ABSTRACT
This chapter describes the relevant camera models as well as overviews the so-called perspective transformation matrix method. It describes Tsai's radial alignment constraint (RAC) method for camera calibration. The chapter presents a simplified RAC-based algorithm, a procedure that handles a near singular case, calibration simulation and experimental results, and outlines Weng's two-phase nonlinear optimization approach. The camera calibration problem is to identify the unknown camera model coefficients given the known/measured data. The idea behind the simplification of Tsai's RAC algorithm is that during the first stage of camera calibration, if one selects only calibration points which lie along the x and y axes of the world coordinate system, the camera radial alignment constraint equation becomes greatly simplified. The cost function associated with a camera calibration process is in general non-convex. Camera calibration requires the computation of the projection of the center of each calibration point onto the image plane.
