Automatic bundle adjustment of thermal infrared images

Abstract

Thermal infrared images show the temperature change of sensed scenes. Therefore, thermal infrared camera can sense some important information that optical images cannot do, especially in night. Thus, thermal infrared images are used not only in the domain of surveying, but also in the disaster prevention, and the environment monitoring. Before thermal infrared images can be used for advanced analysis based on the 3-D reference coordinate system, e.g. TWD97 coordinate system in Taiwan, the precise position and orientation of those images should be determined by bundle adjustment after tie points are extracted either manually or automatically. However, the manual measurement of tie points is much more cost-consuming due to the point features of thermal infrared images cannot correspond to the actual and visible point location of the scene. Therefore, in this study improved SIFT(Scale Invariant Feature Transform) Algorithm developed by Chen and Chio (2011) will be used to extract the tie points automatically, and bundle adjustment will be performed after the thermal infrared camera parameters are calibrated. This study will also discuss the relevant problems and give some suggestions from the test results.