2

I'm trying to recognize rectangular boxes using Kinect, I know I can use OpenCV or PCL to do this but my C++ is a bit rusty, I'm new to Computer Vision in 3D and I want to do it as quick as possible, it's only a demo for a bigger project.

Is there anyway to do it using C# or at least Python, I don't care if I have to learn 3D.

If not, which library should I use, OpenCV or PCL?

Lucas
  • 13,679
  • 13
  • 62
  • 94
user1860045
  • 213
  • 2
  • 5
  • What exactly are you wanting to do, and on what platform? – Nicholas Pappas Nov 28 '12 at 14:39
  • Right now the demo has to recognize a box, I thought of a depth map and then coloring only the box, leaving everything else in a gray scale. I tried to do it with the SDK for Windows, and I'd like to stay in Windows if possible. – user1860045 Nov 28 '12 at 15:02
  • You can use the official SDK to access the Kinect hardware, but there are no built-in abilities for doing what you seek. As @Sacx mentions below, OpenCV can help with some of the leg work (passing it the image data obtained by the SDK). There are C# wrappers that will let you stay in Windows. – Nicholas Pappas Nov 28 '12 at 15:39

1 Answers1

2

I recommend OpenCV.

If you want to work in C# or python with OpenCV then you have wrappers for both of them:

http://www.emgu.com/wiki/index.php/Main_Page

http://www.neuroforge.co.uk/index.php/getting-started-with-python-a-opencv

Sacx
  • 6,276
  • 4
  • 22
  • 29
  • 1
    There are also multiple existing examples of using Kinect and OpenCV together to do many different things. OpenCV will likely do what you want, but will certainly give you a good introduction if it somehow comes up short. – Nicholas Pappas Nov 28 '12 at 14:38