Adaptation of TIN-based Ortho-Mosaicking for DEM Error Mitigation

초록

Ortho-mosaicking is an essential process for generating large-area image products from UAV imagery. Conventional LiDAR-based DSMs provide high accuracy but require expensive equipment and significant processing time. Stereo-based DSMs are relatively efficient yet they are still computationally costly. To address these limitations, this study proposes a TIN-based ortho-mosaicking method with adaptive grid generation using, a nationwide low resolution DEM. The method first constructs a rough DSM from sparse PCD obtained through tiepoints and bundle adjustment. It then analyzes terrain gradients and elevation to detect error-prone regions. PCD is first extracted from the DEM grid across the study area. Points located in the identified error-prone regions are adaptively expanded or merged to stabilize the corresponding grid cells. Finally, the grid corners are back-projected onto the image plane. The patches are then assembled using homography to form a mosaic while maintaining geometric consistency. Experimental results demonstrate that the proposed approach enables fast and seamless ortho-mosaicking, particularly in areas containing high-rise buildings. © ACRS 2025.All rights reserved.

키워드