{"id":250,"date":"2009-02-27T15:44:25","date_gmt":"2009-02-27T20:44:25","guid":{"rendered":"https:\/\/blogs.mathworks.com\/steve\/2009\/02\/27\/using-ismember-with-the-output-of-regionprops\/"},"modified":"2019-10-28T15:27:31","modified_gmt":"2019-10-28T19:27:31","slug":"using-ismember-with-the-output-of-regionprops","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/steve\/2009\/02\/27\/using-ismember-with-the-output-of-regionprops\/","title":{"rendered":"Using ismember with the output of regionprops"},"content":{"rendered":"
\r\n

I recently heard a MathWorks application engineer (hi, Brett<\/a>!) say that a short little paragraph in the regionprops documentation<\/a> was very \"powerful.\" Here it is:\r\n <\/p>\r\n

The function ismember<\/tt> is useful in conjunction with regionprops<\/tt> for selecting regions based on certain criteria. For example, these commands create a binary image containing only the regions\r\n whose area is greater than 80.\r\n <\/p>

   idx = find([stats.Area] > 80);\r\n   BW2 = ismember(L,idx);<\/pre>
When I heard Brett's comment, I immediately pulled out my high-tech note-taking device (index card) from my back pocket and\r\n      wrote: \"regionprops - ismember - blog topic.\"\r\n   <\/p>\r\n   
The basic idea is this: If I have a binary image containing objects, how can I make a new binary image containing a subset\r\n      of the objects satisfying certain criteria? The answer is in three parts:\r\n   <\/p>\r\n   
1. Use bwlabel<\/tt><\/a> and regionprops<\/tt> to compute measurements of the objects.\r\n   <\/p>\r\n   
2. Compute a vector of labels for the objects satisfying the criteria. Sometimes a special structure syntax called the \"comma-separated\r\n      list\" is very useful for this step; I'll show that syntax below.\r\n   <\/p>\r\n   
3. Use ismember<\/tt><\/a> to compute the desired binary image.\r\n   <\/p>\r\n   
I've posted several examples using bwlabel<\/tt> and regionprops<\/tt>, so I won't go into much detail about them.  You label objects in a binary image using bwlabel<\/tt>, and then you compute geometric measurements of the labeled objects using regionprops<\/tt>.\r\n   <\/p>\r\n   
Let's use the blobs.png image for our example.<\/p>
bw = imread('blobs.png'<\/span>);\r\nimshow(bw)<\/pre> 
L = bwlabel(bw);\r\ns = regionprops(L, 'Area'<\/span>, 'BoundingBox'<\/span>);\r\ns(1)<\/pre>
\r\nans = \r\n\r\n           Area: 35\r\n    BoundingBox: [0.5000 0.5000 14 14]\r\n\r\n<\/pre>
regionprops<\/tt> returns a structure array containing the desired measurements for each object.\r\n   <\/p>\r\n   
The comma-separated list syntax for structures offers an easy way to convert a particular field from a structure array into\r\n      a vector, assuming that the field value is always a scalar.  Here's how it works.  When s<\/tt> is a structure array with field foo, and you write the following expression:\r\n   <\/p>
   s.foo<\/pre>
MATLAB translates that into this \"comma-separated list\":<\/p>
   s(1).foo, s(2).foo, ..., s(end).foo<\/pre>
Therefore, if s(k).foo<\/tt> is a scalar, you can create a vector of all the foo<\/tt> values by enclosing s.foo<\/tt> in brackets, like this:\r\n   <\/p>
   foo_fector = [s.foo];<\/pre>
Here's how it works for the 'Area'<\/tt> measurements we computed above for the blobs.png image.\r\n   <\/p>
area_values = [s.Area]<\/pre>
\r\narea_values =\r\n\r\n  Columns 1 through 7\r\n\r\n          35         393          19         139          36          58         736\r\n\r\n  Columns 8 through 14\r\n\r\n          53           2           2           1           2           2           2\r\n\r\n  Columns 15 through 21\r\n\r\n           1         196        3940        4326         504          40         422\r\n\r\n  Columns 22 through 28\r\n\r\n         966         655         175          40          39        1108          35\r\n\r\n<\/pre>
Suppose we want a list of all the objects whos area is between 100 and 1000. We can do that using find<\/tt> and relational operators, like this:\r\n   <\/p>
idx = find((100 <= area_values) & (area_values <= 1000))<\/pre>
\r\nidx =\r\n\r\n     2     4     7    16    19    21    22    23    24\r\n\r\n<\/pre>
And that brings us to ismember<\/tt>. We can construct a binary image containing all the objects whos area is between 100 and 1000 by passing L<\/tt> and idx<\/tt> to ismember.\r\n   <\/p>
bw2 = ismember(L, idx);\r\nimshow(bw2)<\/pre> 
Let's try that with another measurement, solidity<\/i>.  Solidity is defined as the area of the region divided by the area of the convex hull of the region.\r\n   <\/p>
s = regionprops(L, {'Solidity'<\/span>});\r\nsolidity = [s.Solidity]<\/pre>
\r\nsolidity =\r\n\r\n  Columns 1 through 9\r\n\r\n    0.2397    0.9680    0.6786    0.0700    0.8000    0.1321    0.4461    0.7910    1.0000\r\n\r\n  Columns 10 through 18\r\n\r\n    1.0000    1.0000    1.0000    1.0000    1.0000    1.0000    1.0000    0.0972    0.1681\r\n\r\n  Columns 19 through 27\r\n\r\n    1.0000    0.2235    0.0386    0.1505    0.9791    0.9669    0.2260    0.2191    0.5604\r\n\r\n  Column 28\r\n\r\n    0.2397\r\n\r\n<\/pre>
Here are all the objects with a solidity of 0.9 or higher:<\/p>
imshow(ismember(L, find(solidity >= 0.9)))<\/pre> 
And here are all the objects with a solidity of 0.25 or lower:<\/p>
imshow(ismember(L, find(solidity <= 0.25)))<\/pre>