How to apply custom filter to a grayscale image?

I need to have grayscale image and develop a program that applies the filters given below to that grayscale image. I implemented following code but not sure if it is ok or not?

grayImage = imread("cameramanImage.tiff");
h = [ 2  -1  -1 
      -1  2  -1
      -1 -1 2 ];
  im_outFirst = filter2(h, grayImage);

Answer 1

[Image Analyst]

I'd do this, using imfilter():

grayImage = imread("cameraman.tif");
h = [ 2  -1  -1 
     -1   2  -1
     -1  -1   2 ];
im_outFirst = imfilter(grayImage, h);

I'm not sure what filter2 does or how it differs from imfilter().

Answer 2

[Ameer Hamza]

For images, use conv2(): https://www.mathworks.com/help/releases/R2020a/matlab/ref/conv2.html. filter2 is similar to conv2 but applies a rotated filter. Also, I suggest using image as first input and filter as second

im_outFirst = conv2(grayImage, h);

In the general case, it does not matter, but if you specify 'same' or 'valid' option, then it makes a difference.

[Atalay Asa]

So you say the solution is right and also I can use conv2 method instead of filter2?

[Ameer Hamza]

Sorry, I wanted to write conv2() in the line of code in my answer. If you use this syntax

im_outFirst = filter2(grayImage, h);
im_outFirst = conv2(grayImage, h);

Then the difference is that filter2 applies a rotated filter. In the context of image processing, conv2() is more appropriate, or use imfilter() as suggested by Image Analyst.

[Atalay Asa]

Is it possible to implement without using built-in functions like filter2 or conv2?

[Ameer Hamza]

Something like this

im_gray = rgb2gray(im2double(imread('pears.png')));
h = [ 2  -1  -1
     -1   2  -1
     -1  -1   2];

im_filt = zeros(size(im_gray)-2);
for row=2:size(im_gray,1)-1
    for col=2:size(im_gray,2)-1
        im_filt(row-1, col-1) = sum(im_gray(row-1:row+1, col-1:col+1).*h, 'all');
    end
end

[Atalay Asa]

Thanks its worked!