{"id":5588,"date":"2022-06-08T16:08:01","date_gmt":"2022-06-08T20:08:01","guid":{"rendered":"https:\/\/blogs.mathworks.com\/steve\/?p=5588"},"modified":"2022-06-08T16:08:01","modified_gmt":"2022-06-08T20:08:01","slug":"find-all-the-highest-pixel-values-using-ismember","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/steve\/2022\/06\/08\/find-all-the-highest-pixel-values-using-ismember\/","title":{"rendered":"Find all the highest pixel values using ismember"},"content":{"rendered":"
Today's post shows how to use <\/span>ismember<\/span> to conveniently locate all of the highest pixel values in an image. For example, in the following Hough transform image, what are the 20 highest values, and <\/span>where are all the pixels that have those values<\/span>?<\/span><\/div>
\"hough-transform-image.png\"<\/img><\/div>
I got the idea from some example code written by Image Processing Toolbox writer Megan Mancuso. Megan says she got the idea to use <\/span>ismember<\/span> this way from <\/span>Ahmed's answer<\/span><\/a> on MATLAB Answers.<\/span><\/div>
Here's where the Hough transform image above comes from.<\/span><\/div>
A = imread(<\/span>\"gantrycrane.png\"<\/span>);<\/span><\/span><\/div><\/div>
imshow(A)<\/span><\/span><\/div>
<\/div>
<\/div>
<\/div>
<\/div><\/div><\/div><\/div><\/div>
bw = edge(rgb2gray(A),<\/span>\"canny\"<\/span>);<\/span><\/span><\/div><\/div>
imshow(bw)<\/span><\/span><\/div>
<\/div>
<\/div>
<\/div>
<\/div><\/div><\/div><\/div><\/div>
[H,T,R] = hough(bw);<\/span><\/span><\/div><\/div>
imshow(H,[],XData=T,YData=R)<\/span><\/span><\/div><\/div>
daspect <\/span>auto<\/span><\/span><\/div><\/div>
axis <\/span>on<\/span><\/span><\/div>
<\/div>
<\/div>
<\/div>
<\/div><\/div><\/div><\/div><\/div>
Now let's sort to figure out the highest 20 values of <\/span>H<\/span>.<\/span><\/div>
sorted_H_values = sort(H(:),<\/span>\"descend\"<\/span>);<\/span><\/span><\/div><\/div>
highest_H_values = sorted_H_values(1:20)<\/span><\/span><\/div>
highest_H_values = <\/span>20\u00d71<\/span><\/div><\/div>
293\r\n 250\r\n 209\r\n 205\r\n 182\r\n 177\r\n 173\r\n 172\r\n 171\r\n 168\r\n<\/div>
<\/div>
<\/div><\/div><\/div>
<\/div>
<\/div>
<\/div><\/div><\/div><\/div><\/div>
Finally, I'll use <\/span>ismember<\/span> to compute a binary image showing all the elements of H that have one of those 20 highest values. (And I'll use <\/span>xlim<\/span> and <\/span>ylim<\/span> to zoom in so that we can see a few of those elements clearly.)<\/span><\/div>
mask_20_highest_values = ismember(H,highest_H_values);<\/span><\/span><\/div><\/div>
imshow(mask_20_highest_values,XData=T,YData=R)<\/span><\/span><\/div><\/div>
daspect <\/span>auto<\/span><\/span><\/div><\/div>
axis <\/span>on<\/span><\/span><\/div><\/div>
xlim([-10 60])<\/span><\/span><\/div><\/div>
ylim([141 452])<\/span><\/span><\/div>
<\/div>
<\/div>
<\/div>
<\/div><\/div><\/div><\/div><\/div>
Any element-wise logical function, such as <\/span>ismember<\/span>, can be used in this way to produce a binary image. So add this to your bag of MATLAB tricks!<\/span><\/div>\r\n<\/div>