Hello, Let's assume I have an image with 0's as background B and a connected set S made of 1. Is there an easy way with scikit-image to find the maximum (in terms of area) subset of S that is convex. This question is also described here: - http://math.stackexchange.com/questions/455687/find-the-maximum-convex-area - http://stackoverflow.com/questions/17943482/find-the-maximum-convex-area/179... I was wondering if there was an algorithm already implemented in the library to easily do that. Otherwise I guess I'll have to implement it by myself as described in the links above. Many thanks,
On Wed, 20 Apr 2016 at 01:40 <amiksvi@gmail.com> wrote:
Let's assume I have an image with 0's as background B and a connected set S made of 1. Is there an easy way with scikit-image to find the maximum (in terms of area) subset of S that is convex.
I may be wrong, but my gut feeling is that this would require a search over quite a large space of options. There may be some ways to simplify the search, e.g. you could fit the largest circle that inscribes your object, and then only consider all combinations of the remaining pixels; still a lot of work. [... time passes ...] Well, I guess I should have asked Bruce Merry first: http://stackoverflow.com/a/23639901/214686 Care to implement this for us as a PR? The algorithm is O(N^3). Stéfan
participants (2)
-
amiksvi＠gmail.com -
Stéfan van der Walt