Extracting image patch within an oriented bounding box of MSER

2 views (last 30 days)
I have a number of MSER regions obtained in an image. I have parameters of its bounding ellipse. I want to find out the image formed within its oriented bounding box. So, If the ellipse orientation is 60 degrees, I want to find out the mask corresponding to the bounding rectangle of the ellipse(oriented at 60 degrees) (please note that this is different from the bounding box given by regionprops !!). I then want to use that rectangle mask, and want to extract the region of the original image within that rectangle.
Now, I have solved part 1 where I am computing the bounding box rectangle, and am able to compute its four corner points when the rectangle is rotated to have an orientation of 0 degrees. However the trouble is the following :-
a) I get a large number (~1000) MSER regions in an image. The way I am computing the image patch within the bounded rectangle is :- In a temporary array rotate the image by (-60 degrees), and then capture the image portion within the four corner points
b) But consider the waste of this method. For each MSER region i am unnecessarily rotating the whole image for this work, which is wasteful.
Is there an efficient way of doing it, where I can directly do this without rotating the whole image?

Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!