Masking by pre-selected ROI and subtracting all other areas

I'm trying to create an algorithm that can generate a binary mask within a pre-determined ROI that is different for each image in the set. For example, the input images would look something like this, with a red rectangle drawn around the ROI. I need to do calculations on the area and intensity of the pixels within the ROI, so I first need to isolate the region by essentially designating the intensity of all pixels outside of the ROI to be 0. My best idea for accomplishing this is to filter using the shift in contrast between the image and the rectangle which is drawn over the image, but I'm not sure of the matlab functions that will allow me to do this. Any help would be appreciated! Thanks!

Answer 1

[Image Analyst]

You just need poly2mask() and regionprops(). Assuming you have your ROI in a list of (x,y) coordinates:

[rows, columns, numberOfColorChannels] = size(rgbImage);
mask = poly2mask(x, y, rows, columns); % Turn (x, y) into a binary image.
[r, g, b] = imsplit(rgbImage);
redProps = regionprops(mask, r, 'Area', 'MeanIntensity');
greenProps = regionprops(mask, g, 'Area', 'MeanIntensity');
blueProps = regionprops(mask, b, 'Area', 'MeanIntensity');

[Lindsey Lovitt]

So, in a batch set of images, the ROI will be in different positions, rather than at a concrete set of coordinates for every image.

[Image Analyst]

That's fine. It will still work. It really doesn't matter what the (x,y) values are or where they came from. You can have different sets of (x,y) values for each ROI and my code will still work. Did you even try it?

[yanqi liu]

yes，sir，use poly2mask get logica mask，and then use it to other images，if the image size not equal，may be use imresize to get mask to image size，and do segment