Robotics StackExchange | Archived questions

How to interpret depth data from sensor_msgs/Image

Hi,

I am using iai_kinect2 to get a depth cloud in the form sensor_msgs/Image on the topic /kinect2/sd/image_depth_rect. The topic contains :

header: 
  seq: 0
  stamp: 
    secs: 1451490491
    nsecs: 411043474
  frame_id: kinect2_ir_optical_frame
height: 424
width: 512
encoding: 16UC1
is_bigendian: 0
step: 1024
data: [176, 11, 0, 0, 176, 11,0, 52, 11, 29, 11, 52, 11, 29, 11, 122, 11, 170, 11, 221, 11, 174, 11, 191, 11, 166, 11, 187, 11, 0, 0, 0, 0, 131, 11, 0, 0, 0, 0, 180, 11, 0, 0, 0, 0, 0, 0, 0, 0, 114, 11, 86, 11, 131, 11, 0, 0, 131, 11, 0, 0, 0, 0, 50, 11, 43, 11, 138, 11, 251, 11, 152, 11, 85, 11, 47, 11, 68, 11, 122, 11, 124, 11, 123, 11, 109, 11, 0, 0, 199, 10, 0, 0, 117, 11, 140, 11, 0, 0, 64, 11, 91, 11, 0, 0, 0, 0, 241, 10, 0, 0, 223, 10, 19, 11, 115, 11, 121, 11, 102, 11, 136, 11, 173, 11, 169, 11, 93, 11, 73, 11, 123, 11, 151, 11, 124, 11, 96, 11, 49............]

As you see, data contains numbers between 0 and 255. How can I make sense of it? I want to use depthimage_to_laserscan to convert this depth image to laser scan but before that I need to perform some operations on the depth image (or depth_cloud) like removing points on the ground and noise and therefore, I need to know how to interpret this depth data. Can someone tell how to interpret this data?

Thanks in advance.
Naman Kumar

Asked by Naman on 2015-12-30 11:31:29 UTC

Comments

Answers

in very basic , "it's a lidar data". Your node converted the 3d into 2d for slam. Read more on robopeak lidar data for better understanding.

Asked by denim_patel on 2015-12-30 12:08:36 UTC

Comments

What do you mean by "Its a lidar data" ? This data is the direct output of iai_kinect2 before running depthimage_to_laserscan. It is a depth cloud data from Kinect 2 of the form sensor_msgs/Image.

Asked by Naman on 2015-12-30 12:13:56 UTC

Hi,

Here is a link to the documentation of the sensor_msgs/Image. About 16UC1 coding, 16U means that the colour is coded in 16 unsigned bits per channel, and C1 that there is one channel. Then, convert your colors in hexadecimal and see the table with some 16U colors at this link.

Asked by est_CEAR on 2016-02-09 11:11:36 UTC

Comments