高分辨图像二维FFT正/反变换实时处理方法及硬件实现*

高精度密集型数值计算和大规模数据缓存，是高分辨率图像二维FFT(快速傅里叶变换)实时实现中的主要难点。利用实信号傅里叶变换的周期对称性和频域数据的共轭对称性，提出了一种高效且易于硬件实现的二维FFT正/反变换的实时处理方法，将实值图像二维FFT中的一维FFT计算和存储需求缩减了近一半。在以4片TS201为计算核心的DSP处理平台上，使用该方法实现了二维FFT正/反变换和图像频域滤波。实验表明，无须片外存储，单片TS201可处理最大512×512像素的图像；该尺寸图像的正/反变换总处理时间为49.6 ms，

Real-time processing method of 2D-FFT/IFFT for high-resolution image and hardware implementation

High precision calculation of massive data and requirement for large-memory, are main difficulties in the real-time implementation of 2D-FFT(fast Fourier transformation) for high-resolution image. Periodic symmetry of the real FFT and conjugate symmetry of frequency data was utilized to improve the 2D-FFT/IFFT of image.This paper proposed a real-time proces-sing method of 2D-FFT/IFFT, which was fit for hardware realization and reduced about half of the 1D-FFT computation and memory demand. 2D-FFT/IFFT and image frequency filtering based on this method was implemented on the DSP platform, which carried ADSP-TS201 chips as the computation core. Experimental results indicate that, without off-chip memory, inputting 512×512 pixels image,the time required for 2D-FFT /IFFT is 49.6 ms with single TS201,and the processing frame rate is over 80 fps by four parallel TS201.The method is feasible and efficient, which can satisfy real-time demand in engineering.