# -*- coding: utf-8 -*-
"""
Created on Sun Jul 8 21:05:51 2018
@author: Diko
"""
import numpy
def FFT_v1(Img,Wr):
if Img.shape[0]==2:
pic = numpy.zeros([2],dtype=complex)
pic = pic*(1+0j)
pic[0]=Img[0]+Img[1]*Wr[0]
pic[1]=Img[0]-Img[1]*Wr[0]
return pic
else:
pic = numpy.empty([Img.shape[0]],dtype=complex)
pic[0:Img.shape[0]//2] = FFT_v1(Img[::2],Wr[::2])+Wr*FFT_v1(Img[1::2],Wr[::2])
pic[Img.shape[0]//2:Img.shape[0]]=FFT_v1(Img[::2],Wr[::2])-Wr*FFT_v1(Img[1::2],Wr[::2])
return pic;
def FFT_1d(Img):
Wr = numpy.ones([Img.shape[0]//2])*[numpy.cos(2*numpy.pi*i/Img.shape[0])-1j*numpy.sin(2*numpy.pi*i/Img.shape[0]) for i in numpy.arange(Img.shape[0]/2)]
return FFT_v1(Img,Wr)
def FFT_2d(Img):
pic = numpy.zeros([Img.shape[0],Img.shape[1]],dtype=complex)
图像处理 之 二维快速傅里叶变换(FFT2)
最新推荐文章于 2024-07-05 21:05:15 发布
本文对比了numpy.fft.fft2()函数与自实现的二维快速傅里叶变换(FFT2)。尽管两者结果一致,但自实现版本在计算时间上翻倍,尤其是在处理大规模数据时,性能差距显著。文章探讨了优化FFT2效率的需求。
摘要由CSDN通过智能技术生成