******************
Camera Calibration
******************


Camera calibration is the **FIRST** step for any practical vision applications, aiming at removing image distortions caused by adopted camera itself.

- On the one hand, most `image files (.png, .jpg, .tiff, .bmp, etc.) <https://en.wikipedia.org/wiki/Image_file_formats>`_ and `video files (.ogg, .mp4, .h264, .avi, etc.) <https://en.wikipedia.org/wiki/Video_file_format>`_ are taken by cameras for further processing off-line.
- On the other hand, a lot real-time applications, such as `video surveillance <https://www.sciencedirect.com/topics/computer-science/video-surveillance>`_ and `vision inspection <https://www.visiononline.org/vision-resources-details.cfm/vision-resources/What-Are-Vision-Inspection-Systems/content_id/4586>`_, `live video streams <https://en.wikipedia.org/wiki/Live_streaming>`_ captured from cameras are to be processed on the fly.

All these captured `image files <https://en.wikipedia.org/wiki/Image_file_formats>`_, `video files <https://en.wikipedia.org/wiki/Video_file_format>`_, and `live video streams <https://en.wikipedia.org/wiki/Live_streaming>`_ are expected to be **undistorted**. Most of the high-quality cameras have been precisely calibrated before being launched onto the market. Namely, the `camera intrinsic parameters <https://en.wikipedia.org/wiki/Camera_resectioning#Intrinsic_parameters>`_ have been provided by the manufacturer. However, professional computer vision developers always carry out their own calibration to guarantee a better image and video quality.

This chapter is composed of 6 sections. First of all, a couple of popular types of consumer-level cameras are to be briefly introduced. Afterwards, we'll discuss how to use `OpenCV <https://opencv.org/>`_ to calibrate monocular vision cameras with narrow-angle lens, monocular vision cameras with fisheye lens, and how to estimate camera postures after camera calibration. As an extension, binocular camera calibration and Kinect calibration are to be discussed by using `MRPT <https://www.mrpt.org/>`_.

By the end of this chapter, you will be able to generate **calibration XML files** for a couple of monocular vision cameras. And, you’ll be able to use these XML files to generate undistorted images, and further estimate camera postures. Finally, you’ll learn how to calibrate binocular cameras and `Kinect <https://developer.microsoft.com/en-us/windows/kinect/>`_ .


.. toctree::
    chapter2/01_popular_types_of_cameras
    chapter2/02_calibrate_monocular_vision_cameras_with_narrow_angle_lens
    chapter2/03_calibrate_monocular_vision_cameras_with_fisheye_lens
    chapter2/04_pose_estimation_for_monocular_vision_camera
    chapter2/05_stereo_vision_calibration
    chapter2/06_kinect_calibration