Hybrid Representative Coding Scheme for AMBTC Compressed Images

Abstract

Among the representative image compression techniques, absolute moment block truncation coding (AMBTC) offers acceptable image quality for the reconstructed image with a relatively low computational cost. Nevertheless, this approach has its constraints. In instances where an image contains numerous objects, the quality of reconstructed images or compression performance may be suffered. In order to solve this problem, we applied the idea of Huffman coding and proposed an adaptive image compression method with hybrid representative coding based on absolute moment block truncation coding (HRC-AMBTC). To achieve our object, the image blocks in an image are classified into three categories: flat, smooth, and complex. To maintain the image quality of the restored image while providing good compression performance, our designed representative coding is applied to the smooth block. Moreover, complex blocks are encoded by three quantized values, and their bitmaps are derived from our proposed difference clustering. In addition, to further improve HRC-AMBTC, we compress codes of the flat block, which bit per pixel is lower than our original HRC-AMBTC. Finally, the experiments confirm the effectiveness and reliability of the proposed compression methods after comparing them with other variants of the AMBTC method.