Author(s)

Liyu Yan, Qianyan Meng

Corresponding Author

Qianyan Meng

Abstract

Aiming at the challenges in SAR image registration technology, this article aims to construct a SAR image registration algorithm based on machine learning to improve the accuracy and efficiency of registration. In order to achieve this goal, this article firstly analyzes the characteristics of SAR images and summarizes the advantages and disadvantages of traditional SAR image registration algorithms. On this basis, a framework of registration algorithm based on machine learning is designed, which includes key steps such as feature extraction and selection, machine learning model selection and training, and registration parameter optimization. Experimental results show that the proposed algorithm is robust and adaptive in SAR image registration tasks with different scenes and resolutions. Moreover, compared with RF and decision tree algorithms, this algorithm can achieve the spatial alignment of SAR images more accurately. Therefore, this article concludes that SAR image registration algorithm based on machine learning has broad application prospects and can bring substantial technical progress to the field of remote sensing image processing.

Keywords

SAR image registration; Machine learning; Feature extraction; Registration accuracy