Is there a way to use a detectfeatures function without converting an image to grayscale?
1 view (last 30 days)
I'm trying to write code that will grab features out of a color image, but pretty much all code I see doing this involve first converting the image to grayscale. Right now I'm using detectSURFfeatures for grabbing the features from a grayscale image. However, I was wondering if there's anyway to detect features of an image that is a say NxMx3 or NxMx5 without reducing it to a NxM? It doesn't have to be a detectXXXXfeatures function, anything will do that grabs features out of an image.
Dima Lisin on 15 Feb 2016
Edited: Dima Lisin on 15 Feb 2016
Currently, all point feature detectors in the Computer Vision System toolbox work on grayscale images. If you want to somehow use color, you would have to choose an algorithm (such as CSIFT), and then either find an implementation or write it yourself.