Feasibility Study for 1:1000 Scale Map Generation from various UAV images

초록

An UAV(Unmanned Aerial Vehicle) files at lower altitudes than airplanes or satellites and hence generate images at resolution. It opens a possibility for new applications such as 3D object modeling and large scale map generation at low cost. In this study, our goal is to evaluate the feasibility of UAV images for 1:1000 scale map generation by stereo plotting. We calculated orientation parameters of UAV images from various SW and analyzed the 1:1,000 mapping possibility by applying them a DPW (Digital Photogrammetric Workstation). For UAV processing SW, Pix4DMapper, PhotoScan, and 3D-UAV were used. Mapping accuracy was predicted by stereoscopic viewing capability and intermodel accuracy. Tiepoints were generated automatically from UAV images and used for testing. Stereoscopic viewing capability was tasted by the amount of Y-parallex after epipolar transformation at tiepoint locations. Inter-model accuracy was tested by the difference of object coordinates calculated at tiepoints observable in several stereo models. The analysis results showed that Y-parallex accuracy within 0 ~ 2 pixels was acquired for stable stereoscopic viewing and some stereo models produced by each software showed an accuracy suitable for 1:1,000 mapping.