A hybrid approach of wavelet packet and directional decomposition for image compression

In this paper, a novel image compression technique, the combination of wavelet packet transform and directional decomposition, is proposed. Wavelet packet transform is an increasingly remarkable approach which outperforms the standard wavelet transform in image coding. It can be employed to exploit the image redundancy efficiently and therefore obtain a high compression ratio. Directional filtering coding technique, as one of the "second generation image coding techniques", first introduced the concept of directional decomposition. By placing emphasis on edge detection to preserve edge information to exploit the fact that the human visual system is more sensitive to image edge features, it is generally capable to obtain a relatively high compression ratio. The approach proposed in this paper decomposed an image into a low frequency component and a number of high frequency components, with the edges on each high frequency component at its own directions. By a combination process of Cartesian coordinate rotation transform, interpolation, wavelet packet transform, and a coding algorithm, the image is reconstructed at an improved visual quality at the same bitrate compared to the common wavelet pyramid algorithm.