多源遥感图像融合技术研究

发布时间：2018-01-22 03:49

本文关键词： 图像融合遥感方向性信息测度 IHS变换 YIQ变换 PCA变换金字塔结构多分辨率分析小波变换区域特征区域边界强度自适应融合容错评价准则　出处：《西北工业大学》2004年硕士论文　论文类型：学位论文

【摘要】： 作为图像融合领域的一个重要分支，，多源遥感图像融合研究的是如何综合利用不同航空遥感传感器所获取的图像信息，来产生新的数据，以获取对同一事物或目标的更为全面、客观及本质上的认识。在高度信息化的今天，遥感图像融合已经成为图像处理和图像信息理解领域中不可或缺的技术，并在很多军事和民用方面有着重要应用。本文应用图像处理和现代信号处理技术中的多种手段，研究了不同层次上多源遥感图像的融合方法。通过对不同航空遥感传感器所获取的图像数据进行融合，从而提高图像的分辨率、图像分析结果的准确性和置信度，并最终提高对特定航空目标进行自动检测、识别的有效性。本文的主要研究内容和工作可总结如下： 1．介绍了图像融合的基本概念和原理，进而系统分析了多种传统的图像融合方法，并通过实验对其特点和性能做了细致的对比，总结出了一系列有用的结论； 2．在特征级融合方面，提出了一种基于方向性信息测度和IHS变换的图像融合算法。实验表明该算法产生的光谱畸变很小，并且有良好的抗噪性，非常适合处理多光谱图像的融合； 3．在多分辨率图像融合算法中，提出了基于小波变换的自适应图像融合算法(DWT_EI)。与传统算法比较，该算法在提高图像信息含量方面表现得最好。同时它的运算也相对简单，并且和原始图像的相关性也很好(即光谱畸变小)，是一种非常好的融合算法； 4．归纳并给出了基于信息量的评价、基于统计特性的评价、基于相关性的评价和基于梯度值的评价四类十项融合结果评价指标。这些指标被用于对融合实验结果的实际评价中，使得对算法的评价从定性到定量两方面都有了一定的评价标准； 5．不仅从实际应用角度验证了图像融合技术能够增加图像信息含量、提高图像分割、分类和识别的有效性这一结论，而且从理论角度出发，进一步探讨了图像融合中的识别与决策问题。同时对图像融合中的可靠性与容错性问题也进行了简要分析； 6．除多源遥感影像作为实验数据以外，本文还选用了一组多聚焦可见光图片来验证和评价各种算法的融合效果，从而使得各种算法的有效性及优劣性更加直观。
[Abstract]:As an important branch in the field of image fusion, multi-source remote sensing image fusion studies how to synthetically utilize the image information obtained by different aerial remote sensing sensors to generate new data. In order to obtain a more comprehensive, objective and essential understanding of the same thing or object, remote sensing image fusion has become an indispensable technology in the field of image processing and image information understanding. And in many military and civilian areas have important applications. In this paper, image processing and modern signal processing technology are used in a variety of means. The fusion methods of multi-source remote sensing images at different levels are studied. The image resolution is improved by fusion of image data obtained by different aerial remote sensing sensors. The accuracy and confidence of image analysis results, and finally improve the effectiveness of automatic detection and recognition of specific aeronautical targets. The main contents and work of this paper can be summarized as follows: 1. The basic concepts and principles of image fusion are introduced, and then various traditional image fusion methods are systematically analyzed, and the characteristics and performance of these methods are compared in detail through experiments. A series of useful conclusions are summarized. 2. In the aspect of feature level fusion, an image fusion algorithm based on directional information measure and IHS transform is proposed. The experimental results show that the spectral distortion is very small and the algorithm has good noise resistance. It is very suitable for multispectral image fusion. 3. In the multi-resolution image fusion algorithm, an adaptive image fusion algorithm based on wavelet transform is proposed, which is compared with the traditional algorithm. The algorithm has the best performance in improving the information content of the image. At the same time, its operation is relatively simple, and the correlation with the original image is very good (that is, the spectral distortion is small, it is a very good fusion algorithm; 4. The evaluation based on information quantity and statistical characteristic is summarized and given. Evaluation based on correlation and evaluation based on gradient value. These indicators are used in the actual evaluation of fusion experiment results. Make the evaluation of the algorithm from qualitative and quantitative aspects have a certain evaluation criteria; 5. The conclusion that image fusion technology can increase the content of image information, improve the validity of image segmentation, classification and recognition is verified from the practical application. The problem of recognition and decision in image fusion is discussed, and the problems of reliability and fault tolerance in image fusion are also analyzed briefly. 6. In addition to multi-source remote sensing images as experimental data, this paper also selected a group of multi-focused visible light images to verify and evaluate the fusion effect of various algorithms. In order to make the effectiveness of various algorithms and advantages and disadvantages more intuitive.
【学位授予单位】：西北工业大学
【学位级别】：硕士
【学位授予年份】：2004
【分类号】：TP751

【相似文献】