A New Structural Matching Method Based on Linear Features for High Resolution Satellite Images

  Along with commercial accessibility of high resolution satellite images in recent decades, the issue of extracting accurate 3D spatial information in many fields became the centre of attention and applications related to photogrammetry and remote sensing has increased. To extract such information, the images should be geo-referenced. The procedure of georeferencing is done in four main steps of extracting some control information such as point, line or areal features, matching, transformation function estimation and finally resampling. Among different control features, the lines are more considered due to their unique characteristics such as easier procedure of automatically extracting and matching as well as abundance of linear features in satellite images especially in urban areas. To reach an automatic georeferencing procedure, it is inevitable to automate the matching step as one of the most challenging process especially in heterogonous spaces such as image and map spaces. So, in this paper, the automatic matching procedure is performed and assessed using a new structural linear feature-based matching method with no need to any initial information. This method is based on using the most inherent conceivable information of the extracted features. The purpose of this paper is to find all possible match-lines through finding the correspondence of two specific patterns. The proposed method is done in three main phases of high quality pattern selection, matching as well as the final-phase. Additionally, new concepts of mathematically-generated-lines which are produced by extension and intersection of line-segments in two spaces as well as mathematically-generated-points which are the key-points of that lines are introduced and used to find the match-lines. In this paper, the impact of different numbers of stopping-criteria as well as one of the thresholds is studied experimentally. The results show the high potential of the proposed method to find more than 80 percentages of match-lines with 100 percentages of accuracy and reliability in a low computational time.