Vector quantization (VQ) has recently emerged as a powerful and efficient technique for digital speech and image coding. The goal of such a process is data compression: to minimize communication channel capacity or digital storage memory requirements while maintaining an acceptable fidelity level of the data. A review of various VQ algorithms and their respective design considerations as applied to color images is given. Fidelity measurements and signal-to-noise ratio calculations are discussed. A modified mean-residual vector quantizer using the LBG design algorithm with color signal preprocessing is described. The algorithm is developed to yield a bit rate of 0.709 bits per pixel per color with the goal of easy implementation even using a simple microcomputer . Photographic and numeric results of original versus compressed-uncompressed color images are presented. Several modifications to the described algorithm are tested with good results .
Electrical Engineering (MS)
Department, Program, or Center
Electrical Engineering (KGCOE)
Kristy, Stephen H., "Vector Quantization of True-Color Images" (1986). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus