COMPRESSION OF COLOR IMAGES USING QMBTC AND DSQ

Abstract

The major objective of compressing the image is to represent the image with as few bits as possible while preserving the quality and intelligibility needed for a particular application. In the conventional compression schemes for monochrome images, the gray scale values are considered whereas in color images, the three primary color components red, green and blue or any transformed components are considered for compression. A color image can be viewed as three monochrome images each corresponding to the three primary color components red, green and blue. Block Truncation Coding (BTC) is an effective tool for image compression. A color BTC algorithm called Quaternion Moment Block Truncation Coding (QMBTC) is discussed in detail. This color BTC algorithm can adaptively truncate a pixel block into one (or) two output classes depending on the distribution of color. values inside the pixel block. This results in an increase in compression ratio than that achieved by previous color BTC algorithms. A post BTC compression scheme is used to further compress the color image. The post BTC compression scheme comprises of two parts — block merging and color quantization. The dependent scalar quantization is slightly modified in order to reduce the computational complexity. The advantage of the image compression scheme is that the decoding is fast. Therefore, this compression scheme is suitable for transmission of color images. Using a lookup table to display decoded data, this image compression scheme can output images acceptable to human eyes. Several color images have been compressed and results are presented. The work is done on Indy Workstation in C language.