Research On An Automatic Camera Calibration Method Based On Minimum Convex Hull Theory

Vision is an important means which biology obtain outside information, especially for humans. Information about the outside world that people acquires of 80% is provided by the visual organ. Along with the computer become more and more powerful, computer vision has made great development, and is applied widely in various fields. Among them, based on the two-dimensional image to get structure of various elements in the image, namely three-dimensional reconstruction. Threedimensional reconstruction is one of important research tasks of intelligent vision,three-dimensional reconstruction has been widely used in various field. Camera calibration is the basis link of three-dimensional reconstruction, which is also an indispensable step. The extraction of feature points take most time in whole calibration processing, The speed of feature point extraction affect the speed of the camera calibration directly. Therefore researched feature points extraction algorithm which have high automatic level, that achieve camera calibration fast and precisely, it has important theoretical meaning and practical value.There are many shortages about existing camera calibration kit, such as high degree of manual intervention, etc. Aiming at this, this paper based on the research of the camera calibration’s knowledge, with the typical calibration object--checkerboard template as the research object, this paper presents a new corner detection algorithm based on morphological processing and a new automatic corner sorting algorithm based on the theory of the minimum convex hull. Calculation of camera parameters are completed via using Zhang’s calibration model.Through the research of related technologies, use the gaussian filter to smooth the images, method about determining the threshold of binarization image have been designed, the adaptive binary image segmentation have been realized. The checkerboard corner detection algorithm based on image morphology processing have been designed,checkerboard corners have been located in pixel level successfully. Using the theory ofminimum convex hull to identified four poles of checkerboard automatically, through the calculation of homography matrix, the automatic sorting of corners have been completed. Through analyzing and comparing with classical corner location algorithm,gray gradient algorithm have been used to obtain the subpixel corner of checkerboard,thus the precise location of corners have been realized. Combining with the classical Zhang’s calibration model, the calculation of parameters have been completed, the precise and fast calibration of the camera have been accomplished. Experimental results showed that the automatic calibration method is simple, the results is similar to the traditional calibration results, but time-consuming is less, and it has good accuracy and reliability in both indoor and outdoor environment.