of images. Four static Huffman code tables are used to convert the quantized DCT coefficients into variable length codes for both monochrome and colour images. The algorithm is capable of obtaining varying compression ratios by simply changing the scaling factor of the "JPEG Quantization Matrix". The bit rate achieved was in the range of 1 bit/pixel for images indistinguishable from the original. Higher compression ratios can be obtained at the cost of lower image quality.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1993 .S465. Source: Masters Abstracts International, Volume: 32-02, page: 0691. Adviser: M. A. Sid-Ahmed. Thesis (M.A.Sc.)--University of Windsor (Canada), 1993.
Shlimon, Napiluon Petrus., "Image coding for monochrome and colour images." (1993). Electronic Theses and Dissertations. 2377.