Optimum Bit Allocation for a Given Signal-to-Noise Ratio in Transform Coding of Images

Abstract

Transform coding is frequently used in the high-efficient coding of images. In transform coding, a bit allocation map is determined according to the statistic of images. Usually, the map is calculated to minimize the coding error for a given bit rate. However, this paper proposes an other scheme in which a bit allocation is determined to minimize the bit rate for a given coding error. In general, the usual method is suitable for producing devices when transmitting television images. However, the second method is effective in maintaining image quality, when transmitting many freeze frame images having different statistic and storing many images in the storage mass. An analytical formula was derived for the proposed method and good performance was obtained by computer simulation. A desired quality (SNR) of images can be easily obtained using this method.