Application Research On Contourlet In Image Processing

Multiscale geometric analysis is one of the key technologies and has drawn tremendous attention in the research of digital image processing, digital video processing, and computer vision. Contourlet transform (CT), one of the multiscale and multidirection analysis methods, is studied in this thesis. Aiming to investigate the theory and property of CT, we discuss the extensive applications on CT at the fields including digital image watermarking, digital image coding, digital image fusion, and deblocking for coded images. Making full use of the theory and technologies of digital image processing, we propose some novel methods and technologies based on CT, and improve the performance of CT in real application. The main contributions of this thesis are discussed in detail as follows:Firstly, in CT, Laplacian pyramid (LP) decomposes image into a low-frequency (LF) subband and a high-frequency (HF) subband. The LF subband is created by filtering the original image with a 2-D low-pass filter. However, the HF subband is created by subtracting the synthesized LF subband from original image but not by 2-D high-pass filtering original image. In this thesis, we propose a contourlet based image adaptive watermarking (CIAW) scheme, in which the watermark is embedded into the contourlet coefficients of the largest detail subbands of the image. The transform structure of LP makes the embedded watermark spread out into all subbands including LF subbands with great possibility when we reconstruct the watermarked image based on the watermarked contourlet coefficients. Since both LF subbands and HF subbands contain watermarking components, our watermarking scheme is expected to be robust against both LF image processing and HF image processing attacks. The corresponding watermarking detection algorithm is designed to distinguish whether the watermark is present or not by exploiting the unique transform structure of LP. With the new proposed concept of spread watermark, the watermark is detected by computing the correlation between the spread watermark and the watermarked image in all contourlet subbands fully. The proposed CIAW scheme is superior to conventional watermarking schemes thanks to the watermarking components preserved in LF subbands, particularly when the watermarked image is attacked by those image attacks methods which destroy the HF subbands. Experimental results demonstrate the validity of CIAW in terms of both watermarking invisibility and watermarking robustness.Secondly, CT is a new image representation method, which can efficiently represent contours and textures in images. However, CT is a kind of overcomplete transform with a redundancy factor 4/3. If it is applied to image compression straightforwardly, the encoding bit-rate may increase to meet a given distortion. This fact challenges the coding community to develop CT-based image compression techniques with satisfactory performance. In this thesis, we analyze the distribution of significant contourlet coefficients in different subbands and propose a new contourlet-based embedded image coding (CEIC) scheme. The well-known wavelet-based embedded image coding (WEIC) algorithms such as EZW, SPIHT and SPECK can be easily integrated into the proposed scheme by constructing a virtual low frequency subband, modifying the coding framework of WEIC algorithms according to the structure of contourlet coefficients, and adopting a high-efficient significant coefficient scanning scheme for CEIC scheme. The proposed CEIC scheme can provide an embedded bit-stream, which is desirable in heterogeneous networks. Our experiments demonstrate that the proposed scheme can achieve the better compression performance. Furthermore, thanks to the contourlet adopted in the proposed scheme, more contours and textures in the coded images are preserved to ensure the superior subjective quality.Thirdly, recent studies show that wavelet-based image fusion methods provide a high spectral quality in the fused satellite images. However, the fused images obtained by most wavelet-based methods have lower spatial resolution because of the critical down-sampling. In this thesis, we propose a useful fusion method based on CT and local average gradient (LAG) for multispectral and panchromatic satellite images. Since there are few reference available in this field, our approach helps to uncover the application of contourlet for image fusion processing.Contourlet provides a better representation of edges and texture than wavelet. Because edges and texture are fundamental elements in image representation, it is important to enhance them to get high spatial resolution. Based on LAG, the proposed fusion method further reduces the spectral distortion of the fused image. Experimental results show that the proposed fusion method is able to increase the spatial resolution and reduce the spectral distortion of the fused image simultaneously.Finally, a new post-processing method is proposed based on a modified contourlet scheme and the scale space edge detector (ss-edge detector) for suppressing blocking artifacts (BA) in block-based discrete cosine transform (BDCT) compressed images. Since there are few reference available in this field, our approach helps to uncover the application of contourlet for image deblocking processing. Firstly, by exploiting ss-edge detector, blocking map (BM) and edge information (EI) in the compressed image are extracted and protected at the same time. Then the compressed image is transformed by modified contourlet, and the adaptive thresholds are obtained according to BM. Finally, the contourlet coefficients in different subbands are filtered with adaptive thresholds in an adaptive range around the BM and EI. Experimental results show that our deblocking algorithm achieves better performance than the other iterative and noniterative methods reported in the literature.