1
$\begingroup$

I would like to write a simple program which processes the depth feed from an Asus xtion depth sensor using OpenNI. The sensor will be fixed like a CCTV camera and count multiple targets moving around.

The processing would involve some simple functions on each frame (background subtraction, level sets, connected components filter), and some multi-target tracking across frames. I have searched the web, but it is hard to see how best to get started (and I'm also quite new to programming in C).

Can anyone recommend any existing code that can help to get started / any libraries which would be suitable for this real-time application? Or perhaps there is some opensource code which already does such a thing? Would really appreciate any pointers from anyone with experience. Thanks!

Improve this question
$\endgroup$
2
  • $\begingroup$ Is this really a real-time application? Are you coding microcontrollers or using a real-time operating system? $\endgroup$
    – Shahbaz
    Feb 18, 2014 at 12:53
  • $\begingroup$ It's real-time in the sense that the counting data will be acted upon, so it needs to be calculated as soon as new data arrives (although a small lag would be acceptable) $\endgroup$
    – user3840
    Feb 18, 2014 at 13:13

1 Answer 1

Reset to default
0
$\begingroup$

I work with ASUS Xtion as well. The best library that I found so far and I am using daily is Point Cloud Library (PCL). You can find, download, and Install it from here.

After installing and getting it to work, you can go to the tutorial section of the website, and there is a very good tutorial similar to what you want. There is a pcl_people module that you can start from.

This tutorial aims at explaining how to detect people from RGB-D data with the pcl_people module. With the proposed method, people standing/walking on a planar ground plane can be detected in real time with standard CPU computation. This implementation corresponds to a people detection algorithm for RGB-D data.

Improve this answer
$\endgroup$
1
  • $\begingroup$ thanks - the website seems to be a bit slow at th moment but i'll check this out $\endgroup$
    – user3840
    Feb 18, 2014 at 14:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.