ABSTRACT
This chapter introduces conventions for camera coordinate frames in robot operating system (ROS), camera calibration, and low-level image-processing operations with OpenCV. In addition to accommodating OpenCV classes, ROS also has customized publish and subscribe functions for working with images. The image transport framework in ROS comprises both classes and nodes for managing transmission of images. ROS provides support for intrinsic camera calibration through the camera calibration package. Calibration can be as simple as waving the calibration target in front of the camera while running the calibration process. The simple camera model is extended to dual (stereo) cameras in the model file multicameramodel.xacro in package simple camera model. Stereo cameras offer the opportunity to infer 3-D from multiple images. Commonly, stereo cameras will be mounted such that the transform between their respective optical frames is constant. The value of calibrating stereo cameras is that it becomes possible to infer 3-D coordinates of points of interest using multiple camera views.
