image_pipeline 0.9.0 released

| No Comments | No TrackBacks


image_pipeline-0.9.0 is now out. image_pipeline fills the gap between camera_drivers and higher-level vision processing. It includes a camera calibration GUI, a lightweight viewer for image topics, and nodes to perform color interpolation, rectification and stereo processing on camera images.

This is a stable release and has passed our user testing requirements. For changes see the changelist below. This release finally removes much long-deprecated code. In particular, the old stereoproc node is gone, as are all of the stereo messages excepting DisparityImage (which has a new definition).



  • Raw images may now be 8-bit mono, RGB, BGR, or any of the four bayer encodings.
  • Now using OpenCV for bayer interpolation.
  • Now using image_geometry for rectification.
  • Better error reporting when started in global namespace or input topics do not exist.


  • Inherited improvements from image_proc.
  • Now using OpenCV's block matching implementation to calculate the disparity image.
  • Parameters have changed to more closely follow OpenCV's options.
  • Publishing floating-point disparity image.
  • No longer publishing image_disparity, use stereo_view for visualization instead.
  • REMOVED deprecated stereoproc node and other obsolete files.


  • stereo_view shows color-mapped disparity image along with left/right.


  • New definition of DisparityImage. Disparities are now floating-point, not fixed-point. Better comments.
  • REMOVED all deprecated messages (RawStereo, DisparityInfo, StereoInfo).
  • Added script to convert RawStereo bags to Image & CameraInfo bags.


  • GUI improvements.


No TrackBacks

TrackBack URL:

Leave a comment

Find this blog and more at

Monthly Archives

About this Entry

This page contains a single entry by kwc published on January 11, 2010 9:19 PM.

laser_drivers 0.3.0 released was the previous entry in this blog.

diagnostics 0.5.0 released is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.