An acceleration method for correlation-based high-speed object tracking

An acceleration method for correlation-based high-speed object tracking

Blog Article

We propose an acceleration method for 6 Piece Queen Storage Bedroom correlation-based object tracking.Correlation-based tracking methods involve an inverse Fourier transform, which is a bottleneck in acceleration.We exploit a trait of high-speed vision to accelerate this computation: namely, we assume that displacements between consecutive frames do not change dramatically within a short frame acquisition interval.

By limiting a small region where the displacement is assumed to be, we propose to reduce computational cost in the inverse Fourier transform.We implemented the proposed method in phase-only correlation and conducted experiments Spawn using both simulated and real data.We achieved a computation speed around five-times faster than the conventional method without sacrificing accuracy.

The proposed method provides as a useful building block in accelerating correlation-based object tracking.