宽幅全色与多光谱遥感图像的高保真融合技术

项目名称： 宽幅全色与多光谱遥感图像的高保真融合技术

项目编号： No.61262036

项目类型： 地区科学基金项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 胡蕾

作者单位： 江西师范大学

项目金额： 45万元

中文摘要： 高分辨率全色遥感图像和低分辨率多光谱遥感图像融合能获得高分辨率多光谱遥感图像，对于准确分析环境污染、违规用地等具有重要作用。精确配准是融合的基础，而目前提出的配准方法主要用于小幅图像的配准，融合处理也存在空间细节损失与光谱信息失真问题，因此宽幅畸变遥感图像的精确配准与高保真融合是当前的难题之一。本项目在粗配准的基础上，通过合适的分块处理，结合地物特性均匀准确地选取图像块的特征点，并在空间约束下准确快速匹配特征点，实现图像块的准确配准，并有效利用评价手段，对未精确配准的局部区域重新配准，从而达到全图精确配准。在此基础上，项目结合低频系数及高频系数的局部特性，研究基于非下采样的Contourlet变换的自适应加权融合准则，使得融合图像清晰度好、光谱保持度高，实现宽幅全色和多光谱遥感图像的高保真融合。

中文关键词： 全色遥感图像；多光谱遥感图像；图像配准；图像融合；

英文摘要： The remote sensing images fusion aiming at high spatial resolution panchromatic images and low spatial resolution multispectral images usually produce fused images with high spatial and spectral resolutions, which can be used to accurately detect and identify the environmental pollutions and illegal land occupation. The accurate registration is the foundation of image fusions. However, existing registration methods only fit for small images, and existing fusion processing cannot avoid spatial detail information loss and spectrum distortion. Therefore, the accurate registration and high-fidelity fusion for large distorted remote sensing images become the puzzle. In this project, images will be divided into small blocks, from which the key feature points are extracted according to the topographical properties and matched by spatial constraints efficiently. Registration evaluation is adopted to identify the local unregistered areas, which are to be re-registered, thereby the whole image could be matched with high precision. Local characteristics for low and high frequency coefficient would be taken into account and self-adaptive rules based on nonsubsampled Contourlet transform are proposed to ensure that the fusion image would be of high clarity and spectral maintenance. The project would provide a high-fidelity f

英文关键词： Panchromatic remote sensing image；Multi-spectral remote sensing image；Image registration；Image fusion；