Today there are several formal and experimental methods for image compression, some of which have grown to be incorporated into the Joint Photographers Experts Group (JPEG) standard. Of course, many compression algorithms are still used only for experimentation mainly due to various performance issues. Lack of speed while compressing or expanding an image, poor compression rate, and poor image quality after expansion are a few of the most popular reasons for skepticism about a particular compression algorithm. This paper discusses current methods used for image compression. It also gives a detailed explanation of the discrete cosine transform (DCT), used by JPEG, and the efforts that have recently been made to optimize related algorithms. Some interesting articles regarding possible compression enhancements will be noted, and in association with these methods a new implementation of a JPEG-like image coding algorithm will be outlined. This new technique involves adapting between one and sixteen quantization tables for a specific image using either a genetic algorithm (GA) or tabu search (TS) approach. First, a few schemes including pixel neighborhood and Kohonen self-organizing map (SOM) algorithms will be examined to find their effectiveness at classifying blocks of edge-detected image data. Next, the GA and TS algorithms will be tested to determine their effectiveness at finding the optimum quantization table(s) for a whole image. A comparison of the techniques utilized will be thoroughly explored.
Library of Congress Subject Headings
Computer algorithmsImage compression; JPEG (Image coding standard)
Department, Program, or Center
Computer Science (GCCIS)
Champion, Michael, "Genetic algorithm and tabu search approaches to quantization for DCT-based image compression" (1999). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus