Automatic facade texture generation from helicopter dual-camera oblique imagery

The applications of 3D location-based service are getting popular in our daily life. In which, the photo-realistic city model is one of the most important geo-information infrastructure. Therefore, a cost-effective way to create photo-realistic city model has becoming an important research topic in the field of photogrammetry and computer vision. Conventionally, in close-range photograph the building façade would be easily occluded by ground objects and cost lots of time in texture mapping. For airborne vertical imaging system, it is designed for topographic mapping and the façade texture distortion is very serious thus it is improper for façade texture generation. In this research, we propose a helicopter dual-camera oblique imaging system for automatic façade texture generation provided that the camera's exterior/interior orientation parameters and 3D polyhedral building models are existed. The developed automatic texture generation algorithm includes three major steps. They are (i) filtering out the image which the target façade is not within the range of image frame or doesn't face the camera, (ii) excluding the image which has serious perspective distortion, and (iii) selecting out the image which has the best quality from the remaining images. Some case studies will be demonstrated in the final paper including the comparisons with manual texture generation through close-range photograph.