Research on the registration and fusion algorithm of terahertz SAR images

The processing of terahertz (THZ) synthetic aperture radar (SAR) images holds significant application value in target characteristic analysis and intelligent recognition, rendering it a current research focal point. The article investigates image registration methods based on SIFT, SURF, and KAZE operators. An enhanced RANSAC algorithm incorporating forward preprocessing is developed to accomplish the registration and fusion of terahertz SAR images, addressing the research gap in high-noise and low-contrast terahertz image registration. The research findings suggest that the KAZE operator is better suited for terahertz SAR image registration. Additionally, this study conducted fusion on various terahertz images, demonstrating that image fusion can enhance information content and reduce noise. This research contributes to the application of terahertz SAR images in fields such as target characteristic analysis and recognition.