load calibration parameters into camera_info

asked 2016-11-18 02:19:25 -0600

timku
1 ●6 ●7 ●7

I am using the Intel RealSense R200 for my vision based project. I have managed to run the stereo calibration for the infrared stream and I have also obtained the calibration data file (.yaml) for the left and right infrared stream.

Currently, I am subscribing to the raw left and right infrared stream ('/camera/ir/image_raw' and '/camera/ir2/camera_info') from the default nodelet (r200_nodelet_default.launch)

I would like to know how do I load the calibration parameters that I have obtained into the respective camera_info topics namely ('/camera/ir/camera_info' and '/camera/ir2/camera_info') so I can run the stereo_image_proc to obtain the rectified infrared stream.

answered 2016-11-18 05:14:38 -0600

afranceson
497 ●9 ●16 ●22 http://www.geduino.org/

Dear timku,

the standard way is to use the camera_info_manager (wiki).

BTW I'm quite confused by your goal: the Intel RealSense R200 manage the stereo vision by itself and its node provide the disparity map, depth point cloud and camera_info. So you don't need to calibrate it: this is a task for the camera itself.

Greetings Alessandro

edit flag offensive delete link

Comments

Hi Alessandro,

Do you have any example of using the camera_info_manager as I can't seem to find any example online. Do you mean that the camera is factory calibrated?

timku ( 2016-11-18 07:49:12 -0600 )edit

And I have managed to sync the left and right IR stream topic and would like to generate the disparity map through stereo_image_proc. However, the disparity window freeze after using the image_view disparity view method. Is this the proper method?

timku ( 2016-11-18 07:56:04 -0600 )edit

The stereo_image_proc node is designed to elaborate two images taken at same time from two different and separated camera (like two webcam) in order to create a disparity map and 3D point cloud.

The RealSense R200 instead is a stereo camera and produce those info by itself.

afranceson ( 2016-11-18 08:08:26 -0600 )edit

This means that the algorithm provided in stereo_image_proc is INSIDE the RealSense. Look at the topic published by the realsense_camera node: depth/camera_info (sensor_msgs/CameraInfo) depth/image_raw (sensor_msgs/Image) - basically the disparity map depth/points (sensor_msgs/PointCloud2)

afranceson ( 2016-11-18 08:11:32 -0600 )edit

Ok, thanks for the above infor! That clearly clarify some of my doubts on the Realsense camera!

timku ( 2016-11-18 10:14:30 -0600 )edit

Btw, I am working on my final year project (vision-based obstacle detection in GPS-denied outdoor environment) using the RealSense R200 through selecting appropriate stereo detector and stereo matching algorithms. After which, I will generate the disparity/depth map to detect any nearby obstacle.

timku ( 2016-11-18 10:21:19 -0600 )edit

Just to check is there any feature detector function and stereo matching algorithms that is inside the ROS package? Or I can just use the OpenCV library in ROS?

timku ( 2016-11-18 10:23:40 -0600 )edit

I guess this is a task for the Point Could Library. As OpenCV also PCL has a porting in ROS: see its wiki page. More info about PCL here.

afranceson ( 2016-11-18 10:54:26 -0600 )edit

add a comment

load calibration parameters into camera_info

1 Answer

Comments

Question Tools

Stats

Related questions

load calibration parameters into camera_info edit

1 Answer

Comments

Question Tools

Stats

Related questions

load calibration parameters into camera_info