Automated DEM Extraction Process Applying Denoised Stereo Images and Optimal Number of Tie Points
Year: 2023
Type: preprint
Abstract: Digital Elevation Model (DEM) has received more attention nowadays attributed to its ability to learn the topography of the Earth's surface. A DEM is acknowledged as a basic geospatial dataset requirement for many environmental and geographical applications. In this study, a high resolution and high accuracy DEM is created using high spatial resolution stereo imagery of the region captured by the image acquisition system of the Cartosat-1 sensor. This research proposed a study to investigate the impact of Noise Reduction Filters (NRFs) on the accuracy of the derived DEM under different topographical circumstances. The most suitable NRFs for stereo imagery of the Cartosat-1 sensor are selected by comparing popular conventional and adaptive spatial domain filters. The DEM is extracted from the denoised stereo imagery of the Cartosat-1 sensor using the Automated Tie Point (ATP) selection method. The selection and positioning of the ATPs on the stereo imageries have a considerable impact on the accuracy and quality of the DEM. Additionally, the current work provides information on the optimum number of ATPs on stereo imageries in the image matching process for various topographic situations. The accuracy of the extracted DEM is evaluated quantitatively using a variety of statistical measures.
Source: Research Square
Institution: COEP Technological University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4366998713