经典与新型小波理论及其在图像处理中的应用

发布时间：2018-05-07 11:41

本文选题：经典小波 + 多分辨分析　；参考：《中国地质大学（北京）》2014年博士论文

【摘要】：本文致力于研究小波理论在图像尤其是遥感图像处理领域的应用.文章首先总结了小波理论的源流与发展,介绍了连续小波变换、离散小波变换、多分辨分析和几类经典小波.对小波理论在信号和图像处理领域的应用指出了其优缺点.进而,针对经典小波的缺陷改良,介绍了新型小波系统的发展沿革和现状,主要包括脊波变换(Ridgelet)、曲波变换(Curvelet)、轮波变换(Contourlet)等,阐述了新型小波系统的主要思想和算法. 本文研究了数字音频和图像信号的小波去噪理论,对一维音乐信号和二维标准图像重点讨论了小波阈值去噪算法,并进行了去噪仿真实验,根据峰值信噪比(PSNR)等判别参数,将实验结果进行了对比. 一个明确而新颖的观点是从纹理图像的角度理解地图,对遥感图像作卡通和纹理分解.本文给出了离散小波框架变换的定义和用它来进行纹理刻画的方法,由此,结合Gabor小波和支持向量机SVM提出了图像分割算法,并对标准纹理图像库图像和北京地区遥感图像进行了纹理图像分割实验. 本文建立了数字脊波和数字曲波变换的图像分解算法,比较了基于全局和局部对偶框架的GDF算法和LDF算法.给出了二维Gabor滤波器基函数的一般形式和图形,讨论了滤波器选取原则、设计方法和特征提取步骤,并用基于纹理分析的Gabor方法对遥感图像目标区域进行特征提取和边缘刻画. 轮波变换具有良好的多分辨率、局部化及方向性特征,本文重点讨论了方向滤波器组的设计.针对影像融合,探讨了基于多孔atrous小波与广义IHS变换结合的融合方法.此外,本文把测地线活动轮廓(GAC)用曲波演化方法进行多尺度改进,获得曲波测地线蛇形活动轮廓(CGS).探讨了边界探测函数的构造.文章结尾提出“全息地图”的概念构想.
[Abstract]:This paper focuses on the application of wavelet theory in image processing, especially in remote sensing image processing. In this paper, the source and development of wavelet theory are summarized, and the continuous wavelet transform, discrete wavelet transform, multi-resolution analysis and several classical wavelets are introduced. The advantages and disadvantages of wavelet theory in signal and image processing are pointed out. Then, aiming at the defect improvement of classical wavelet, this paper introduces the development and current situation of the new wavelet system, including ridgelet transform, Qu Bo transform and wheel wave transform, etc. The main ideas and algorithms of the new wavelet system are expounded. In this paper, the wavelet denoising theory of digital audio and image signals is studied. The wavelet threshold denoising algorithm for one-dimensional music signal and two-dimensional standard image is discussed, and the simulation experiment is carried out, and the discriminant parameters such as PSNR are analyzed according to the peak signal-to-noise ratio (PSNR). The experimental results are compared. A clear and novel view is to understand the map from the point of view of texture image and to decompose the remote sensing image by cartoon and texture. In this paper, the definition of discrete wavelet frame transform and the method of texture characterization are given. Therefore, an image segmentation algorithm based on Gabor wavelet and support vector machine (SVM) is proposed. The texture image segmentation experiments of standard texture image library and remote sensing image in Beijing area are also carried out. In this paper, the image decomposition algorithm of digital ridgelet and digital Qu Bo transform is established, and the GDF algorithm and LDF algorithm based on global and local dual frame are compared. In this paper, the general form and graph of the basic function of two-dimensional Gabor filter are given, and the selection principle, design method and feature extraction procedure of the filter are discussed, and the feature extraction and edge characterization of the target region of remote sensing image based on the Gabor method based on texture analysis are carried out. The wheel-wave transform has good multi-resolution, localization and directionality. The design of directional filter banks is discussed in this paper. For image fusion, the fusion method based on porous atrous wavelet and generalized IHS transform is discussed. In addition, in this paper, the geodesic active contour (GAC) is improved by using Qu Bo's evolution method, and then we obtain the active contours of Qu Bo geodesic snake-like activity. The construction of boundary detection function is discussed. At the end of the paper, the concept of "holographic map" is put forward.
【学位授予单位】：中国地质大学（北京）
【学位级别】：博士
【学位授予年份】：2014
【分类号】：P237;TP751

【参考文献】