{"id":358,"date":"2011-01-07T18:18:58","date_gmt":"2011-01-07T23:18:58","guid":{"rendered":"https:\/\/blogs.mathworks.com\/steve\/2011\/01\/07\/feature-and\/"},"modified":"2019-10-29T13:51:01","modified_gmt":"2019-10-29T17:51:01","slug":"feature-and","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/steve\/2011\/01\/07\/feature-and\/","title":{"rendered":"Feature AND"},"content":{"rendered":"
\r\n

I'm going to take a break from looking at M&Ms<\/a> this week and instead consider a question that Richard asked yesterday<\/a>:\r\n <\/p>\r\n

\"I have two binary images of the same size. I want to select regions from image one, where that region has to partially overlap\r\n with a region from image two.\"\r\n <\/p>\r\n

I've seen this called the \"Feature-AND\" operation, and you can use the Image Processing Toolbox function imreconstruct<\/tt><\/a> to do it. I'll use one of the toolbox sample images to show you how.\r\n <\/p>

bw = imread('text.png'<\/span>);\r\nimshow(bw)<\/pre> 
Suppose I've got a \"marker\" image that has a thick line running down the diagonal, like this:<\/p>
stripe = logical(eye(256));\r\nstripe = imdilate(stripe, ones(3,3));\r\nimshow(stripe)<\/pre> 
Now ask the question, \"Which objects in the original image touch the line?\" It's easy to use a simple elementwise AND to see\r\n      which pixels<\/b> in the original image touch the line:\r\n   <\/p>
touching_pixels = bw & stripe;\r\nimshow(touching_pixels)<\/pre> 
Now to get the entire objects that are touching the line, apply an operation called morphological reconstruction<\/i>, which is what imreconstruct<\/tt> does. This function takes two inputs, a marker image and a mask image. Use the touching pixels as the marker image, and use\r\n      the original image as the mask image.\r\n   <\/p>
marker = touching_pixels;\r\nmask = bw;\r\nout = imreconstruct(marker, mask);\r\nimshow(out)<\/pre> 
Neat!<\/p>\r\n   
Here's another variation: \"Which objects in the original image contain a vertical stroke that's at least 11 pixels tall?\"<\/p>\r\n   
Start by eroding the original image by a vertical line structuring element.<\/p>
se = strel(ones(11,1));\r\nbw_eroded = imerode(bw, se);\r\nimshow(bw_eroded)<\/pre> 
You can see that many characters are gone completely. If we define the pixels that are left as the marker image, then reconstruction\r\n      gives us back the complete objects:\r\n   <\/p>
marker = bw_eroded;\r\nout = imreconstruct(marker, mask);\r\nimshow(out)<\/pre> 
This is a very useful operation, and I recommend that you add it to your bag of image processing tricks.<\/p>