MATLAB: I am trying to implement chain code starting from a pixel where intensity value is one I am successfully getting its chain code as 7 and the new co-ordinate should be this point but instead of this updation is not going on this way instead I am get

chain code

I am trying to implement chain code starting from a pixel where intensity value is one I am successfully getting its chain code as 7 and my new co-ordinate should be this point but instead of this updation is not going on this way instead I am getting all the chain codes of coordinates having intensity value 1. I am new to matlab.Image on which I am working is also attached.Any kind of help will be highly appreciated.Thanks in advance. Code is as follows:

   clc;
  clear;
  close all;
  I=imread('tool.png'); 
  % imshow(I)
  I2 = imcrop(I,[180 174 250 450]);
  figure
  imshow(I2)
  im2double(I2);
  t=graythresh(I2);
  BW=im2bw(I2,t);
  figure
  imshow(BW)
  J=bwperim(BW);
  B=im2double(J);
  imshow(B)
  [M, N]=size(B);
      p=7;
      q=6;
      h=5;
      s=4;
      t=3;
      u=2;
      v=1;
      w=0;
    for j=2:N-1
    for i=2:M-1
       if B(i,j)==1
           fprintf('Element(%d,%d) = %d.\n',i,j,B(i,j))
          x=i;
          y=j;   
   while (x>=2 && y<=246) 
  if B(x+1,y+1)==1
   fprintf('The chain code for (%d,%d)= %d.\n',x,y,p)
   newx=x+1;
   newy=y+1;
   break;
   elseif B(x+1,y)==1
         fprintf('The chain code for (%d,%d)= %d.\n',x,y,q)
  newx=x+1;
  newy=y;
   break;
         elseif B(x+1,y-1)==1
              fprintf('The chain code for (%d,%d)= %d.\n',x,y,h)  
  newx=x+1;
  newy=y-1;
  break;
  elseif B(x,y-1)==1
       fprintf('The chain code for (%d,%d)= %d.\n',x,y,s)
     newx=x;
      newy=y-1;
      break;
  elseif B(x-1,y-1)==1
      fprintf('The chain code for (%d,%d)= %d.\n',x,y,t)
       newx=x-1;
      newy=y-1;
        break;
  elseif B(x-1,y)==1
       fprintf('The chain code for (%d,%d)= %d.\n',x,y,u)
       newx=x-1;
    newy=y;
     break;
  elseif B(x-1,y+1)==1
       fprintf('The chain code for (%d,%d)= %d.\n',x,y,v)
    newx=x-1;
       newy=y+1;
         break;
  elseif B(x,y+1)==1
       fprintf('The chain code for (%d,%d)= %d.\n',x,y,w)
       newx=x;
      newy=y+1;
       break;  
  end
    end
  end
  end
  end

Best Answer

Well, you have a while loop that runs if x or y are within a certain range. Within that loop, you never modify x or y, so in theory, once in the loop it would never exit, since the loop condition never changes. Instead, you've created two new variables in the loop newx and newy that you never use again. That's one of the bug. Simple fix is to update x and y instead of creating newx and newy.

The reason your loop does not run for ever is that you immediately terminate the loop when one of the neighbouring pixel is 1. You're just lucky that none of the pixels are completely isolated otherwise you'd have encountered a forever running loop. I assume you actually want the opposite of what you've coded in that you want the loop to continue if there is a neighbour and exit if there is none, so actually your break statement should be within a final else that you'd hit if no neighbour is found.

Note that the conditional statement for your while loop is also wrong. You never check for an upper x.

Finally, note that you use x to denote rows of your image and y to denote columns. This is contrary to normal convention, x is normally columns and y rows.

Related Solutions

MATLAB: How to convert this to a for loop

% your while loop
x = input('Enter a number between 1 and 100: ');
i = 0;
while x>0
    x=x-i;
    i=i+1;
end 
fprintf('The while loop ran for %d interations\n' ,i)
% my for loop
x = input('Enter a number between 1 and 100: ');
for i = 0:100
    x = x - i;
    
    if x <= 0
        break;
    end
end 
fprintf('The while loop ran for %d interations\n', i + 1)
Enter a number between 1 and 100: 30
The while loop ran for 9 interations
Enter a number between 1 and 100: 30
The while loop ran for 9 interations

MATLAB: Matrix Value Replacement Problem

so the E and W directions work but for some reason I can't get the N and S directions to work at all

As far as I can tell the code you've posted is bound to error at some point as it will inevitably try to index at 0 or past the size M.

