Error while calculating the edge of an image

Question

I'm trying to extract edge from an image. I used the following algorithm.Input image(e11) which is a 512 * 512 grayscale image is also given.

1. Find the morphological gradient of input image(gradientim)
2. Find the negative image of the gradient image(negativeim)
3. Subtract original image from the closed image using bottom-hat transformations(bottomhatim).
4. Compute average pixel of input image(AVG)
5. Find the binary image based on AVG to smoothen the image.
6. Find the largest connected region to find the larger objects(CC).
7. Subtract largest region is from the smoothened image(edge).

The matlab code I'm written is given below

e11 = imread("lena.jpg");
e11= double(e11);
gradientim = imdilate(e11,se) - imerode(e11,se);
negativeim = imcomplement(gradientim);
bottomhatim = imclose(negativeim,se) -  e11 ;
AVG = mean2(e11);
%-------Computing binary image--------
for i=1:(height)
     for j=1:(width)
         if(AVG > bottomhatim(i,j,:))
              bottomhatim(i,j,:) = 1;
         else
              bottomhatim(i,j,:) = 0;
         end
     end
end
CC = bwconncomp(bottomhatim);
edge = bottomhatim - CC;

While doing step 7, Since the type of the connected component(CC) is 'struct' ,I'm getting an error as follows

"Undefined function 'minus' for input arguments of type 'struct'".

Is "bwconncomp" function can be used to find the largest connected region?Is there any alternate function for this?Please help me to correct this code.Thanks in advance.

Answer 1

You are assuming CC is an array but it's actually a structure. Specifically, this is what the docs say about bwconncomp:

 bwconncomp Find connected components in binary image.
    CC = bwconncomp(BW) returns the connected components CC found in BW. 
    BW is a binary image that can have any dimension. CC is a structure 
    with four fields:

       Connectivity   Connectivity of the connected components (objects).

       ImageSize      Size of BW.

       NumObjects     Number of connected components (objects) in BW.

       PixelIdxList   1-by-NumObjects cell array where the kth element
                      in the cell array is a vector containing the linear
                      indices of the pixels in the kth object.

By your algorithm description, you want to find the largest connected component and subtract this from your image to get the final image stored in edge. I recommend you use bwlabel instead, which returns a label map that labels each distinct object with a unique ID. The second output of bwlabel returns how many objects were detected.

I would use bwlabel, combined with histc to count up how many pixels belonged to each region. We would use this to determine the region that has the largest area. You'd then make a mask that consists of this object, then use this and subtract with your image to get the final output edge.

Specifically, do this at the end of your code:

%// Do bwlabel on image
[L, num] = bwlabel(bottomhatim);

%// Count how many values belong to each object, ignoring background
counts = histc(L(:), 1:num);

%// Find which object gives us the largest area
[~,max_id] = max(counts);

%// Make a mask and then subtract
CC = L == max_id;
edge = imsubtract(bottomhatim, CC);