基于非抽样复轮廓波变换的图像去噪算法研究

将二维双树复小波变换(DT-CWT)与非抽样方向滤波器组(NSDFB)相结合,构造一种新的非抽样复轮廓波变换(NSCCT),并对其平移不变性作了相应证明。同时利用对称的正态逆高斯(NIG)分布先验概率模型和贝叶斯最小均方算法,提出一种基于NSCCT的图像去噪算法。实验结果表明,本文构造的NSCCT能够有效地抑制伪Gibbs现象,并且具有更丰富的方向分量,因而在图像的细节和纹理表现方面有一定的优势。     

基于NSCT和SVM的脑MRI医学图像识别

利用NSCT变换具有多尺度和平移不变性,能够稀疏地表示纹理图像的特点,将具有丰富纹理信息的人体脑部核磁共振(MR)图像,从空间域变换到频率域表示。提取变换后表征图像特性的低频子带均值、方差及高频16个方向子带能量作为特征向量,输入SVM分类器进行分类识别。实验结果表明该方法对非病变脑部MR图像识别准确率达到100%,病变脑部MR图像的识别率达到90.90%,综合识别率达到95.45%。且该方法提取的特征维数少,识别速度快,识别率高,能够快速区分病变与非病变脑部MR图像。     

基于无抽样Contourlet变换的图像增强方法

为了有效增强图像的细节信息,研究了基于无抽样Contourlet变换的图象增强方法。首先将待增强图像进行无抽样Contourlet变换,然后使用映射函数对无抽样Contourlet系数进行增强处理,最后进行无抽样Contourlet反变换得到增强后的图像。实验结果表明,该方法可以有效增强图像的微弱边缘信息。     

基于Contourlet变换的稳健性图像水印算法

提出了基于Contourlet变换的数字图像水印算法。与小波变换不同的是,Contourlet变换采用类似于线段(contoursegment)的基得到一种多分辨、局部化、方向性的图像表示。水印信号通过基于内容的乘性方案加载到Contourlet变换系数。在采用零均值广义高斯分布拟合Contourlet变换系数的基础上,提出采用极大似然估计实现水印的盲检测。依据Neyman-Pearson准则,在给定虚警率的情况下对判决准则进行了优化。实验结果表明在保证水印隐蔽性的前提下,水印对常见的信号处理手段以及几何变换具有很好的稳健性。     

Stereo matching cost computation based on nonsubsampled contourlet transform

A new matching cost computation method based on nonsubsampled contourlet transform (NSCT) for stereo image matching is proposed in this paper. Firstly, stereo image is decomposed into high frequency sub-band images at different scales and along different directions by NSCT. Secondly, by utilizing coefficients in high frequency domain and grayscales in RGB color space, the computation model of weighted matching cost between two pixels is designed based on the gestalt laws. Lastly, two types of experiments are carried out with standard stereopairs in the Middlebury benchmark. One of the experiments is to confirm optimum values of NSCT scale and direction parameters, and the other is to compare proposed matching cost with nine known matching costs. Experimental results show that the optimum values of scale and direction parameters are respectively 2 and 3, and the matching accuracy of the proposed matching cost is twice higher than that of traditional NCC cost.     

一种新的基于非下采样Contourlet变换的图像去噪方法

提出了一种新的结合非下采样Contourlet变换(NSCT)和高斯比例混合模型的图像去噪方法.采取的主要方法为:1)通过NSCT对图像进行分解;2)根据高斯比例混合模型建立图像模型;3)利用贝叶斯估计进行图像去噪.实验结果表明,相对于已有算法,本文方法降噪效果好,在去噪性能指标和边缘保持的主观视觉上都表现出优异的性能.     

基于非下采样轮廓小波变换的图像去噪

针对采用下采样滤波器结构的轮廓波、轮廓小波在图像去噪过程中会引入伪吉布斯现象,利用小波变换(WT)和非下采样方向滤波器组(NDFB)构造了一种新的多尺度、多分辨率图像的非下采样轮廓小波变换(NWCT)。WT去除了拉普拉斯金字塔滤波器(LPF)的计算冗余,NDFB保证了该变换具有平移不变性。为了验证该变换的有效性,对其进行了图像去噪实验。实验结果表明,所提出方法能获得比WT、轮廓波变换(CT)、轮廓小波变换(WCT)更高的峰值信噪比(PSNR),并且能够很好地抑制伪吉布斯现象。     

基于特征点的红外与可见光图像配准研究

提出了一种基于特征点的红外和可见光图像配准方法。首先根据边缘图像的结构分别提取红外图像特征点和可见光图像特征点,其次根据结合了形状结构和灰度、梯度角信息的配准准则寻找两幅图像的对应特征点,最后利用三对特征点根据刚体变换得到图像的缩放倍率、旋转角度和平移坐标,从而实现两幅图像的配准,并通过实验给予了证明。     

