A novel saliency based image compression algorithm using low complexity block truncation coding

Image features are better captured by saliency values, which can be deployed for subsequent high level processing at a reduced computational complexity. In this paper, we proposed a novel algorithm that integrates saliency and a low complexity block truncation coding (LCBTC) in a single framework. The proposed LCBTC framework has great potential in low power hardware implementation. The two phases of the algorithm are computation of the saliency strength of image blocks based on the probability of the pixels followed by a low complex Block Truncation coding method.The blocks having high saliency value are encoded using LCBTC and blocks having low saliency value are encoded using mean value of the blocks, thereby increasing the computational efficiency than the traditional BTC method. The efficacy of the proposed algorithm is evaluated based on objective fidelity criteria considering SSIM, QSSIM, FSIM, PSNR and bpp as well as subjective evaluation. The proposed method outperformed recent and baseline BTC methods in terms of objective and subjective measures. Proposed method shows significant improvements in performance over traditional BTC and recent approaches at lower bpp. It achieved an average PSNR of 33.03 dB and an average FSIM of 0.92,QSSIM of 0.91 and SSIM of 0.90 at a bpp of 1.65 and better perceptual quality with lower visual artifacts.