search

RIVeR-Lab / aero_srr_13

Repository for storing the aero_srr ROS packages
2 stars 2 forks source link

IOImages in vision_comm must also handle the camera paramerters topic published in ros #13

Open bpwiselybabu opened 11 years ago

bpwiselybabu commented 11 years ago

currently IOImages does not give any camera parameters. I must either load these from file for connected camera or load it from appropriate rostopic

randomlogic78 commented 11 years ago

Should we put these on the parameter server

On Feb 14, 2013, at 7:40 PM, bpwiselybabu notifications@github.com wrote:

currently IOImages does not give any camera parameters. I must either load these from file for connected camera or load it from appropriate rostopic

— Reply to this email directly or view it on GitHubhttps://github.com/RIVeR-Lab/aero_srr/issues/13.

aruisdante commented 11 years ago

I would think so. Also, look into using dynamic_reconfigure for anything that could change at run-time (or might be useful to change at run-time during testing)

Adam Panzica

— Reply to this email directly or view it on GitHub.

bpwiselybabu commented 11 years ago

usually ros publishes a camera_info topic where the calib data is published so it makes sense to use IOImages to handle it and give it as opencv Mat that th users can use.

turgaysenlet commented 11 years ago

Dynamic reconfigure is nice but as you said you may not need it for everything. Especially if something is for initialization time only command-line parameters may work fine.

Turgay Senlet, M.Sc.

Ph.D. Candidate

E-mail: tsenlet@cs.rutgers.edu Web site: http://paul.rutgers.edu/~tsenlet

Computational Biomedicine Imaging and Modeling Center Department of Computer Science Rutgers, The State University of New Jersey

On Feb 14, 2013, at 10:44 PM, Adam Panzica notifications@github.com wrote:

I would think so. Also, look into using dynamic_reconfigure for anything that could change at run-time (or might be useful to change at run-time during testing)

Adam Panzica

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

turgaysenlet commented 11 years ago

Synchronization between published camera image messages and the camera_info messages are important. Usually camera image is of type sensor_msgs::Image. I am not sure if you are planning on using something else.

Turgay Senlet, M.Sc.

Ph.D. Candidate

E-mail: tsenlet@cs.rutgers.edu Web site: http://paul.rutgers.edu/~tsenlet

Computational Biomedicine Imaging and Modeling Center Department of Computer Science Rutgers, The State University of New Jersey

On Feb 14, 2013, at 10:45 PM, bpwiselybabu notifications@github.com wrote:

usually ros publishes a camera_info topic where the calib data is published so it makes sense to use IOImages to handle it and give it as opencv Mat that th users can use.

— Reply to this email directly or view it on GitHub.

bpwiselybabu commented 11 years ago

For the april tag detection and estimation of the 6DOF pose i need to know the intrinsic camera calib. And i dont want this to be a parameter only loaded for becon detect. am sure most of the code where we want to estimate 3d pose will require camera calib. so i feel its critical that it is served by vision_comm

turgaysenlet commented 11 years ago

I think every camera image message should be accompanied with a related camera_info message and they should be "almost" synchronous.

Turgay Senlet, M.Sc.

Ph.D. Candidate

E-mail: tsenlet@cs.rutgers.edu Web site: http://paul.rutgers.edu/~tsenlet

Computational Biomedicine Imaging and Modeling Center Department of Computer Science Rutgers, The State University of New Jersey

On Feb 15, 2013, at 10:27 AM, bpwiselybabu notifications@github.com wrote:

For the april tag detection and estimation of the 6DOF pose i need to know the intrinsic camera calib. And i dont want this to be a parameter only loaded for becon detect. am sure most of the code where we want to estimate 3d pose will require camera calib. so i feel its critical that it is served by vision_comm

— Reply to this email directly or view it on GitHub.

bpwiselybabu commented 11 years ago

i dont think synchronization is very important. as the info topic only provides intrinsic parameters which are fixed. the extrinsic parameters are usually calculated using the tf. http://www.ros.org/doc/api/sensor_msgs/html/msg/CameraInfo.html

bpwiselybabu commented 11 years ago

only part that might change is the roi. but i feel we can handle rois if we transmitting the images between nodes in better ways

turgaysenlet commented 11 years ago

It is like this throughout ROS as I have seen. For example check the stereo_proc, it requires synchronous left camera_info, right camera_info, left image and right image.

Turgay Senlet, M.Sc.

Ph.D. Candidate

E-mail: tsenlet@cs.rutgers.edu Web site: http://paul.rutgers.edu/~tsenlet

Computational Biomedicine Imaging and Modeling Center Department of Computer Science Rutgers, The State University of New Jersey

On Feb 15, 2013, at 10:36 AM, bpwiselybabu notifications@github.com wrote:

i dont think synchronization is very important. as the info topic only provides intrinsic parameters which are fixed. the extrinsic parameters are usually calculated using the tf. http://www.ros.org/doc/api/sensor_msgs/html/msg/CameraInfo.html

— Reply to this email directly or view it on GitHub.

bpwiselybabu commented 11 years ago

i agree with left and right image sync for stereo camera but for a normal camera with just a single intrinsic parameter being set i dont think its important and also this sync with camera_info in stereo camera is imp only if both the cameras are changing pose with respect to one another because rectification and projection matrix is going to change.

turgaysenlet commented 11 years ago

Normally image and camera info are sent synchronously whether it is stereo, or mono whether the camera parameters change or not.

Turgay Senlet, M.Sc.

Ph.D. Candidate

E-mail: tsenlet@cs.rutgers.edu Web site: http://paul.rutgers.edu/~tsenlet

Computational Biomedicine Imaging and Modeling Center Department of Computer Science Rutgers, The State University of New Jersey

On Feb 15, 2013, at 10:57 AM, bpwiselybabu notifications@github.com wrote:

i agree with left and right image sync for stereo camera but for a normal camera with just a single intrinsic parameter being set i dont think its important and also this sync with camera_info in stereo camera is imp only if both the cameras are changing pose with respect to one another because rectification and projection matrix is going to change.

— Reply to this email directly or view it on GitHub.