MATLAB: How to count black pixels in a region of an image that can only have 1 white neighbor pixel

pixel count

I have a binary image (black background and white pixels). How do i count the number of blackpixels witch in it's neighbourhood of 5*7 (5 lines and 7 colums) can only have a white pixel.

I can try using bwhitmiss() but i would need 35 matrixs with one white pixel in each "matrix slot".

Can you tell me an easyest way to achieve this?

Best Answer

You could try leveraging conv2.

Here is the idea:

in = [
    0 0 0 0 0 0 0 0 
    0 1 0 0 0 0 0 0 
    0 0 0 0 0 0 0 0 
    0 0 0 0 0 0 0 0
    0 0 1 0 0 0 0 0 
    0 0 0 0 0 0 0 0 
    0 0 0 0 0 0 0 0 ]
k = [
    1 1 1
    1 1 1
    1 1 1]
r = conv2(in,k,'same')

The result r will have 0 at all locations which DO NOT have 1's in the immediate 3x3 neighborhood.

(Have a look at conv2 and play with the positions of 1's in in to get a better idea)

Related Solutions

MATLAB: Serious problem with “createMask” function.

Here is some code illustrating how to use bwboundaries to get X-Y polygon vertices that work better for mask creation via createMask or poly2mask (which createMask calls). The idea is to get a polygon that includes the centers of the desired mask pixels inside the polygon instead of along its edges. First, here is your code with some graphics to show the area of each pixel and to show the output of bwboundaries going through pixel centers.

BW = [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0];
   B = bwboundaries(BW);
   b = B{1};
   imshow(BW,'InitialMagnification','fit');
   hold on
   [M,N] = size(BW);
   x = [0.5 (N+0.5)];
   for k = 1.5:(M-0.5)
       y = [k k];
       plot(x,y,'Color',[.7 .7 .7],'LineStyle','--');
   end
   y = [0.5 (M+0.5)];
   for k = 1.5:(N-0.5)
       x = [k k];
       plot(x,y,'Color',[.7 .7 .7],'LineStyle','--');
   end
   plot(b(:,2),b(:,1),'r')
   hold off

And here is modified code that produces X-Y values that do not go directly through pixel centers.

 BW3 = imresize(Z,3,'nearest');
 B3 = bwboundaries(BW3);
 b3 = B3{1};
 b3 = (b3 + 1)/3;
 imshow(BW,'InitialMagnification','fit');
 hold on
 [M,N] = size(BW);
 x = [0.5 (N+0.5)];
 for k = 1.5:(M-0.5)
     y = [k k];
     plot(x,y,'Color',[.7 .7 .7],'LineStyle','--');
 end
 y = [0.5 (M+0.5)];
 for k = 1.5:(N-0.5)
     x = [k k];
     plot(x,y,'Color',[.7 .7 .7],'LineStyle','--');
 end
 plot(b3(:,2),b3(:,1),'r')
 hold off

MATLAB: Cannot get the same result in GA as I get in linprog

GA solver is stochastic in nature, there should be no reason to expect the same results as that of linprog. Infact you may not even get the same results for two successive runs of GA.

However there are ways to make the results closer by including a hybrid solver. Change the opts as follows:

opts = gaoptimset('StallGenLimit',500,'Generations',500,'TolFun',1e-1,'HybridFcn',@fmincon);

More info here:

http://www.mathworks.com/help/gads/gaoptimset.html

Best Answer

Related Solutions

MATLAB: Serious problem with “createMask” function.

MATLAB: Cannot get the same result in GA as I get in linprog

Related Question