IMAGE PROCESSING, PATTERN RECOGNITION

Information technology for digital terrain model reconstruction from stereo images

V. A. Fursov^ab, E. V. Goshin<sup>ab

a</sup> Samara State Aerospace University
^b Image Processing Systems Institute, Russian Academy of Sciences

Abstract: This paper considers an information technology of digital terrain model (DTM) reconstruction from the pair of stereo images. For the case when the camera parameters are known, image matching algorithms based on epipolar constraints are thoroughly considered. In the matching we use weight coefficients as a penalty function for distance from the matching point to the epipolar line. Implementation of the technology when cameras’ parameters are not known is also considered. At the initial stage the problem of identification of the fundamental matrix from the corresponding points is solved. The major advantage of the proposed technology is the lack of the image rectification stage. This improves the reliability of the matching. The results of the proposed technology application for the DTM reconstruction from remote sensing stereo images are given.

Keywords: stereoimages, image matching, digital terrain model, disparity map, conformed identification, fundamental matrix, epipolar geometry.

Received: 06.03.2014