Research On The Camera Calibration Algorithm

Vision is the important way for humans to understand the outer world. With the emergence and development of digital technology, computer vision has gradually become one of the most active and challenging field of study. As the important part of the computer vision on obtain three-dimensional information, camera calibration not only lays a good foundation for the visual inspection and stereo vision research, but also improves the accuracy of image measurement and motion analysis. Therefore, research on the camera calibration has important theoretical study and practical value.The relative theory and algorithms of camera calibration are researched in this thesis. In order to obtain more accurate camera parameters, the existing algorithms are improved to build the relationship between the 3D object and the 2D image and lay a foundation for the subsequent application of vision. The main works are as follows:Firstly, the foundational knowledge of camera calibration is introduced, and several major calibration methods are introduced and analyzed from two directions of traditional calibration method and self-calibration method.Secondly, in order to obtain more accurate results of calibration for traditional calibration methods, an improved camera calibration method based on the adaptive harmony search algorithm is proposed. Because the harmony search algorithm is not sensitive and needs less math, these characteristics could be used to change this algorithm’s control parameters into adaptive way so that the objective function is not limited to local minima. Then the adaptive harmony search algorithm is combined with the process of camera calibration to optimize the camera’s intrinsic parameters and the coefficient of radial distortion. Experimental result shows the accuracy and robustness of algorithm.Thirdly, an improved camera self-calibration method based on the geometry of vanishing point is proposed. This method not only preprocesses the detection lines by taking advantage of the characteristics of the template straight line, improving the calculation accuracy of the vanishing points, but also establishes the constraint equation by taking advantage of the geometry of vanishing point, simplifying the process of solving intrinsic parameters. Simulation experiment shows the accuracy and robustness of algorithm, and the real experiment also proves its effectiveness in the follow application.Finally, two improved algorithms are applied to visual detection. Experiment through take 10 pictures of work piece under the same environment and different conditions to compare its detection size, and the result shows the accuracy and effectiveness of the proposed method.