You would probably be better off creating a simple new node that reads the
images from the file(s) and publishes them seperately for the
stereo_image_proc node to use.

There is a good tutorial at
http://www.ros.org/wiki/image_transport/Tutorials/PublishingImages that
reads in an image from a file and publishes it.  You could modify it to
publish two images( you will want to synchronize the timestamps).

You probably will also want to look at
http://www.ros.org/wiki/camera_info_manager and use that for getting and
setting the camera info.

There may already be a package around to do what you need, but I didn't find
one with a cursory search.

I have been using the uvc_stereo package which reads images from usb cameras
and publishes them for stereo processing.  You could get that and look at
the code as an example of how to publish for stereo correctly.

Hope this helps.

-Jeff Hyams
RE2, Inc.

On Wed, Oct 20, 2010 at 8:55 AM, kollok kollok <kolloktn@gmail.com> wrote:

> Hello,
>
> I am working on a stereoscopy vision project and I have the intention to
> use ROS in this project. But I have a couple of questions before starting
> and I would be very grateful for your help.
>
> 1- Is it possible to edit the source code of a package (for example
> stereo_image_proc) and recompile it to be used in an other project?
>
> 2- For stereo_image_proc, could I specify the paths of two images (from my
> hard disk) and make them inputs for the stereo_image_proc in order to
> calibrate, rectify and extract disparity map from them?
>
> Thank you in advance
>
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>