基于FRFT和Gabor小波的遥感图像变化检测研究
发布时间：2018-10-10 20:16
【摘要】：变化检测技术是遥感图像处理的重要应用之一。遥感图像变化检测是通过对同一区域不同时期的两幅已配准的遥感图像进行分析,检测出该区域地表变化信息的过程。变化检测技术在自然灾害监测、生态环境监测、战场动态监视等领域得到了广泛的应用。 本文主要研究了基于分数阶Fourier变换(Fractional Fourier Transform, FRFT)和Gabor小波的遥感图像变化检测算法,主要内容如下所述： 1、本文将非平稳信号处理理论的重要分支之一,分数阶Fourier变换应用到遥感图像变化检测中,提出一种新的无监督变化检测算法。对同一地区不同时期获得的多时相遥感图像采取低阶次的分数阶Fourier变换,然后根据图像类型计算差异图像。该算法结合每个像素的邻域信息,并利用主成分分析(Principal Component Analysis, PCA)产生与每个像素对应的基于邻域信息的特征向量。变化区域检测问题可以转化为一个二分类问题,利用K-means算法将特征向量分为两类：变化类和不变化类,得到变化检测图。最后使用光学遥感图像和合成孔径雷达(Synthetic Aperture Radar, SAR)图像数据进行仿真实验,实验结果证实了本文提出算法的有效性。 2、本文提出一种基于Gabor小波和两级聚类的多时相遥感图像变化检测算法,利用同一地区不同时期的多时相遥感图像获得差异图像,然后得到差异图像不同尺度和不同方向的Gabor小波变换,提取差异图像中每一个像素多尺度和多方向数据组成特征向量。同时本文提出一种基于模糊C均值(Fuzzy C-means, FCM)聚类的两级聚类算法,提高了变化检测的效果。通过与几种现有算法检测结果的对比,可以看到本文提出的算法具有更好的检测效果。 3、本文提出一种基于Gabor小波和PCA的多时相遥感图像变化检测算法,利用Gabor小波变换获得差异图像的多尺度和多方向数据,但是为了减少计算的复杂度,我们使用PCA对多尺度和多方向Gabor特征数据进行降维。本文分别使用了K-means和FCM两种算法实现聚类获得变化检测结果,进而得到两种聚类算法的性能对比。最后使用光学遥感图像和SAR图像进行仿真实验,证实了本文提出算法的有效性。
[Abstract]:Change detection technology is one of the important applications of remote sensing image processing. Remote sensing image change detection is a process of detecting the surface change information of the same region by analyzing two registered remote sensing images in different periods of the same area. Change detection technology has been widely used in natural disaster monitoring, ecological environment monitoring, battlefield dynamic monitoring and other fields. In this paper, the algorithm of remote sensing image change detection based on fractional Fourier transform (Fractional Fourier Transform, FRFT) and Gabor wavelet is studied. The main contents are as follows: 1. One of the important branches of the theory of non-stationary signal processing is introduced in this paper. Fractional Fourier transform is applied to remote sensing image change detection, and a new unsupervised change detection algorithm is proposed. The multitemporal remote sensing images obtained from different periods in the same area were obtained by fractional Fourier transform of low order, and then the differential images were calculated according to the image types. The algorithm combines the neighborhood information of each pixel and uses principal component analysis (Principal Component Analysis, PCA) to generate feature vectors corresponding to each pixel based on neighborhood information. The problem of variable region detection can be transformed into a two-classification problem. The feature vector can be divided into two categories by using K-means algorithm: the change class and the invariant class, and the change detection graph can be obtained. Finally, the optical remote sensing image and synthetic aperture radar (Synthetic Aperture Radar, SAR) image data are used for simulation experiment. The experimental results prove the validity of the proposed algorithm. 2. This paper proposes a multi-temporal remote sensing image change detection algorithm based on Gabor wavelet and two-level clustering. The multitemporal remote sensing images of the same region are used to obtain the differential images, and then the Gabor wavelet transform with different scales and directions is obtained to extract the multi-scale and multi-direction data of each pixel in the differential image. At the same time, a two-level clustering algorithm based on Fuzzy C-means (FCM) clustering is proposed, which improves the effect of change detection. By comparing the detection results with several existing algorithms, we can see that the algorithm proposed in this paper has better detection effect. 3. This paper proposes a multi-temporal remote sensing image change detection algorithm based on Gabor wavelet and PCA. Gabor wavelet transform is used to obtain the multi-scale and multi-directional data of the differential image, but in order to reduce the computational complexity, we use PCA to reduce the dimension of the multi-scale and multi-directional Gabor feature data. In this paper, two algorithms, K-means and FCM, are used to achieve the change detection results, and then the performance of the two clustering algorithms is compared. Finally, the simulation results of optical remote sensing images and SAR images show that the proposed algorithm is effective.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751