非抽样轮廓波变换域图像融合方法研究

轮廓波变换由于不具备平移不变性,应用于图像融合会产生褶皱现象。针对轮廓波变换的这一不足,提出了一种基于非抽样轮廓波变换的图像融合算法。首先由非抽样金字塔和非抽样方向滤波器组对源图像进行多尺度、多方向分解,然后对得到的变换系数进行融合处理得到融合系数,最后对融合系数进行重构即可得到融合图像。算法将具备平移不变性的非抽样轮廓波变换应用于图像融合,消除了褶皱现象,获得了更好的融合效果。实验结果表明,与其他基于多分辨率分解的融合算法相比,该算法显著减小了融合图像的MSE及ΔH,提高了Correlation值及融合图像的清晰度,是一种有效的融合算法。     

基于非亚采样轮廓波的舰船尾迹检测

介绍了海洋表面合成孔径雷达（SAR）图像的舰船尾迹检测算法。该算法首先对图像进行对数变换，将乘性噪声转化为加性噪声，然后对图像进行非亚采样contourlet 变换，根据变换后的像素均值对变换系数进行调整，以便去除噪声和提取尾迹轮廓。接着对经非亚采样contoulet调整后的图像进行局部Radon变换，即在对像素积分过程中沿着线性特征被分割后的若干短线段进行积分，而非对整幅图像进行积分。最后通过采用形态学膨胀和腐蚀的方法将临近的分离短线段连接起来便于后续的处理。采用该算法对实际SAR图像进行舰船尾迹检测，结果表明：该算法在背景噪声很强时仍能检测出略微有弯曲的尾迹信号。     

基于非亚采样轮廓波的舰船尾迹检测

介绍了海洋表面合成孔径雷达(SAR)图像的舰船尾迹检测算法.该算法首先对图像进行对数变换,将乘性噪声转化为加性噪声,然后对图像进行非亚采样contourlet变换,根据变换后的像素均值对变换系数进行调整,以便去除噪声和提取尾迹轮廓.接着对经非亚采样contoulet调整后的图像进行局部Radon变换,即在对像素积分过程中沿着线性特征被分割后的若干短线段进行积分,而非对整幅图像进行积分.最后通过采用形态学膨胀和腐蚀的方法将临近的分离短线段连接起来便于后续的处理.采用该算法对实际SAR图像进行舰船尾迹检测,结果表明:该算法在背景噪声很强时仍能检测出略微有弯曲的尾迹信号.     

Despeckling SAR images based on a new probabilistic model in nonsubsampled contourlet transform domain

A method for synthetic aperture radar (SAR) image despeckling based on a probabilistic generative model in nonsubsampled contourlet transform (NSCT) domain was proposed. The shrinkage estimator in NSCT domain consists of a new type of likelihood ratio and prior ratio, both of which are dependent on the estimated masks for the NSCT coefficients. While the previous probabilistic approaches are restricted to parametric models, the limitation is eliminated and the hybrid density model is applied in this paper. The suggested approach does not make heavy assumptions on the NSCT coefficient distribution, so that it can handle complex NSCT coefficient structures. The likelihood ratio is composed of the hybrid density, and the prior ratio is equipped with the selective neighborhood systems to enhance the detail information. The method can effectively adapt the shrinkage estimator to the redundancy property of the NSCT. The proposed approach was applied to real SAR images despeckling and compared through the SAR image vision effect, the equivalent number of looks, and the edge sustain index. Experimental results show that the proposed approach outperforms previous works involved in the paper with the better despeckling result and edge preservation.     

A nonsubsampled countourlet transform based CNN for real image denoising

The state of the art deep learning based denoising methods can achieve great denoising results. However, due to the lack of clean training data, the ground truth and noise level are unknown, traditional denoising methods are difficult to remove blind noise in general images. Furthermore, deep learning methods require specific noise levels to train the model, and specific models are built only deal with one noise level. In this paper, we propose a Nonsubsampled Countourlet Transform based convolutional network model (CTCNN) to deal with Gaussian noise and the noise of real images. The model is modified by U-Net, nonsubsampled Countourlet Transform (NSCT) and inverse NSCT are used to instead of sum pooling layer and up-convolution operation. NSCT can decrease the size of feature maps and preserve details of images without information loss. Different training strategies are adopted to train models in order to handle blinding noise such as underwater images which contain noise naturally. Simulation results show the proposed method is effective in standard images dataset and blind noisy images. The model we proposed has been compared with other state of the art denoising methods, and better subjective representation and PSNR improvement are obtained.     

