Fast Fourier Transform & Assignment 2 Yong-Fong Lin Visual Communications Lab Department of Communication Engineering National Central University Chungli, Taiwan Oct. 4, 2007
Outline Assignment Description Discrete Fourier Transform Review Introduction to Fast Fourier Transform(FFT) Example:4 Point FFT & FFT Butterfly Experimental Result Notice Reference
Assignment Description Part 1 Transform an gray image from spatial domain into frequency domain using Fast Fourier Transform(FFT). And show the spectrum of the image. Part 2 Rotate the gray image by , and show the spectrum of the rotated image.
Discrete Fourier Transform(DFT) Review One-Dimensional DFT for u = 0,1,2, …. ,M-1
Fast Fourier Transform (1/5) for u = 0, 1, 2, … ,M-1 If (n N) then M can be expressed as M=2K (K N) (Keep this in mind !!)
Fast Fourier Transform (2/5) (1) for u = 0,1, … , K-1
Fast Fourier Transform (3/5) By (1)
Fast Fourier Transform (4/5) (2) for u = K, K+1, … , 2K-1
Fast Fourier Transform (5/5) Conclusion: We can perform the DFT by using FFT as follow steps for the first K points (u = 0 ~ K-1) for the rest K points (u = K ~ 2K-1) the rest K points doesn’t need extra computation, it can just be obtained by the result of first K points.
Example:4 Point FFT & FFT Butterfly Consider a sequence:f(x) for x= 0 ~ 3 need to be transformed. The transformed result is F(u) for u = 0 ~ 3. F(0) F(2) F(1) F(3) F(0) F(2) F(1) F(3) F(0) F(1) F(2) F(3) + -
Experimental Result (1/2) Original image and the corresponding spectrum
Experimental Result (2/2) Rotated image and the corresponding spectrum
Notice Don’t forget to multiply According to the property of “ Separability ” , we can perform two-dimensional DFT by using one-dimensional DFT. (p197) is just To deal with complex number , we must have 2 buffer. One for the real part The other for the imaginary part
Reference Rafael C. Gonzalez , Richard E. Woods , “Digital Image Processing, ” second edition , pp.208-213