Real-time Sequence Images Registration based on Fast Least Squares Feature Point Tracking

Abstract

Abstract This study addresses the issues of diminished precision and sluggish computational pace observed in current sequence image registration algorithms. A sequence image registration algorithm based on adaptive feature point selection (AFPS) and fast least squares tracking (FLST) is proposed. The corner points are detected by the FAST algorithm. The feature points are selected according to the distribution and intensity of the corner points. The fast least squares tracking technique is employed to compute the displacement of feature points between successive images. The proposed feature point selecting approach and the tracking method substantially increased the efficiency and performance of the registration algorithm. The registration algorithm is evaluated by sequence images taken by drones. The test outcomes demonstrate that, in comparison to existing algorithms, the proposed algorithm significantly enhances both the accuracy and real-time capabilities of sequence image registration.