Known almost from the following: https: //www.zhihu.com/question/29448299/answer/102658379
Speaking from monocular vision, usually do visual recognition, measurement, and then went photographs, do various kinds of processing on the digital image, such as color processing, graying, filtering, edge detection, the Hough transform, to give the final desired feature, but please note, proceed to the pixel this step we just get interested in it ,
But how to convert pixels into the real world it? In other words, these things would happen to the remaining images were measured in order to get the data have practical significance and moderate it? ? ? ? ?
The meaning of existence camera calibration is to solve this pain egg problem! ! ! ! !
Continue to look through the camera calibration can be aware of the following:
1. external parameter matrix. Tell the real world point (world coordinate system) is how through the rotation and translation, and then fell another real world point (camera coordinate system) on;
2. the parameter matrix. Tell basis points above the 1 on how through the camera lens, and through electronic pinhole imaging and converted into pixels;
3. distortion matrix. Why not tell the position falls theory, but also TMD had a little deviation (displacement and deformation).
4. structural parameters. Tell how the camera has to do with the camera rotation and translation through to reach the current position. By structural parameters, you can put the relationship between each pixel camera left and right of the image acquired using mathematical language to describe, to ensure that two banana hi desire in our state.
Overall: Who camera calibration by finding objects into images and mathematical relationships in the real world, and find out the amount of contact in order to achieve the purpose of measuring the actual data from the image! ! ! !