Adaptive Directional Lifting Based Wavelet Transform And Applications

With the widely spread of computer, fast development of internet and emergence of all kinds of digital equipments, digital images have been used in many applications. How to represent and compress digital images efficiently becomes a hot research topic. Researchers have been developing image compression schemes with better coding performance and more features. The wavelet based state-of-art image compression standard JPEG2000 achieves much better coding performance than DCT based scheme JPEG. JPEG2000 also offers much functionality such as spatial scalability, SNR scalability, ROI and so on. How to improve the coding performance of wavelet becomes a hot research topic during the past few years.We first present a novel two-dimensional wavelet transform scheme of adaptive directional lifting (ADL) in image coding. Instead of alternately applying horizontal and vertical lifting as in present practice, ADL performs lifting-based prediction in local windows in the direction of high pixel correlation. Hence, it adapts far better to the image orientation features in local windows. The ADL transform is achieved by existing one-dimensional wavelets and is seamlessly integrated into the global wavelet transform. The predicting and updating signals of ADL can be derived even at the fractional pixel precision level to achieve high directional resolution, while still maintaining perfect reconstruc-tion. To enhance the ADL performance, a rate-distortion optimized directional segmentation scheme is also proposed to form and code a hierarchical image partition adapting to local features. Experimental results show that the proposed ADL-based image coding technique outperforms JPEG 2000 in both PSNR and visual quality, with the improvement up to 2.0 dB on images with rich orientation features.Directional transforms have been studied for many years. However, their performances on image coding are seldom analyzed and compared. In this paper, we choose three representative transforms: Contourlet, directional filter banks (DFB) and adaptive directional lifting (ADL) to compare their performances for image coding in terms of both theoretic analyses and numeric experiments. For a uniform evaluation framework, we adopt a 2D anisotropic image model and compare the coding gain of different transforms on the model. The analyses show that ADL generally performs the best amongst the compared transforms. And directional transforms have better performance than traditional wavelet on images with rich diagonal texture.We also develop a directional lifting-based 3D wavelet transform for volumetric medical datasets. The proposed wavelet transform exploits orientation features of volumetric datasets by performing the transform along the direction of strong correlation in local area. When coupled with 3d-EBCOT, the proposed scheme can generate embedded bit stream and present decoded quality from lossy to lossless. The experimental results on various medical volumetric datasets show that the introduction of directional 3d wavelet transform improves the image quality in terms of both visual quality and PSNR and visual quality with up to 2db gain compared with traditional 3d wavelet. The ring effects artifacts around edges are also reduced significantly.