Automatic coregistration of three-dimensional building models with image features

The aim of this article is to investigate methods for\ud the automatic extraction of the infrared (IR) textures\ud for the roofs and facades of existing building models.\ud We focus on the correction of the measured exterior\ud orientation parameters of the IR camera mounted on\ud a mobile platform. The developed method is based on\ud point-to-point matching of the features extracted from\ud IR images with a wire-frame building model. Firstly,\ud the extraction of different feature types is studied on\ud a sample IR image; Förstner and intersection points\ud are chosen for a representation of the image features.\ud Secondly, the three-dimensional (3D) building model\ud is projected into each frame of the IR video sequence\ud using orientation parameters; only coarse exterior\ud orientation parameters are known. Then the automatic\ud co-registration of a 3D building model projection into\ud the image sequence with image features is carried out.\ud The matching of a model and extracted features is\ud applied iteratively, and exterior orientation parameters\ud are adjusted with least square adjustment. The method\ud is tested on a dataset of a dense urban area. Finally,\ud an evaluation of the developed method is presented\ud with five quality parameters, i.e. efficiency of the\ud method, completeness and correctness of matching\ud and extraction