Automatic Co-Registration of Aerial and Satellite Imagery using optimized minimum distance thresholds for matching
Prachi Goel, Pratyush Dayal, Tushar Kanti Patra, Charu Gupta · 2022 International Mobile and Embedded Technology Conference (MECON) · 2022
Imagery is captured by different satellites, Unmanned Aerial Vehicles (UAV’s), equipment’s or sensors, from different angles and at different time frames. Difference in nature of data, orientations or settings of satellites, equipment’s or sensors may lead to introduction of errors in the imagery. Due to various such errors it becomes difficult for the imagery to be deployed in various remote sensing applications such as fusion of images, change detection panorama creation etc. Hence, we need image registration which geometrically aligns images belonging to the same area. This paper provides a solution for removal of the registration errors automatically for multi temporal and multi resolution imagery, hence facilitating Automatic Image co-registration. Feature extraction utilizes the SURF (Speeded-Up Robust Features). Feature matching is done using Fast Library for Approximate Nearest Neighbours (FLANN) matcher. The work focuses on solving the problems faced in matching of valid key point descriptors of the sensed and reference imagery, particularly the panchromatic satellite imagery. Multiple scenarios for minimum distance thresholds have been visualized for eliminating the bad matches. Very high accuracy of less than a meter in most cases has been achieved for satellite imagery, using the above method of image registration.