微波与光学遥感图像的分类与重建
发布时间：2018-10-05 11:09
【摘要】：拍摄条件和天气气候等客观因素的干扰造成的数据缺失,对光学遥感影像的解译和应用造成了巨大的限制。常见的数据缺失类型包括云层与云层阴影遮挡、条带状噪声和其他噪声。光学遥感图像重建方法利用相关的图像数据,对图像缺失部分的数据进行还原,使得遥感图像在主观视觉上有着更高的辨识度,对后续诸如遥感图像解译、目标检测与监督、地物分类和变化检测等应用带来了极大的便利。目前,常用的光学遥感图像重建算法可以分为基于辅助图像信息的重建方法和基于图像修复技术的重建方法。其中,基于辅助图像信息的重建方法是指利用多光谱或多时域的图像之间的相关性信息,对缺失区域进行重建。基于图像修复技术利用待修复图像数据完好区域的图像信息,对缺失区域进行估计。文章所提出的算法基于图像修复技术。这一类重建方法的瓶颈之一是缺乏待修复数据的图像先验信息,制约了其重建准确度。考虑到微波图像不受气候影响,以及对云层的穿透性,文章利用常用的微波图像聚类分类方法对微波遥感图像进行分割,得到其图像结构信息,提出一种结合微波图像先验结构信息的图像修复方法,应用于光学遥感图像重建。实验证明,在海岸线的应用场景下,文章提出的算法更好地保持了图像结构的连贯性,相较于几种常用的重建方法,文章提出的算法具有更高的重建精度。文章所做主要研究工作内容如下:1、基于Criminisi图像修复算法,文章将其应用于光学遥感图像云层移除的具体重建问题,并使用微波图像替换掉了原有算法中的光学亮度图像,提出了一种基于微波遥感图像的等高照线数据项计算方法,增强了原有算法对地表不规则结构的重建精度。2、文章研究分析了常用的聚类图像分类算法,对微波遥感图像进行了分割,分析比较了不同聚类算法之间的分割效果和各自的优劣之处。针对海岸线应用场景的重建问题,文章将分割后的微波图像结果作为先验信息,提出一种先分类后重建的重建方法模式,在地表分界线类型区域的重建中,取得了良好的重建精度。与其他常用光学图像重建方法相比,文章所得重建图像与原始纯净图像具有更好的相似程度。
[Abstract]:The lack of data caused by the interference of objective factors such as shooting conditions and weather and climate has greatly restricted the interpretation and application of optical remote sensing images. Common data loss types include cloud and cloud shadows, banded noise and other noises. The method of optical remote sensing image reconstruction uses the relevant image data to restore the missing part of the image, which makes the remote sensing image have higher recognition degree in subjective vision, such as interpretation of remote sensing image, target detection and supervision. The application of ground object classification and change detection has brought great convenience. At present, the commonly used optical remote sensing image reconstruction algorithms can be divided into auxiliary image information based reconstruction method and image restoration technology based reconstruction method. The reconstruction method based on auxiliary image information refers to the reconstruction of missing region by using the correlation information of multi-spectral or multi-time domain images. Based on the image restoration technique, the missing region is estimated by using the image information of the intact region of the image data to be repaired. The proposed algorithm is based on image restoration technology. One of the bottlenecks of this kind of reconstruction method is the lack of image prior information of the data to be repaired, which restricts its reconstruction accuracy. Considering that microwave images are not affected by climate and are penetrating to clouds, the microwave remote sensing images are segmented by the commonly used methods of microwave image clustering and classification, and their image structure information is obtained. An image restoration method based on the prior structure information of microwave image is proposed, which is applied to the reconstruction of optical remote sensing image. The experiments show that the algorithm proposed in this paper can better maintain the coherence of image structure in the application scene of shoreline. Compared with several commonly used reconstruction methods, the algorithm proposed in this paper has higher reconstruction accuracy. The main work of this paper is as follows: 1. Based on the Criminisi image restoration algorithm, this paper applies it to the reconstruction of optical remote sensing image cloud removal, and uses microwave image to replace the optical luminance image in the original algorithm. In this paper, a method of calculating the isometric line data item based on microwave remote sensing image is proposed, which enhances the reconstruction accuracy of the original algorithm to the irregular structure of the earth's surface. In this paper, the commonly used clustering image classification algorithm is studied and analyzed. The segmentation of microwave remote sensing image is carried out, and the segmentation effect of different clustering algorithms and their advantages and disadvantages are analyzed and compared. Aiming at the reconstruction problem of shoreline application scene, this paper takes the result of microwave image segmentation as a priori information, and proposes a reconstruction method mode of classification and reconstruction, which is used in the reconstruction of the surface boundary type area. Good reconstruction accuracy has been obtained. Compared with other commonly used optical image reconstruction methods, the reconstructed image has a better similarity with the original pure image.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751
