The Landscape of Dense Image Matching Algorithms

Both improvements in camera technology and the rise of new matching approaches triggered the development of suitable software tools for image based 3D reconstruction by research groups and vendors of photogrammetric software. Based on dense pixel-wise matching, the photogrammetric generation of dense 3D point clouds and Digital Surface Models from highly overlapping aerial images has become feasible. In order to evaluate the quality of these matching algorithms in terms of accuracy and reliability, the European Spatial Data Research Organisation (EuroSDR) started a benchmark on image based DSM generation in February 2013. This test is based on two representative image blocks, which were processed by different groups with different software systems. The results provided from the different groups give a profound insight to the landscape of dense matching algorithms and are used within the paper to evaluate the potential of image based photogrammetric data collection.