image_geometry

`image_geometry` contains C++ and Python libraries for interpreting images geometrically. It interfaces the calibration parameters in sensor_msgs/CameraInfo messages with OpenCV functions such as image rectification, much as cv_bridge interfaces ROS sensor_msgs/Image with OpenCV data types.

Homepage: http://wiki.ros.org/image_geometry

image_geometry contains camera model classes that simplify interpreting images geometrically using the calibration parameters from sensor_msgs/CameraInfo messages. They may be efficiently updated in your image callback:

image_geometry::PinholeCameraModel model_;
void imageCb(const sensor_msgs::ImageConstPtr& raw_image,
             const sensor_msgs::CameraInfoConstPtr& cam_info)
{
  // Update the camera model (usually a no-op)
  model_.fromCameraInfo(cam_info);
  // Do processing...
}

Code API

image_geometry contains two classes:

image_geometry::PinholeCameraModel - models a pinhole camera with distortion.
image_geometry::StereoCameraModel - models a stereo pair of pinhole cameras.