0
$\begingroup$

Rosanswers logo

Hi,

I'm currently writing a program which is to use data from Kinect in order to perform image processing. I'm using the OpenNI-Kinect driver with Ubuntu 10.10 and ROS Diamondback.

The image processing needs color image, depth image and point cloud data, then I'm subscribing to the topics "camera/rgb/image_color", "camera/depth/image" and "camera/depth/points", which allows me to save pointers to the data, and - if needed - to save the different data as images or binary files.

My problem is that I don't know how to get a consistent set of these data, i.e. how to get RGB, depth and point cloud data which have the same time stamp.

The program is looping with a while(true) which contains:

  • a cvWaitKey(1) to set the saving and to exit the loop,
  • and a ros::spinOnce() to actually loop.

For the moment the program saves point cloud and depth data which have the same time stamp but the RGB never matches. Would you see how to get a RGB captured at the same moment ?


Originally posted by Matthieu on ROS Answers with karma: 21 on 2011-05-02

Post score: 2

$\endgroup$

2 Answers 2

0
$\begingroup$

Rosanswers logo

You can use the approximate time syncronization policy, which mostly gets the right data bundle. Depth image and point cloud should have the exact same time stamp, the rgb image shouldn't be off more than 1/30 sec. I have recently introduced a filter that checks for time offsets in rgbdslam, and does get triggered every now and then.

Btw, better loop with while(ros::ok()){ ... } instead of while(true). If your node gets killed, your program will stop.


Originally posted by Felix Endres with karma: 6468 on 2011-05-02

This answer was ACCEPTED on the original site

Post score: 3

$\endgroup$
0
$\begingroup$

Rosanswers logo

In general you can't expect to have perfect synchronization between RGB and depth for the Kinect. From the openni_camera wiki page:

Synchronization

Other than the Primesense devices, the Kinect does not support hardware synchronization. Thus the time points where the RGB image and the depth image are captured may be up to ~16ms away from each other. For all devices we use a Approximatetimesynchronizer to generate the point clouds /camera/rgb/points. For Primesense devices the hardware synchronization is turned on automatically if both streams are used (subscribing to image

  • depth).

Note that /camera/depth/image and /camera/depth/points both contain essentially the same information--the former is the depth of each point and the latter is the point cloud derived from that depth data. Subscribing to both of them simultaneously is probably overkill.


Originally posted by Patrick Bouffard with karma: 2264 on 2011-05-02

This answer was NOT ACCEPTED on the original site

Post score: 3

$\endgroup$

Your Answer

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