An Image Coding Algorithm Based on Nonredundant Contourlet Transform and Visual Characteristics

This paper presents an image coding technique based on contourlet transform. The redundancy of the contourlet transform is eliminated by switching the LP decomposition step to separable wavelet transform. The transform is then optimized through a wavelet coefficient weighting process in order to enhance the visual quality in the image domain, and with that the directional filter bank(DFB) are employed to high frequency wavelet subbands. At last, all the resulted subbands are efficiently coded with EBCOT. At the same bit rates, a gain of up to 0.3~0.6 dB over JPEG2000 has been observed for standard images with directional features.