High-precision multimodal remote sensing image matching based on Log-Gabor filtering

Multimodal remote sensing image matching is challenging due to nonlinear radiative differences. This paper proposed an effective high-precision matching method based on Log-Gabor filtering, which including two stages: coarse matching and fine matching. In the coarse matching stage, key feature points are detected and initially matched using a phase congruency model and log-Gabor filters; in the fine matching stage, based on the estimated affine transformation model, the matching effect is further optimized using log-Gabor based template matching and three-dimensional phase correlation techniques. The experimental comparison of the algorithm studied in this paper with other image recognition algorithms shows that this algorithm has a significant performance advantage in terms of the number of correct matches and matching precision.