基于多分辨率分析的SAR图像和多光谱图像融合算法研究

发布时间：2018-10-16 20:04

【摘要】：随着传感器技术的飞速发展,现在人们获取图像信息的方式多种多样,但是仅靠单一类型的图像很难能够对目标有一个全面的呈现,所以图像融合处理技术应运而生。图像融合其主体思路是通过一系列的算法将来自不同成像系统的于同一场景拍摄的多源图像融合成一幅新的图像,从而使得使用者可以通过新的图像对目标或场景有一个更全面的感知,或者为图像的后续处理(如图像分割,提取特征,识别目标)提供帮助。本论文着重研究了基于非下采样轮廓波变换的多源图像融合算法,并通过将模糊聚类分析(FCM)和HIS彩色模型变换与非下采样轮廓波变换相结合来进行合成孔径雷达图像(SAR图像)和多光谱图像(MS图像)的融合处理。在这个框架下,本文提出了两种不同的融合方式:一是基于多窗口的融合方式,该算法在低频系数融合时将根据不同的区域采取不同的窗口融合规则,以保证获得更多的源图像信息,同时在高频系数融合时则采用区域块能量取大的方式来确保图像的细节信息能够最大程度得到保留;二是基于区域的融合方式,该算法将采用基于多种区域特性的新型加权平均方式进行低频子带系数的融合,而在高频子带系数融合方面将根据相应高频子带系数的相关系数的大小而选择不同的融合方式,一面可以减少高频子带系数融合过程中的计算量,同时也能够保留图像的细节信息。
[Abstract]:With the rapid development of sensor technology, there are a variety of ways to obtain image information, but it is difficult to get a comprehensive representation of the target by a single type of image, so the image fusion technology emerges as the times require. The main idea of image fusion is to fuse multi-source images from different imaging systems into a new image by a series of algorithms. Thus, the user can have a more comprehensive perception of the target or scene through the new image, or provide help for the subsequent processing of the image (such as image segmentation, feature extraction, target recognition). This paper focuses on the multi-source image fusion algorithm based on non-downsampling contour wave transform. The fusion of synthetic Aperture Radar (SAR) image and multispectral image (MS image) is carried out by combining fuzzy clustering analysis (FCM) and HIS color model transform with non-downsampling contour wave transform. In this framework, this paper proposes two different fusion methods: one is based on multi-window fusion, the algorithm will adopt different window fusion rules according to different regions in low frequency coefficient fusion. In order to ensure that more source image information can be obtained, at the same time, when the high frequency coefficients are fused, the region block energy is used to ensure that the details of the image can be preserved to the maximum extent; second, the fusion method based on the region, In this algorithm, a new weighted averaging method based on multiple region characteristics is used to fuse the low frequency subband coefficients, and different fusion methods are chosen according to the correlation coefficients of the corresponding high frequency subband coefficients. On the one hand, it can reduce the computation in the fusion process of the high frequency subband coefficients, but also can preserve the details of the image.
【学位授予单位】：沈阳航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】