how to convolved two image ?

hi

i want to convolved 2 images.

i tried with conv2 but its not working.

i have 2 images one is Gaussian filterd image and other is sharped image. and i would like to convolved this 2 image. i write a code

im = imread('my original.jpg');

myfilter = fspecial('gaussian',[3 3], 0.5);

a = imfilter(image, myfilter);

b = imsharpen(image,'Radius',0.5);

c=conv2(a,b);

plz help me

Answer 1

[Image Analyst]

You DON'T want to use image as the name of a variable - it's the name of a built in function. Plus you need to convert to double when you call the convolution.

grayImage = imread('Cameraman.tif');
grayImage = double(grayImage);
subplot(2,3,1);
imshow(grayImage, []);
axis on;
title('Original Image', 'FontSize', 15);

myfilter = fspecial('gaussian',[3 3], 0.5);
subplot(2,3,2);
imshow(myfilter, []);
axis on;
title('myFilter', 'FontSize', 15);

a = imfilter(grayImage, myfilter);
subplot(2,3,3);
imshow(a, []);
axis on;
title('a', 'FontSize', 15);

b = imsharpen(grayImage,'Radius',0.5);
subplot(2,3,4);
imshow(b, []);
axis on;
title('b', 'FontSize', 15);

c=conv2(a,b, 'full');
subplot(2,3,5);
imshow(c, []);
title('c', 'FontSize', 15);
axis on;

[Image Analyst]

Why do you want to do this anyway - what does it mean? It doesn't make sense to me.

[SAM]

i would like to calculate the blurring effect...

[Image Analyst]

That's what image "a" is. It makes no sense whatsoever to convolve a blurred full sized image with a sharpened full size image. It's basically meaningless.

[SAM]

Undefined function 'conv2' for input arguments of type 'double' and

attributes 'full 3d real'.

again same error...

Error in tryingf11_2 (line 18)

c=conv2(a,b, 'same');

[SAM]

To estimate the blurring effect in a painting,

to model the blurred image I b as the result of Gaussian smoothing filter Gσ applied on a hypothetic sharp image I s ,

i.e. I b =G σ ∗ I s .

The symbol ∗ here means convolution.

Here the parameter σ of Gaussian filter and the sharp image I s are both unknown. Assuming that the frequency distribution for I s is approximately the same, we have the parameter σ of Gaussian filter to represent the degree of blurring. By taking Fourier-Transform on I b , this method looks for the highest frequency whose power is greater than a certain threshold and assumed it inverse-proportioned to the smoothing parameter σ . If the highest frequency is small, it can be considered to be blurred by a largeσ . So the blurring feature is measured as:

[Image Analyst]

Well that's fine, but that's not at all what you are doing.

[Anand]

Looks like your image, is 3D. Is it an RGB image, in that case use rgb2gray to convert it to grayscale:

im = double(rgb2gray(imread('my original.jpg')));

[SAM]

ya i got it

[SAM]

@image analyst

so what should i do?

Answer 2

[Matt J]

You didn't show your error messages, so I'm just guessing, but this

 im = imread('my original.jpg');

should probably be this

 image = double(imread('my original.jpg'));

[SAM]

Undefined function 'conv2' for input arguments of type 'double' and

attributes 'full 3d real'.

again same error...

[Image Analyst]

They both have to be gray scale images, not color. What does this say:

whos a whos b

Answer 3

[Taimoor Zafar]

I need to convolve two jpg images by using conv2 command but it's not working . Images are attached.

I am new to this field.

Plz help me

Answer 4

[Mahfuj]

The code works fine. But how we can reconstruct two images from the convolved image ?

[Image Analyst]

You'd have to know one of them, and then use an "inverse filter".