# Understanding /camera/depth/points

What does the value in the 'data' field of the data published by /camera/depth/points represent? They are supposed to represent the distance of every pixel from the Kinect center point? How to I convert the data generated as point cloud into corresponding cartesian 3D coordinates ? An array of size [4915200x1] is generated. Which means distance to each pixel is represented using 16 entries in matrix (because 480 x 640 x 16 = 4915200). How do I understand this. Please explain the program behind visualization of this pointcloud data in rviz .

I kept my kinect in front of an plain 2D wall parallel with it. I thought that I will get a matrix with uniform value through out as the wall is equidistant from kinect. But I get a data which I could not understand and decode!

Originally posted by gaussian on ROS Answers with karma: 88 on 2015-01-28

Post score: 1

The data is all the point clouds data fields (x, y, z) encoded into that array. Usually you don't read that by hand. Use a PCL Subscriber that allows you to directly get point data with x,y,z coordinates.

Originally posted by dornhege with karma: 31395 on 2015-01-29

This answer was ACCEPTED on the original site

Post score: 0

Comment by gaussian on 2015-01-29:

Comment by dmngu9 on 2015-03-26:
i have converted pointcloud2 from kinect to pointcloudxyz. Are x,y,z coordinates in meters or mm?

Comment by dornhege on 2015-03-26:
m ,

Comment by dmngu9 on 2015-04-07:
for the camera/depth/points topic. does it publish coordinates of pixel or world coordinate. Im a bit confused

Comment by dornhege on 2015-04-08:
These are x,y,z points in the frame that is passed in the message header.

Comment by dmngu9 on 2015-04-08:
my frame id is camera_depth_optical_frame. are the xyz in my frame world coordinates? which will give depth from kinect to objects?

Comment by dornhege on 2015-04-08:
They are very likely relative to the kinect sensor.

Comment by dmngu9 on 2015-04-08:
i still dont get what you mean