基于自适应阈值SIFT算法的TOFD图像配准

文中运用超声TOFD的方法对焊缝进行检测,通过对同一缺陷进行检测,获取多组数据,产生一系列TOFD图像,将其中一幅图像作为参考图像,其他的作为待配准图像,运用SIFT的方法进行配准。对SIFT算法中距离比阈值参数进行分析,采用自适应距离比阈值的方法,寻找到最优的距离比阈值参数,对图像进行特征匹配,提高配准精度。同时,将配准后的图像,通过迭代相加的方式,使检测图像的检测误差和噪声进一步降低。实验表明,该方法能显著地改善TOFD图像的质量,信噪比达到24d B。     

结合图像信息熵和特征点的图像配准方法

在分析当前主要的图像配准技术之后,针对图像特征点的分布和同名点的匹配问题,提出了结合图像信息熵和特征点的图像配准方法。首先对图像进行一定程度的分块,根据信息论的方法,计算每一块的信息熵,信息熵的大小基本反映了各个模块的纹理变换情况。然后根据各个模块的信息熵大小,进行图像的粗匹配。之后在各个模块提取出一定数目的特征点,信息熵大,纹理信息丰富,选取的特征点就相应较多,反之则纹理信息变化不大,选取的特征点数目较少。最后根据这些具有代表性的同名点进行精确匹配。为验证该方法的有效性,对两幅图像进行传统方法和改进的图像配准方法的比较。     

A hybrid feature extraction method for SAR image registration

Extracting and matching correct correspondence between two images are significant stages for feature-based synthetic aperture radar (SAR) image registration. Two methods of feature extraction were employed in this study. Blob features were obtained by combining a Gaussian-guided filter (GGF) with a scale invariant feature transform, and corner features were obtained from the GGF. A GGF can store edge information and operate more effectively than a Gaussian filter. The ratio of average was used to compute gradients in order to reduce the speckle effect. Fast sample consensus (FSC) algorithm was combined with complete graph method for feature correspondence matching. Although FSC algorithm can extract valid correspondence, it may not be efficient enough to deal with SAR images due to its random nature and the large number of outliers in the data. Therefore, a graph-based algorithm was employed to solve the problem by eliminating outliers. The proposed hybrid method was tested on several real SAR images having different properties. The results showed that the proposed method performed the automated registration of SAR images more accurately and efficiently.     

基于非下采样剪切波变换的红外与可见光图像融合

由于传统稀疏表示(SR)冗余字典单一,脉冲耦合 神经网络(PCNN)模型参数设置复杂,为了解决上述问题,提 出了基于非下采样剪切波变换(NSST)的红外与可见光图像融合算法。该算法首先通过NSST将 源图像分解成低频子 带和高频子带。然后,使用自适应稀疏表示(ASR)模型进行NSST域低频部分稀疏系数的融合 ;同时,采用参数自适 应脉冲耦合神经网络(PA-PCNN)模型融合相应的高频部分。最后,对融合后的低频和高频波 段进行NSST反变换,重 建得到融合结果图。实验结果表明:该算法解决了传统SR模型的“块效应”问题,克服了PC NN模型中自由参数的设置难点,在主观视觉和客观评价上均优于现有算法。     

非降采样Contourlet图像边缘检测算法

介绍了非降采样Contoudet变换(NSCT)的结构特点与系数分布特性,并提出了基于NSCT的图像边缘检测算法,实现了对图像高频边缘的有效去噪和提取.该方法具有多向选择性、高低频信息并行检测、信息利用率高、两类边缘相互补偿修正精确定位及抗噪声干扰能力强等特点.通过仿真实验并与Prewitt、Sobel、Canny及多尺度小波边缘检测算法进行比较,验证了该方法的有效性和优越性.     

改进非下采样轮廓波在散斑条纹中的滤波处理

为了改进传统多尺度变换滤波在电子散斑干涉（ESPI）条纹图中去噪效果和边缘细节保护不理想问题，提出改进非下采样轮廓波（NSCT）滤波算法。采用离散平稳小波变换和NSCT变换模型，联合非线性扩散和改进的快速非局部均值滤波算法，进行了理论分析和实验验证，取得了将本文中算法应用于模拟和实验ESPI条纹图滤波效果定量分析的数据。结果表明，本文中的算法在模拟ESPI条纹图和实验图相比其它算法散斑指数最小分别为0.41121，0.38043，0.35362，对应峰值信噪比最大；该算法在提升去噪能力的同时，能够更好地恢复条纹细节信息。研究结果为以后应用多尺度变换滤波在ESPI条纹图打下了基础。     

基于Contourlet变换和SVM的SAR图像目标识别

针对SAR图像具有斑点噪声,特征提取较难的特点,提出了基于Contourlet变换和SVM的SAR图像目标识别分类算法.该算法的在特征提取时利用了Contourlet域的标准偏差进行特征提取的方法,后端用支持向量机分类器,提高分类精度.实验结果证明该分类算法能够减少SVM的特征维数,具有较好的分类性能.