摘要
以2为基(即N=2m)的FFT算法在N不等于2的幂2m时,需要用补0的方法将x(n)延长为2m,这种补0的方法对于某些对图像直接在频域进行的处理会产生影响。本文给出当图像大小N为偶数但不为2的幂时的不需补0的结合DFT和FFT的实用傅里叶变换算法。实验结果表明对于常用的图像存储格式,该算法速度接近FFT,并且不会对图像频谱进行的直接处理带来负面影响,在图像处理方面和快速傅里叶变换相比具有一定的优势。
Using FFT needs the size of the image being some power of 2.Otherwis e,lets of zero need to be replenished to meet it.So it will influence some proce ssing of frequency spectrum.This paper proposed a useful Fourier transform algor ithm combining FFT with DFT.It only needs the size of image being even.As concer ning the common image format,the experiments shows that it’s computational spee d is almost same as FFT when it without any bad effect in processing of frequenc y spectrum.So it has some superiority than FFT in image processing.
出处
《计算机应用与软件》
CSCD
北大核心
2005年第5期38-39,116,共3页
Computer Applications and Software
基金
湖北省自然科学基金项目(No.2004ABA068)