An adaptive block truncation coding algorithm for image compression

Abstract

A method of designing a minimum mean square error (MMSE) quantizer for the block truncation coding (BTC) algorithm for image compression is presented. Based on this MMSE quantizer, an adaptive block truncation coding (ABTC) algorithm is designed. This method tries to optimize the quantization output levels based on the locality of the image pixel blocks, and significantly improved reconstructed image quality is achieved at 2 b/pixel. The performance of the ABTC MMSE quantizers is investigated. It is compared to the standard BTC method and the absolute moment BTC (AMBTC) method. The ABTC method is very efficient as compared to the standard BTC and the AMBTC methods.< >