FFT Architecture for Motion Estimation using Phase Correlation

Fast Fourier Transform(FFT) is one of the widely used algorithms in almost all signal processing applications in various contexts. Efficient hardware architectures to implement FFT tailored for a typical application scenario is always challenging and requires a lot of analysis. In this paper an architecture for a 64x64 FFT is proposed for a real time video motion estimation application. Global motion estimation for video data employing phase correlation has many advantages such as its ability to estimate larger motions with low computational complexity in addition to spectrum insensitivity and immunity to noise and illumination variation. The major computational block that decides the efficiency of the phase correlation algorithm is FFT. The proposed FFT architecture is designed to meet the real-time requirements of a video stabilization module employing phase correlation for the motion estimation where the videos are assumed to be captured at a rate of 30 frames per second or above. The proposed architecture was implemented on a Kintex-7 FPGA board and met the required specifications.