Integration of Kinematic Parameters into Object-based Image Sequence Matching

Abstract Spatio-temporal data of objects are crucial for scientific research and technical analysis in various dynamic applications. Particularly relevant is the measurement of deformations by photogrammetric deformation analyses, which are carried out in a dense and non-intrusive manner. However, the observation of fast-moving phenomena, such as in wind tunnel experiments, is often affected by kinematics and specific experimental circumstances, which can reduce the quality of the results. In addition, acquired sensor data, which provide information about rotational velocities and accelerations, are often not sufficiently integrated into the photogrammetric analysis. This paper presents a generic framework to integrate kinematic information into the photogrammetric deformation analysis. Based on this, we extend object-based image matching to a sub-pixel accurate spatio-temporal matching method, integrating kinematic data as prior knowledge and simultaneously determining unknown (geometric and kinematic) parameters. We applied the approach in wind tunnel tests with a miniaturized wind turbine, where additional kinematic information about the rotation was acquired. The results demonstrate that the new method generates highly accurate and dense information about blade deformation using photogrammetric and kinematic data. The approach is applicable to a variety of dynamic photogrammetric applications.