2D Discrete Fourier Transform with Simultaneous Edge Artifact Removal for Real-Time Applications

摘要

Two-Dimensional (2D) Discrete Fourier Transform (DFT) is a basic andcomputationally intensive algorithm, with a vast variety of applications. 2Dimages are, in general, non-periodic, but are assumed to be periodic whilecalculating their DFTs. This leads to cross-shaped artifacts in the frequencydomain due to spectral leakage. These artifacts can have critical consequencesif the DFTs are being used for further processing. In this paper we present anovel FPGA-based design to calculate high-throughput 2D DFTs with simultaneousedge artifact removal. Standard approaches for removing these artifacts usingapodization functions or mirroring, either involve removing criticalfrequencies or a surge in computation by increasing image size. We use aperiodic-plus-smooth decomposition based artifact removal algorithm optimizedfor FPGA implementation, while still achieving real-time ($\ge$23 frames persecond) performance for a 512$\times$512 size image stream. Our optimizationapproach leads to a significant decrease in external memory utilization therebyavoiding memory conflicts and simplifies the design. We have tested our designon a PXIe based Xilinx Kintex 7 FPGA system communicating with a host PC whichgives us the advantage to further expand the design for industrialapplications.