Fast multiscale SAR images stitching based on improved APAP and OTSU algorithms
Chenyu Zhu, Xiaoyu Cong, Yubing Han, Weixing Sheng · IET conference proceedings. · 2024
In marine synthetic aperture radar (SAR) remote sensing images, the land and ships often only occupy part of the images, and the rest are ocean or coastal backgrounds. Existing SAR image stitching algorithms often directly detect the entire SAR image, which consumes a lot of computing resources. In this paper, an automatic image stitching algorithm based on the As-Projective- As-Possible (APAP) and Otsu algorithm is proposed. Firstly, a new method of image size alignment is proposed to make multiple images have the same size for subsequent processing steps. Secondly, a threshold segmentation method is used to detect the positions of land or ships and filter out most of the sea backgrounds to reduce the computational requirements. Thirdly, a phase correlation algorithm is used to calculate the phase correlation between different images for determining the optimal stitching order and the scale-invariant feature transform (SIFT) algorithm is used to detect feature points from the thresholded images. Finally, the APAP algorithm is employed to perform the actual stitching at the source SAR image level using the order and feature points measured before, using the previously measured order and feature points. Experimental results conducted on the SSDD dataset demonstrate that the performance of our method is close to that of the traditional stitching algorithm, but the number of feature points and time are reduced by 46.83% and 27.87% of the traditional method, respectively.