Performance Evaluation of Local Detectors in the Presence of Noise for Multi-Sensor Remote Sensing Image Matching

Automatic, efficient, accurate, and stable image matching is one of the most critical issues in remote sensing, photogrammetry, and machine vision. In recent decades, various algorithms have been proposed based on the feature-based framework, which concentrates on detecting and describing local features. Understanding the characteristics of different matching algorithms in various applications increases the potential of successful matching in a given application. Numerous studies have evaluated and analyzed many of these algorithms in various applications. However, performance evaluation of image matching methods in multi-sensor images, especially optical-radar and noisy images, is limited. This research will evaluate the performance of the state-of-the-art- detectors, including SURF, KAZE, SIFT, PC, FAST, and Harris detectors for multi-sensor image matching. Moreover, we integrated the employed detectors with the uniform competency algorithm to identify the most reliable features with uniform distribution. Next, we employed a scale-invariant version of the HOSS descriptor to describe extracted features. The results show the superiority of the KAZE detector in the presence of noise and various geometric and radiometric distortions.