Kernel correlation filtering algorithm based on fractional Fourier transform

Robust tracking is an important and challenging task in the field of computer vision. Kernel correlation filtering is one of the most popular tracking algorithms. Diversity and complexity of tracking objects usually lead to degradation of its tracking performance. To improve the tracking performance, we propose a kernel correlation filtering based on fractional Fourier transform. We replace Fourier transform by fractional Fourier transform. The Peak-to-Sidelobe Ratio of the tracking object in the response map becomes higher, which benefits the tracking performance. Extensive experiments on the dataset OTB100 show the advantage of our proposed method.