# power spectrum plot of an image

158 views (last 30 days)
Dsp on 10 Dec 2013
Edited: Prasobhkumar P. P. on 10 Mar 2020
I am quite new to signal processing and matlab, and I have begun (a small project) to calculate the power spectrum of an image in the frequency domain - and plot this against the frequency.
So, what I have till now is the following:
close all; clear all;
img = imread('my_pic.jpg','jpg'); % it is a color image 400px x 400px
img = rgb2gray(img); % change to gray
psd = 10*log10(abs(fftshift(fft2(img))).^2 );
figure(2); clf
mesh(psd)
So far it looks good - I get the mesh plot which resembles the spectra I see in various academic papers.
However, what I am looking for is a graph plot of this power spectra Vs. the frequency - and I am not entirely sure how to get this frequency vector for the image. I could do say:
N=400; % the image is 400 x 400
f=-N/2:N/2-1; % possible frequencies?
.. but I am not convinced this is entirely correct.... I'd really appreciate if someone could point me in the right direction to plot log frequency Vs. the power spectrum.

#### 1 Comment

Jan on 11 Dec 2013
I would agree to your suggestion. The notion of spatial frequencies in image is at a first glance not quite as intuitive as in other digital signal processing applications. However, in my opinion a spatial frequency in an image depends on the geometry of the optics (recording distance, pixel size, pixel pitch etc.) in much the same way as a frequency in a one dimensional signal depends on the sampling rate. Yet I would appreciate an input of more experienced forum members on this point.

Prasobhkumar P. P. on 4 Mar 2020
Edited: Prasobhkumar P. P. on 10 Mar 2020