First, three essential pieces of advice:

Do not use scripts called by scripts. It's a recipe for bug. It makes it very hard to track where the variables used by a subscript come from and it's very easy for a script to accidentally stomp on a variable used by another script. Learn to use functions properly.
Make sure to use meaningful names for your variables. As Image Analyst and I said, your x and y variables are misleading. x is traditionally seen as the horizontal direction which is the 2nd dimension in matlab. So row, column would be better variable names in your case.
Learn to use the debugger. When something doesn't work, step through your code one line at a time with the debugger and see if what happened is what you expected. It's the fastest way to find out what is wrong with your code.

There are plenty of problems with your code. Some minors, others major.

C(x,y)=NumberGenerator;

For once, you're calling a function, not a script. In some ways it is good you're calling a function, except that this function is badly written. The function:

function [r]=RandomNumber(z)
zmin=0;
zmax=1;
N=1;
r=zmin+rand(1,N)*(zmax-zmin);
end

The name of the function RandomNumber does not match the name of the file. Matlab will put a wiggly line under the function to warn you of that. The function name that matlab use will always be the file name, so the RandomNumber name is ignored. The function takes an input but never does anything with it. Just as well, since you didn't pass an input when you called the function. The function seems to be designed to return a variable size array of random number between arbitrary boundaries but since everything is hardcoded, it's simply a wrapper for rand. In effect the initial line is just:

C(x,y) = rand;

A useful version of that function could have been:

function r = NumberGenerator(N, zmin, zmax)
    if nargin < 2   %no zmin and zmax passed, use default 0 and 1
        zmin = 0;
        zmax = 1;
    end
    if nargin < 1  %no N passed
        N = 1
    end
    r = zmin + rand(1, N)*(zmax-zmin);
end

Continuing:

if C(x,y)<.2
    %...








elseif C(x,y)>0
    %...

If C is les than 0.2, we take the first branch. When we arrive to the elseif, we're guaranteed that C >=0.2, so it's definitively greater than 0. The elseif is therfeore not needed and just misleading. This should have been written:

if C(x,y)<.2
    %...
else  %no need for other test
    %...

Moving on, we're in the C>=0.2 branch, we have, in MatrixGenV2:

if C(x,y)<.2000

As said, we're in the branch where C is guaranteed to eb 0.2 or more so that test will never be true

elseif C(x,y)>.8
    %...
elseif (.6>C(x,y))&&(C(x,y)>.4)
    %...
elseif (.8>C(x,y))&&(C(x,y)>.6) 
    %...
elseif(.4>C(x,y))&&(C(x,y))>.2            
   %...
end

Problem here, since you use strict inequalities (>) everywhere, if C is exactly 0.2, 0.4, 0.6 or 0.8 none of the branch will ever be true. You need to use some >= on some conditions. However,the simplest is to scarp thewhole ot and replace it with:

D(x, y) = discretize(C(x,y), [0.2 0.4 0.6 0.8 1]) %return 1 for values >= 0.2 and < 0.4, 2 for >=0.4 and < 0.6, ... 4 for >=0.8 and <= 1

Next, the big problem in Collapse:

if D(x,y)==E && M(x,y+1)==0
%...
elseif D(x,y)==W && M(x,y-1)==0
%...

If y happens to be equal to n, y+1 is an invalid index. If y happens to be 1, y-1 is 0, which is also not valid. This is boudn to result in an error at some point.

To finish, it seems to me that there is a fundamental problem with your algorithm. You carry out until all M is filled, but to fill a location, it and one of its neighbour must be empty (and then both are filled). I can see a scenario where the holes in M are just single zeros surrounded on all sides. In that case, any(M==0, 'all') is true so you keep on looping, but you can never find a location to fill.

Also, the closer you come to a filled matrix, he more often your picked random location is going to be something already filled. It would be much better to always pick a random location among the ones not filled:

while any(M==0,'all')
    unfilledlocations = find(M == 0)'  %linear indices of unfilled location
    randidx = randi(numel(unfilledlocations));  %pick one of them at random
    [row, column] = ind2sub(size(M), unfilledlocations(randidx)); %convert to row,column (your badly named x and y)
    %we're guaranteed that row, column is 0. However, it's not guaranteed that any neighbour is empty
end

Best Answer

Related Solutions

MATLAB: How to convert this to a for loop

MATLAB: Matrix Value Replacement Problem

Related Question