Re: Regarding covariance matrix estimation
Phil Brubaker
Phil
On 10/28/2015 8:21 AM, Sushil Kumar Joshi wrote:
--Hello, I am working on capon algorithm for images using matlab. I have a set of 4 images of dimension r x c where r= no. of rows and c=no. of columns and N_im=no. of images. For the covariance matrix, take L=8 neighboring pixels around a target pixel, so I have a matrix G = [g1, g2, … g8], where g1 … g8 are all vectors of the samples. E.g. if I have 4 images, then each g will be a 4 by 1 vector. Then the sample covariance matrix is C = 1/8 * summation from 1 to 8 (G * G^H) (H is the matrix transpose with complex conjugate) which will be a 4 x 4 matrix for a single pixel.
Initially I am suppose to generate the stacked images, and then perform the above mentioned procedure for each pixel in the stacked images. My image has complex double pixel values.
Till now I have done for single pixel: y_cal is the stack of 4 images i.e 143 X 65 X 4
m=zeros(8,1);
for i=1:8
for j=1:4
m{i}=[y_cal(1,1,j)];
end
end
I am getting error like Cell contents assignment to a non-cell array object. I am unable to resolve it.
My ultimate task is to generate 4 X 4 covariance matrix for a single pixel and then 143 x 65 x 4 x 4 for the entire image. Moreover, here I have randomly chosen 8 pixels. I want the pixels which are the neighboring pixels (Square or rectangular window is mostly preferred).
Kindly provide me some guidance regarding how should I proceed through coding. Your help will be highly appreciated
You received this message because you are subscribed to the Google Groups "Spectral_Estimation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spectral_estima...@googlegroups.com.
To post to this group, send email to spectral_...@googlegroups.com.
Visit this group at http://groups.google.com/group/spectral_estimation.
For more options, visit https://groups.google.com/d/optout.
