Data fusion for 3D thermal imaging using depth and stereo camera for robust self-localization

Infrared thermography is a common method of measuring surface temperature distributions. New approaches to the three-dimensional (3D) recording of these measurements have been developed in the last decade. A given 3D real-time imaging system [1] consisting of an infrared (IR) camera and a near-infrared (NIR) depth sensor is the object of investigation. The system uses an improved KinectFusion processing algorithm and its Iterative Closest Point (ICP) method for self-localization. The aim of this work is to reduce the impact of correspondence problems when scanning objects with few geometric features on 3D modeling. In order to achieve robust self-localization, a stereo camera is integrated into the system and its localization data is fused with the existing system data. By means of indoor and outdoor test scenarios the system shows its ability to bridge sections of ICP localization problems.