Research On Image Matching Based On Marked Point

In the computer-aided close-range photogrammetry, a group of photos from differert directions with the camera, in the computer mark the target point from the group of photos, obtain two-dimensional coordinates of each marker, after a series of algorithms, two-dimensional coordinates into three-dimensional, and complete the whole measurement process, and the point and point matching is an essential setp in this series of algorithms.The matching, that find the makers corresponding to the same objiect in two or more images. This image point matching method will be realized in three steps. Firstly, the marked point is to be locationed and identified. Currently, there is more and more identification methods, this article describes an improve identification method in detail. Each step has been discussed in detail, including image binarization, edge detection and scanning. Experimental results show that the method can efficiently identify the identifying point.Secondly, the coded point is identified and it is to be decoded and matched. This article describes the design steps of the encoding code point in detail, and uniqueness based on coding theory, achieve to decode and match. With the first step to prepare, this step change calculation is simple, easy to show. Experimental results show that the method is feasible, matching high, it is worth drawing.Finally, there are a lot of study on the match of non-coded point. By comparing several non-coded point matching method, proposed an improve method for the match. Due to non-coded point lack of the characteristics of coded point shape and large numbers, we use method of combining coding and non-coding point, and the introduction of similarity and Geometry theorems in the same photo, completed the match of non-coding point. The results show that the method has a good match results and certain value.