Abstract
For geostationary meteorological satellite (GSMS) remote sensing image registration, high computational cost and matching error are the two main challenging problems. To address these issues, this paper proposes a novel algorithm named slope-restricted multi-scale feature matching. In multi-scale feature matching, images are subsampled to different scales. From a small scale to a large scale, the offsets between the matched pairs are used to narrow the searching area of feature matching for the next larger scale. Thus, the feature matching is accomplished from coarse to fine, which will make the matching process more accurate and reduce errors. To enhance the matching performance, the outliers in the matched pairs are rectified by using slope-restricted rectification, which is based on local geometric similarity. Compared with other algorithms, the experimental results show that our proposed method is more accurate and efficient.
Highlights
Image registration is an inherent part of remote sensing image processing, since it has been widely applied in image fusion [1,2], image mosaic [3], change detection [4,5] and 3D reconstruction [6,7].In recent years, geostationary meteorological satellites with higher spatial resolution and higher accuracy have been invented, study on registration for lower accuracy geostationary meteorological satellite (GSMS) images is still needed, especially for historical data re-processing
To measure the matching performance, the ground truth is labeled manually: for each landmark in the landmark image, we accurately mark its corresponding point in the GSMS image, except the landmark covered by clouds
This paper focuses on reducing the time cost and matching error
Summary
Image registration is an inherent part of remote sensing image processing, since it has been widely applied in image fusion [1,2], image mosaic [3], change detection [4,5] and 3D reconstruction [6,7]. Geostationary meteorological satellites with higher spatial resolution and higher accuracy have been invented, study on registration for lower accuracy geostationary meteorological satellite (GSMS) images is still needed, especially for historical data re-processing (e.g., climate change analysis). A number of methods have been proposed for remote sensing image registration. These methods can be coarsely classified into intensity-based and feature-based methods. Compared with intensity-based algorithms, feature-based algorithms have a good ability to handle image distortions and illumination changes, and reduce the computational cost. We focus on the research of feature-based matching methods
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