本文编号：2263104
[Abstract]:Change detection technology is one of the important applications of remote sensing image processing. Remote sensing image change detection is a process of detecting the surface change information of the same region by analyzing two registered remote sensing images in different periods of the same area. Change detection technology has been widely used in natural disaster monitoring, ecological environment monitoring, battlefield dynamic monitoring and other fields. In this paper, the algorithm of remote sensing image change detection based on fractional Fourier transform (Fractional Fourier Transform, FRFT) and Gabor wavelet is studied. The main contents are as follows: 1. One of the important branches of the theory of non-stationary signal processing is introduced in this paper. Fractional Fourier transform is applied to remote sensing image change detection, and a new unsupervised change detection algorithm is proposed. The multitemporal remote sensing images obtained from different periods in the same area were obtained by fractional Fourier transform of low order, and then the differential images were calculated according to the image types. The algorithm combines the neighborhood information of each pixel and uses principal component analysis (Principal Component Analysis, PCA) to generate feature vectors corresponding to each pixel based on neighborhood information. The problem of variable region detection can be transformed into a two-classification problem. The feature vector can be divided into two categories by using K-means algorithm: the change class and the invariant class, and the change detection graph can be obtained. Finally, the optical remote sensing image and synthetic aperture radar (Synthetic Aperture Radar, SAR) image data are used for simulation experiment. The experimental results prove the validity of the proposed algorithm. 2. This paper proposes a multi-temporal remote sensing image change detection algorithm based on Gabor wavelet and two-level clustering. The multitemporal remote sensing images of the same region are used to obtain the differential images, and then the Gabor wavelet transform with different scales and directions is obtained to extract the multi-scale and multi-direction data of each pixel in the differential image. At the same time, a two-level clustering algorithm based on Fuzzy C-means (FCM) clustering is proposed, which improves the effect of change detection. By comparing the detection results with several existing algorithms, we can see that the algorithm proposed in this paper has better detection effect. 3. This paper proposes a multi-temporal remote sensing image change detection algorithm based on Gabor wavelet and PCA. Gabor wavelet transform is used to obtain the multi-scale and multi-directional data of the differential image, but in order to reduce the computational complexity, we use PCA to reduce the dimension of the multi-scale and multi-directional Gabor feature data. In this paper, two algorithms, K-means and FCM, are used to achieve the change detection results, and then the performance of the two clustering algorithms is compared. Finally, the simulation results of optical remote sensing images and SAR images show that the proposed algorithm is effective.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751
【参考文献】
中国期刊全文数据库 前3条
1 李金基;焦李成;张向荣;杨咚咚;;基于两时相图像联合分类的SAR图像变化检测[J];红外与毫米波学报;2009年06期
2 马国锐;李平湘;秦前清;;基于融合和广义高斯模型的遥感影像变化检测[J];遥感学报;2006年06期
3 王桂婷;王幼亮;焦李成;;自适应空间邻域分析和瑞利-高斯分布的多时相遥感影像变化检测[J];遥感学报;2009年04期
，本文编号：2263104
本文链接：https://www.wllwen.com/guanlilunwen/gongchengguanli/2263104.html
