Human Visual System Based Enhanced AMBTC for Color Image Compression Using Interpolation

Abstract

This paper proposes a novel enhanced AMBTC image compression method for digital color images which uses the characteristics of human visual system. The proposed compression method first transforms the RGB image into YCbCr color space and then divides each component of the image into non-overlapping blocks. To improve the imperceptibility of caused distortion due to AMBTC compression, the proposed method computes four quantization levels and 32 bits bit-planes for luminance component (Y) blocks whereas the blocks of the remaining two components (i.e., Cb & Cr) are represented by two quantization levels and only 8 bits bit-planes. At the receiving end, the receiver can use interpolation technique to predict the remaining 8 bits of the bit-planes of Cb and Cr components. Thus, the proposed method provides better quality stego-image than the related existing methods, Further, the required bit-rate is also reduced for all the test images irrespective of their characteristics.