[Abstract]:The lack of data caused by the interference of objective factors such as shooting conditions and weather and climate has greatly restricted the interpretation and application of optical remote sensing images. Common data loss types include cloud and cloud shadows, banded noise and other noises. The method of optical remote sensing image reconstruction uses the relevant image data to restore the missing part of the image, which makes the remote sensing image have higher recognition degree in subjective vision, such as interpretation of remote sensing image, target detection and supervision. The application of ground object classification and change detection has brought great convenience. At present, the commonly used optical remote sensing image reconstruction algorithms can be divided into auxiliary image information based reconstruction method and image restoration technology based reconstruction method. The reconstruction method based on auxiliary image information refers to the reconstruction of missing region by using the correlation information of multi-spectral or multi-time domain images. Based on the image restoration technique, the missing region is estimated by using the image information of the intact region of the image data to be repaired. The proposed algorithm is based on image restoration technology. One of the bottlenecks of this kind of reconstruction method is the lack of image prior information of the data to be repaired, which restricts its reconstruction accuracy. Considering that microwave images are not affected by climate and are penetrating to clouds, the microwave remote sensing images are segmented by the commonly used methods of microwave image clustering and classification, and their image structure information is obtained. An image restoration method based on the prior structure information of microwave image is proposed, which is applied to the reconstruction of optical remote sensing image. The experiments show that the algorithm proposed in this paper can better maintain the coherence of image structure in the application scene of shoreline. Compared with several commonly used reconstruction methods, the algorithm proposed in this paper has higher reconstruction accuracy. The main work of this paper is as follows: 1. Based on the Criminisi image restoration algorithm, this paper applies it to the reconstruction of optical remote sensing image cloud removal, and uses microwave image to replace the optical luminance image in the original algorithm. In this paper, a method of calculating the isometric line data item based on microwave remote sensing image is proposed, which enhances the reconstruction accuracy of the original algorithm to the irregular structure of the earth's surface. In this paper, the commonly used clustering image classification algorithm is studied and analyzed. The segmentation of microwave remote sensing image is carried out, and the segmentation effect of different clustering algorithms and their advantages and disadvantages are analyzed and compared. Aiming at the reconstruction problem of shoreline application scene, this paper takes the result of microwave image segmentation as a priori information, and proposes a reconstruction method mode of classification and reconstruction, which is used in the reconstruction of the surface boundary type area. Good reconstruction accuracy has been obtained. Compared with other commonly used optical image reconstruction methods, the reconstructed image has a better similarity with the original pure image.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP751
【相似文献】
相关期刊论文 前10条
1 曾友州;胡莹;曾伟一;郑晓霞;;提取数字图像边缘的算法比较[J];成都航空职业技术学院学报;2009年04期
2 宋建中;;喷雾图像的自动分析[J];光学机械;1988年04期
3 涂承媛;曾衍钧;;医学图像边缘快速检测的模糊集方法[J];北京工业大学学报;2005年06期
4 常君明;冯，
本文编号：2253152
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2253152.html
