【作者】 王振飞；

【导师】 王能超； 施保昌；

【作者基本信息】 华中科技大学 ， 计算机软件与理论， 2006， 博士

【摘要】 小波理论及其应用的研究是一项十分有意义的课题。经典小波分析方法在理论上和在工程应用中获得了巨大的成功,但它不是十分完美的,存在一些严重的不足。提升格式小波变换与经典小波变换相比具有小波构造简单、变换容易实现和节省存储空间等优点。另外,提升格式提供了一种比Mallat算法更快、更简洁和更易操作的小波变换方法,而且还可以立即得到其逆变换。在讨论带参数双正交整数提升小波构造方法的基础上,探讨了基于提升小波变换的图象压缩、数字水印和图像融合改进算法。针对SPIHT算法中存在中频和高频子带小波系数的不重要数具有相关性,设计了SPIHT改进算法。新算法利用一个组合函数来减少相应位置中频子带的不重要小波系数的冗余,同时构造集合d来减少高频子带带间的冗余。数值实验表明,新算法有较为理想的压缩效果,峰值信噪比和人眼视觉效果均有所提高。在数字水印方面,安全性和鲁棒性一直是研究的热点与难点。分别提出了基于小波和人类视觉系统的稳健数字水印算法和基于混沌置乱和人类视觉系统的小波域数字水印算法。数值实验表明,这些算法在抗压缩和抗剪切攻击方面均有很强的能力,具有较强的鲁棒性。特别是第二个算法使嵌入的水印具有更强的不可见性。混沌密钥的使用与小波域重要系数修改的隐蔽性,加强了水印信息的保密性,性能优于文献上相近的方法。为了提高水印技术性能,一个有效的策略是探索图像和水印的特征而自适应地嵌入水印。作者还提出了基于神经网络的小波域稳健盲数字水印算法。利用神经网络的学习特性,找出嵌入的水印与重要系数的关系。训练后的神经网络可以从水印图像中恢复出水印。提出了一种基于小波对比度和神经网络的多聚焦图像融合算法。首先对各源图像进行小波变换,根据变换后系数计算出图像的小波对比度,选取源图像部分区域小波对比度作为前溃神经网络的训练样本,调整神经网络权重;然后用训练好的神经网络组合融合图像的小波系数,对组合后的系数进行一致性校验;最后对该系数进行小波逆变换,得到融合图像。数值实验表明,该算法能够较好地解决多聚焦图像融合问题,生成的融合图像效果优于传统图像融合方法。曲波是小波理论的发展,在图像处理方面具有小波所不具有的优点。它能够为数字图像中的边缘以及线性特征提供一种比小波多分辨率分析更加精确的分析方法。在研究了曲波变换的特性后,提出了一种基于曲波变换的图像融合方法,并分别将由不同传感器获得的不同分辨率的遥感图像和医学图像利用此方法进行了融合。新方法不仅能够完好地显示了源图像各自的信息,而且能很好地将源图像的细节融合在了一起。数值实验表明采用这种方法可以得到更好的融合效果。更多还原

【Abstract】 Wavelet theories and their application is an important subject. This thesis focuses on the construction of the biorthonormal integer wavelets with parameter based on lifting scheme. Lifting scheme is used for constructing the conventional wavelets.It is a method of constructing wavelets in space domain.The motif of the method is that you can lift the sample multi-resolution to the multi-resolution that you want. Thus, you can build your own wavelets based on what you want according to the lifting scheme. This paper introduces the engineering application in image compression, digital watermark and image fusion based on discussing of the biorthonormal 9/7 integer wavelets with parameter.This paper proposes some modifications to the SPIHT algorithm. It is based on the idea of the insignificant correlation of wavelet coefficients among medium- and high-frequency subbands. In this scheme, insignificant wavelet coefficients that correspond to the same spatial location in the medium subbands can be used to reduce the redundancy by a combined function proposed in associated with the modified SPIHT. In high-frequency subbands, the modified SPIHT proposes a dictator to reduce the interband redundancy. Experimental results indicate that the proposed technique improves the quality of the reconstructed image in terms of both the peak signal-to-noise ratio (PSNR) and the perceptual results over the original SPIHT at the same bit rate.A novel wavelet based watermarking scheme for copyright protection of digital image is presented. With this method, watermark was scrambled through Arnold transform before it was embedded. To embed the watermark robustly and imperceptibly, watermark bits are added to the significant coefficients of each subband selected by considering the human visual system (HVS) characteristics. From the experimental results it can be observed that proposed method is robust to wide variety of attacks.A new robust watermarking algorithm in DWT (discrete wavelet transform) domain is presented. Due to good randomness and easy reproducibility of chaos,the watermark is firstly permuted by the hashed chaotic sequence. To embed the watermark robustly and imperceptibly, watermark bits are added to the significant coefficients of each subband selected by considering the human visual system (HVS) characteristics. From the experimental results it can be observed that proposed method is robust to wide variety of attacks and also superior to the similar method in the literature. The usefulness of multilevel chaotic keys and privacy of modification to frequency coefficients enhance the security of watermark information.A novel multi-focus image fusion method using wavelet contrast and neural network is proposed. Firstly,source images are decomposed with wavelet transform (WT) and wavelet contrasts are obtained. Some of them are selected as sample for training neural network. Then,output wavelet coefficients of fused image though neural network and verify them by consistency. Finally, the fused image is obtained with inverse WT. Experimental results show that the proposed method outperforms traditional wavelet transform (WT) method for image fusion.Curvelet develops from wavelet theory, but it has several incomparable advantages on image processing. As to the edge and curve feature of the image, curvelet analysis offers a more precise way than wavelet multiresolution analysis. In this paper, after a research on the characteristics of curvelet transform, a novel fusion method based on curvelet transform is proposed. The fused images are obtained through merging various remote sensing images and medical images by applying this image fusion method respectively. Comparing with other methods and appraising on objective and subjective performance, this method can display information of the each input image and fuse details of input images successfully. Results show that the proposed method outperforms the traditional approach both visually and quantitatively.In the end, a discussion and analysis of promising research way based on this paper is presented, including lifting scheme of wavelet-based compression techniques for the multi-dimensional image、multicast watermarking protocol based on watermarking algorithm、objective evaluation method for image fusion based on wavelet lifting scheme.更多还原