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.) and video files (.ogg, .mp4, .h264, .avi, etc.) are taken by cameras for further processing off-line.
On the other hand, a lot real-time applications, such as video surveillance and vision inspection, live video streams captured from cameras are to be processed on the fly.
All these captured image files, video files, and live video streams 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 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 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.
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 .
- Popular Types of Consumer-Level Cameras
- Calibrate Monocular Vision Cameras With Narrow-Angle Lens
- Calibrate Monocular Vision Cameras With Fisheye Lens
- Pose Estimation For Monocular Vision Cameras
- Stereo Vision Calibration
- Kinect Calibration