A Coding Scheme for Images with Text and Photograph

Abstract

The standard encoding scheme JPEG shows an excellent performance for still pictures in terms of coding efficiency. In the JPEG system, however, higher frequency components are not used to increase compression efficiency. Therefore, the reconstructed picture quality for pictures including edges such as characters is visually not good. In this paper, a new encoding scheme for high definition images with text and photograph is presented. The proposed algorithm consists of the following two processes : (1) Character segmentation (2) Adaptive encoding for both character regions and back ground region. Computer simulation results using several pictures shows 37.6 dB at 1.01 bits/pel on average while 26 dB at 1.0 bits/pel in the performance of JPEG. From the experimental results, the prospects of using this method can be considered.