I blurred a grayscale image using an average filter and then I am supposed to deblur using an Inverse filter using Constrained Division. Is there a better way to do this?

Hello I need some help. I have to blur a grayscale image using a 5x5 blurring filter. Then I need to attempt to de-blur the result using inverse filtering with constrained division.The error that pops up is this:

 Error using  ./ 
 Complex integer arithmetic is not supported.

 Error in test2 (line 12)
 fbw = fftshift(fft2(blur))./bw;

When I blur the image using a butterworth filter it works just fine yet when I blur using the average filter my code won't work. Below is my attempt:

 clear
 I = imread('buffalo.png');
 H = fspecial('average',[5 5]) ;
 blurred=imfilter(I,H);
 blur=im2uint8(mat2gray((blurred)));
 imshow(blur)

 d = 0.01;           
 bw = blurred;        
 bw(find(bw<d)) = 1; 
 fbw = fftshift(fft2(blur))./bw; 
 ba = abs(ifft2(fbw)); 
 unb01 = im2uint8(mat2gray(ba));

Answer 1

[Matt J]

I think this should work.

   bw = double(blurred);

In general, I think you should be using im2double rather than im2uint8. The operations that you are doing require float precision.