Calibration of close-range thermal imagery for integration into 3D VR models

The appearance of terrestrial laser scanners (TLS) has provided a new data source of geometric information. Several TLS allow to be equipped by a calibrated camera, whose images may be directly mapped on the DSM as photo-texture. Here a further improvement is proposed, i.e. the integration of thermal imagery into the 3D model in order to acquire knowledge about internal stratigraphy of walls, floors, ceilings and other ancient structures. Obviously, a fundamental pre-requisite to obtain this task is the calibration of thermal sensor and the orientation of each image into the object reference system of the TLS data. Unfortunately, due to the poor radiometric and geometric quality of themal images, their integration into the TLS 3D model is a complex task; moreover, looking for control points which could be measured on both 3D model and thermal image is not trivial. This leads to the failure of methods performing calibration and orientation in a unique task, such as self-calibration approaches. Calibration has to be performed in laboratory. We have performed the calibration of a thermal camera NEC Thermotracer TH 7102 WX by means of a calibration dig and the computation of inner calibration in a bundle block l.s. adjustment. Data processing has been performed by using a low-cost photogrammetric commercial software.