FFTに基づくロバストな画像のレジストレーション手法の設計

抄録

Image registration plays a very important role in a camera-based remote sensing system. This paper discusses a 2D fast Fourier transform (FFT)-based method, which is well known for its easy implementation, accuracy, and robustness. The problem with the FFT-based method is that its robustness and accuracy highly depend on some preprocessing such as the design of a bandpass filter and subpixel estimation. This paper provided solutions for its algorithm design and tuning issues. Finally, the evaluation based on the image dataset and the actual images demonstrated its effectiveness.