Research On Registration Between Remote Sensing Image And Vector Data Based On Straight Line Features

The registration between remote sensing image and vector data has extensive application inthe field of photogrammetry and remote sensing, and it is the base and premise of updatinggeography data, change detection, information fusion etc. Since the different character betweenremote sensing image and vector data and obvious rotational and scale change, registrationbetween them is much harder than registration between data of same character, for example,image registration. In this paper, straight line feature is selected as matching primitive andresearch is carried out which is aimed at the registration between image and vector data based onstraight line features. The main finished work and innovations are as follows:1. This article makes an introduction of the research background and significance. Thestatus quo and development tendency of the registration between the image and vector data aresummarized.2. An automatically total registration approach is proposed which is based on straight linefeatures. In this approach, the similarity measure is established according to the distance which iszero between reference line from vector data and corresponding line in the image in the samereference coordinate system. A modified Hough transform is adopted to simultaneously estimatethe exterior orientation parameters and obtain correspondence between registration primitives.This approach is insensitive to rotational and scale transformation. Experimental results provedthe above proposed approach is valid.3. The elements are researched which have influence on the precision of registrationbetween remote sensing image and vector data. The position and direction of control straightlines have influence on the precision of6exterior orientation elements, and the influences areanalyzed academically. The reliability of precision analytical results is proved by making use ofsimulated images. Experimental results proved the precision of the exterior orientation elementscalculated by line features are almost the same as the precision of elements using traditionalmethod.4. A registration method between remote sensing image and vector data, which is based on2D multi-scale line template, is put forward. The initial projective result of vector data straightlines filterd in the image is regarded as prior information. And in the remote sensing image,2Dmulti-scale line template is utilized to search line features in order to establish the correlationbetween matched primitives. The method used to calculate the exterior orientation parameters,which is based on the condition that the straight line on the image space and corresponding lineon the object space are coplanar, is utilized to calculate parameters. Experimental result provedthe method is feasible.