Image Filtering in the Frequency Domain.pdfVIP

1 Image Filtering in the Frequency Domain PROJECT #3 ECE618 Due: 10/14 Handed: 10/14 Abstract: In this project, image filtering in frequency domain is studied. The main two parts of the project are mainly concerned with finding and displaying the Fourier spectrum of an image. For better visualization of the frequency, the Fourier spectrum is shifted to the center when be displayed. The average value of the image is found by conventional averaging methods, and it is compared to the result obtained directly from the spectrum. In the third part an image is corrupted by a sinusoidal noise and a Butterworth filter is designed to clean the noised image. The filter parameters are determined from spectrum analysis. 2 )(2 ),(),( N yv M xuj oo oo eyxfvvuuF + =?? π )()22 (2 )1)(,(),()2/,2/( yx yxj yxfeyxfNvMuF + + ?==?? π Technical Discussion PART 1: For the 2D continuous function, its Fourier Transform can be given as follow: Where F(u,v) is the Fourier Transform result with the frequency components u and v corresponding to x and y, respectively. And f(x,y) is the original continuous function. For the 2D discrete function, the Fourier Transform can be given as follow: Where u=0,1,2,………,M-1 v==0,1,2,………,N-1 From the 2D discrete Fourier Transform equation, one can show that the translation property of discrete Fourier Transform is given as: When uo=M/2, vo=N/2 we get: This equation shows that the Fourier spectrum can be shifted to the central point for better displaying by multiplying the original image by (-1)(x+y) . The 2D inverse Fourier Transform for discrete function can be computed as follow: Where x=0,1,2,………,M-1 y==0,1,2,………,N-1 PART 2: ∫ ∫ ∞ ∞? ∞ ∞? +?= dxdyeyxfvuF uvuxj )(2),(),( π ∑ ∑ ? = ? = +? = 1 0 1 0 )(2 ),(1),( M x N y N vy M uxj eyxf MN vuF π ∑ ∑ ? = ? = + = 1 0 1