Evaluation of invariant feature algorithms on FL-BSAR imagery registration

Imagery registration is an important application of FL-BSAR. Meanwhile, many start-of-art invariant feature image registration methods are proposed in computer vision world. However, few of them have been evaluated in SAR application. Our paper focuses on the evaluation of contemporise registration algorithms on FL-BSAR. According to our experiment, the best interest point detector is the Harris-Laplacian detector which detects corner-like structures. And both of PCA-SIFT and SIFT descriptors are fit for SAR image features. Our work shed some light on its encouraging effect on the image matching for FL-BSAR images using algorithms based on invariant feature.