【作者】 张大明；

【导师】 胡茂林；

【作者基本信息】 安徽大学 ， 应用数学， 2005， 硕士

【摘要】 多源图像融合技术研究的是如何综合利用不同类型传感器所获得的图像,通过一定的算法得到一副信息更加准确、丰富的新图像,从而产生比单一传感器更精确、更完整、更可靠的描述和判决,它在军事、民用等方面有着极为广泛的应用背景,是目前科技界的一个热门研究领域。 从待处理数据的抽象层次上分,融合技术可分为像素级、特征级和决策级三种。本文主要是围绕像素级展开的,深入研究了SAR图像与光学图像融合、多聚焦图像融合和多光谱图像与高分辨率图像融合的技术,并提出了一些改进方法,取得了较好的效果。本文主要在以下方面作了有益的探索和改进: ● 详细讨论了SAR图像成像特点以及于光学图像融合的必要性,并指出了基于高斯—拉普拉斯金字塔方法融合中出现的问题,给出了改进算法和相应的试验图像。 ● 叙述了小波变换及Mallat快速算法,以及在图像融合中的应用,详细介绍了方向可调滤波器的理论内容。 ● 将方向可调滤波器与小波变换方法相结合,提出了一种基于区域方向 能量测量的图像融合方法,并将其应用到多聚焦图像融合中,从尺度参数、小波基的选取、融合规则等方面进行了融合结果的讨论和评价,取得了较好的融合效果;又将该改进方法应用于多光谱图像与高分辨率图像的融合中,进行了类似的分析,取得了较好的融合效果。更多还原

【Abstract】 The purpose of the image fusion technique is how to use different images from many kinds of sensors, through some algorithm, to get a more accurate and abundant image ,which includes more information, then we can achieve more complete and more reliable description and decision. The image fusion technique is used very extensively in military and civil application, and now image fusion is a very popular research filed.From the abstract level of the data which is to be dealed with, the image fusion technique can be divided intoto pixel level, feature level and decision level. This paper is studies the image fusion technique based on pixel level, and the SAR/Optical image fusion multi-focus image fusion and multi- spectrum/high resolution image fusion technique are studied deeply. Also some improved algorithm were put forward and some better results were obtained. This paper mainly makes some studies and improves in some aspects as follows:· Discussing the characteristics of SAR image and the necessary of the SAR/Optical image fusion, pointing out the problems in the algorithm based on Gaussian-Laplacian pyramid and giving out the improved algorithm and experimental fusion image.· Narrating wavelet transform, Mallat algorithm and their implication in image fusion, and also introducing the theory of steerable filters in detail.· By combining wavelet transform with steerable filters, a image fusion algorithm based on the measure of area orientations energy was proposed. On the one hand, the algorithm was applied to the multi-focus image fusion, from the scale parameter, the choose of the basis of wavelet and the rules of fusion, the fusion results were discussed and evaluated, good fusion results were acquired; on the other hand, the algorithm also was applied to the multi-spectrum/high resolution image fusion, some similar analysis was carried through, and also good results were got, the algorithm was proved to be very effective.更多还原