Matching of Aerospace Photographs with the use of Local Features

Abstract

Matching of digital images is very challenging computer vision problem. The aim of investigation was developing of algorithms for matching real aerial and cosmic photographs. In the proposed methods, images are described locally by scale and rotation invariant descriptors. Reliability and high accuracy of the algorithms has been achieved by combining dense keypoint detector and robust descriptor with complex procedure of outlier elimination. The algorithms are capable of making correct decisions when number of local mismatches is more than 99%.