A patch-based method for the evaluation of dense image matching quality

Abstract Airborne laser scanning and photogrammetry are two main techniques to obtain 3D data representing the object surface. Due to the high cost of laser scanning, we want to explore the potential of using point clouds derived by dense image matching (DIM), as effective alternatives to laser scanning data. We present a framework to evaluate point clouds from dense image matching and derived Digital Surface Models (DSM) based on automatically extracted sample patches. Dense matching errors and noise level are evaluated quantitatively at both the local level and whole block level. In order to demonstrate its usability, the proposed framework has been used for several example studies identifying the impact of various factors onto the DIM quality. One example study proves that the overall quality on smooth ground areas improves when oblique images are used in addition. This framework is then used to compare the dense matching quality on three different terrain types. In another application of the framework, a bias between the point cloud and the DSM generated from a photogrammetric workflow is identified. The framework is also used to reveal inhomogeneity in the distribution of the dense matching errors caused by overfitting the bundle network to ground control points.