多源影像融合技术研究

【作者】 王粮库；

【导师】 李成金； 王庆；

【作者基本信息】 苏州大学 ， 光学， 2011， 硕士

【摘要】 影像融合是图像理解和计算机视觉领域中一项重要技术,其中多源影像融合是研究的热门和重点。多源影像融合的目的是充分利用不同传感器所获取的多个待融合源图像中包含的冗余信息和互补信息,产生新的图像,以获取对同一事物或目标更为全面、客观及本质的认识,产生比单一图像信息更精确、更可靠的描述和判决,融合后的图像应该更适合于人类视觉感知和计算机后续处理。随着科学技术的发展,多源影像融合的应用已从军事领域扩展到资源分类与监测、医疗卫生、矿产探测、环境监督等民事领域,如何更好的实现多源影像融合技术及其应用,是本领域亟待解决的重大问题。本文针对多源影像融合的相关技术展开研究,主要研究工作如下:(1)讨论了影像融合的目的与意义、发展历史、研究现状、融合过程分析以及本课题的来源。(2)介绍了影像融合前图像增强、几何变换以及图像配准等预处理工作。(3)综述了常用的影像融合方法以及各自特点,并对各算法存在的问题进行适当改进。(4)分析影像融合主观和客观评价方法,并提出基于联合特征加权求和的评价方法。(5)在介绍Contourlet变换性质和原理的基础上,提出了基于改进的K均值聚类和Contourlet变换的图像融合、基于区域最大能量和Contourlet变换的图像融合以及改进的小波-Contourlet变换的图像融合,并运用主客观评价方法以及联合特征加权求和评价方法对实验结果进行验证评价,评价结果显示了上述方法的有效性。(6)通过实验展示了图像融合方法在地物分类、军事侦查以及病理探测等方面的应用。(7)总结全文工作,并对未来研究方向作以展望。更多还原

【Abstract】 Image fusion is an important technique for image analysis and computer vision,and Multi-source image fusion is popular and important. The purpose of image fusion is to fully use the complementary and redundant information in multi-source images,which comes from different sensors, and to generate new images in order to get more complete, objective and essential cognition for the same thing or target,and to produce more compete and more reliable description and decision than one image, in order that the fused image should be more suitable for human visual perception and computer further processing. With the rapid development of science and technology, the application of mufti-source image fusion has been extended from military field to civil field, such as resources classification and monitoring, medical treatment, mineral exploration,environment surveillance and so on. The major problem to be solved in this area is how to better implement technology of multi-source image fusion and its application. In this paper, we make research on the correlated technologies of multi-source image fusion. The main work is as follows:(1)Discuss multi-source image fusion concerning the purpose and significance, development history, research situation, the process analysis and sources of this topic.(2)Introduce the pretreatment work of image fusion, such as image enhancement, geometric transformation, image registration and so on.(3)Summary common methods and their characteristics of image fusion, and improve appropriately problems in the algorithm.(4)Analyse the subjective and objective evaluation method of image fusion, and propose new evaluation method based on weighted sum of the joint feature.(5)Propose image fusion based on improved K-means Clustering and Contourlet Transform, Max Energy and Contourlet Transform and improved Wavelet-Contourlet Transform,on the basis of introducing properties and principles of Contourlet Transform,and evaluate the experimental results by using the subjective, objective and weighted sum of the joint feature evaluation methods, and evaluation results show the effectiveness of above methods. (6)Show the image fusion method in the application of terrain classification, military reconnaissance and pathology detection through experiments.(7)Summary the full next and prospect for future research.更多还原