目标网格编码支持的多源遥感影像地理匹配

A TGC-based Approach for Multi-source Remote Sensing Image Registration

为了解决多源遥感影像快速匹配问题,针对传统影像配准对控制点及基准影像依赖的问题,引入剖分网格体系建立目标网格特征编码,将网格区域内的地物目标空间分布与地理位置关联,实现多源遥感影像或数据的地理匹配。在分析了图像配准和遥感影像匹配形式局限基础上,提出了目标网格编码的概念,重点对地物编码方案进行了设计,建立多层次基准编码。实际应用时从待匹配遥感影像中提取定位目标的空间分布信息,通过多尺度编码匹配定位过程可以将影像定位至相应网格,从而实现多源影像的相互匹配。基于本方法在江苏及四川进行了多源遥感影像匹配试验,结果表明该方法在信息有限情况下也能求得较高匹配精度,从而为影像匹配提供了新的思路与方法,也进一步拓展了剖分网格的应用。

In order to solve the problem of traditional rapid image registration reliance on control points and reference images,this paper introduces a split grid system to establish the Target Grid Coding(TGC).By this operation,the spatial distribution of ground objects in the grid area is associate with geographic location to achieve geographic matching of multi-source remote sensing images or data.We first propose the concept of TGC after analyzing image matching and then design and build multi-level TGC.In practice,the spatial distribution information of the object is extracted from the unmatched remote sensing images.After positioning images to the corresponding grid by multi-level TGC,the multisource images can be matched.To validate the usability of this method,experiments have been conducted in Sichuan and Jiangsu.The experimental results show that the proposed method achieves high accuracy with less information,which provides a new insight for image matching,and further expands the application of mesh gridding technique.