thesis.pdf

Accurate 3D landscape models of cities or mountains have wide applications in mission

planning, navigation, geological studies, etc. Lidar scanning using drones can provide high

accuracy 3D landscape models, but the data is more expensive to collect as the area of

each scan is limited. Thanks to recent maturation of Very-High-Resolution (VHR) optical

imaging on satellites, people nowadays have access to stereo images that are collected on a

much larger area than Lidar scanning. My research addresses unique challenges in satellite

stereo, including stereo rectification with pushbroom sensors, dense stereo matching using

image pairs with varied appearance, e.g. sun angles and surface plantation, and rasterized

digital surface model (DSM) generation. The key contributions include the Continuous 3D-

Label Semi-Global Matching (CoSGM) and a large scale dataset for satellite stereo processing

and DSM evaluation.