A Study for Efficient Methods of System Calibration between Optical and Range Sensors

Abstract. Recently, interests in 3D indoor modeling and positioning have been growing. Data fusion by using different sensors data is one of the 3D model producing methods. For a data fusion between two kinds of sensors, precise system calibration is essential. If relative geometric location of each sensor can be accurately measured with a system-calibration, it is possible to locate a pixel that corresponds to the same object in two different images, and thus, produce a more precise data-fusion. Purpose of this study is finding more efficient method of system calibration between optical and range sensor. For this purpose, experiment was designed by considering following variables, i) system calibration method, ii) testbed type, iii) and distance data(whether use it or not). So, In this study, test-bed for system calibration was designed by considering the characteristics of sensors. Also, precise simulation was done to find efficient method of system calibration, and its results were reflected in real experiment. Results of simulation show that the bundle adjustment method is more efficient than single photo resection in system calibration between range and optical sensors. And the most efficient case was when using i) the bundle adjustment with ii) the simulated data set which were obtained between 2m to 4m away from the test-bed. These results of simulation were reflected in real system calibration. Finally, real system calibration were performed and its results were compared to results of simulation. And accuracy of system calibration was evaluated by producing fusion data between range and optical sensors.