Development of algorithms for feature matching in the method for reconstucting high-precision digital surface models

Abstract

The method for reconstructing high-precision digital surface models baseed on stereo matching of point and contour features is presented. Methods for extracting points and contours in images, and their subsequent matching are considered. Additionally the specific items of the proposed method are underlined from the point of view of using these image features for solving the final problem of detection features of the surface relief. Special attention is paid to the transition from local stereo matching to the global one. The necessity of this transition for successful solving the problem of a qualitative relief model with a minimal number of outliers is shown. The advantages of the developed method are demonstrated by the example of reconstructing a surface relief based on two stereo pairs of aerial images of different resolutions.