Development of quality indicator for stereoscopic plotting using orientation parameters of UAV images

Abstract

An unmanned aerial vehicle (UAV) acquires images at higher spatial resolution than airplanes or satellites. It opens a possibility for large scale digital map generation at low cost. Recently, studies are being conducted to generate large scale maps using UAVs. Most of researcher are using UAV commercial software such as Pix4D and PhotoScan. While they provide information such as camera calibration and bundle adjustment quality, they do not provide any quality indicators for a large scale digital map generation. In this paper, we develop a technique to indicate the quality to generate a large scale digital map by calculating the quality of stereoscopic plotting. We aim to utilize interior and exterior orientation parameters and tie-points derived from UAV images and to exclude the use of ground control points. For measuring this quality, we propose to use check-points to calculate Y-parallex and absolute-model accuracy among the adjacent stereo models. The Y-parallax affects the quality of stereo-viewing and hence the possibility of stereo-plotting using the images. The absolute model accuracy affects stereoscopic plotting accuracy for digital mapping. The absolute model accuracy can be calculated by the difference of object coordinates at check-points observable in several stereo models. We compared UAV images processed from various UAV SW with the quality of stereoscopic plotting by using orientation parameters. The analysis results showed that the rotary-wing produced the quality of stereoscopic plotting that was more accurate than the fixed-wing. By using this technique, we can conclude that large scale digital map accuracy can be predicted in advance. © 2020 40th Asian Conference on Remote Sensing, ACRS 2019: Progress of Remote Sensing Technology for Smart Future. All rights reserved.