Cameras are hardware that generates image data and are widely used in consumer electronics, automobiles, security and other fields. A series of research and application fields have been derived around the camera, including traditional image processing and intelligent applications based on deep learning. At present, the camera is also an important hardware component in the automatic driving of the big fire, such as a fisheye camera for surround view, and a peripheral view camera for adas.
How does the camera achieve imaging? How are pixels related to objects in the real world? This belongs to the camera imaging problem, which is a classic research content in image processing. Based on this, image quality debugging, camera calibration, image transformation, stereo vision, monocular distance measurement and other research fields are derived.
In the current hot field of autonomous driving, camera calibration and stereo vision based on internal and external parameters of the camera are also basic introductory knowledge. Therefore, it is very important for researchers/engineers in the field of image processing to understand and master the imaging principles of cameras and the skills of camera calibration. If you are interested in the field of image processing, or want to get started in the field of stereo vision/autonomous driving, whether you are a college student/employed, the knowledge/skills of camera imaging principles and calibration are extremely important basic knowledge.
However, the content related to camera imaging and camera calibration includes knowledge such as camera imaging principles, matrix operations, optimization problems, etc. It is not easy to implement a camera calibration solution from scratch. "If you want to do a good job, you must first sharpen your tools." The emergence of OpenCV provides us with more efficient and convenient learning/research conditions.
OpenCV is a set of open source computer vision and machine learning software libraries that provide rich and stable image processing APIs. It is lightweight and efficient, and has high operating efficiency on Windows, Linux, Android and other platforms; therefore, it has a wide range of applications, including image segmentation, face recognition, motion detection and tracking, assisted driving and other fields. In short, familiarity with the use of OpenCV will greatly improve the efficiency of scientific research/work.
At the same time, for camera calibration, OpenCV integrates a set of calibration APIs used with checkerboards, which can efficiently realize the entire calibration function of monocular cameras. However, 1) Due to the high combination of camera calibration and camera imaging principles, it is still difficult to understand; 2) After completing camera calibration, how to use the internal and external parameters of the camera for application may not be particularly clear to beginners.
In view of this, the author has made a set of courses , hoping to provide some help. The main content of the course is to calibrate the monocular camera based on OpenCV, and at the same time cooperate with the target to realize the positioning of the target in the three-dimensional space. Through the study of this series of courses, you will master 1) the imaging principle of the camera; 2) the definition and function of the internal and external parameters of the camera; 3) the data collection and internal reference calibration of the camera on the PC end; 4) how to realize the measurement of the monocular camera distance .
The effect of the actual combat routine is shown in the figure below, which realizes the three-dimensional space positioning of the checkerboard based on the monocular camera . This is an operation that integrates camera imaging, corner detection, optimization solution, and image transformation, and perfectly applies the results of camera calibration. .
Going back to the course itself, the course structure is shown in the figure below. Course materials including PPT, documents, and source code are available for download and study.
Although the duration of this course is not long, the content is refined and the key points are prominent. Complete the camera calibration through practical exercises, and realize the 3D positioning of the checkerboard based on the monocular camera, which will greatly improve your understanding of the meaning and function of the internal and external parameters of the camera. At the same time, this is also the basic knowledge of stereo vision, and it will also help you to carry out in-depth study in fields such as SLAM/visual ADAS .
The course address is OpenCV actual combat monocular camera calibration and three-dimensional positioning--computer vision video tutorial-artificial intelligence-CSDN Programmer Training Institute
Students in need can find out, thank you~