A SAR image fast stitching algorithm based on machine learning

Abstract

Aiming at the splicing problem of Synthetic Aperture Radar (SAR) image, an improved algorithm for SURF is proposed to realize the fast splicing of SAR image. The SURF feature descriptor has scale invariance and rotation invariance, and has strong robustness to light intensity and affine transmission variation. The improved algorithm uses machine learning methods to build a binary classifier that identifies the key feature points in the SURF extracted feature points and eliminates the key feature points. In addition, the relief-F algorithm is used to reduce the dimensionality of the improved SURF descriptor to complete image registration. In the image fusion stage, a weighted fusion algorithm with a threshold is used to achieve seamless image mosaic. Experimental results show that the improved algorithm has strong real-time performance and robustness, and improves the efficiency of image registration. It can accurately mosaic multiple SAR images.