The importance of camera calibration and distortion correction to obtain measurements with video surveillance systems

Video surveillance systems are commonly used as important sources of quantitative information but from the acquired images it is possible to obtain a large amount of metric information. Yet, different methodological issues must be considered in order to perform accurate measurements using images. The most important one is the camera calibration, which is the estimation of the parameters defining the camera model. One of the most used camera calibration method is the Zhang's method, that allows the estimation of the linear parameters of the camera model. This method is very diffused as it requires a simple setup and it allows to calibrate cameras using a simple and fast procedure, but it does not consider lenses distortions, that must be taken into account with short focal lenses, commonly used in video surveillance systems. In order to perform accurate measurements, the linear camera model and the Zhang's method are improved in order to take nonlinear parameters into account and compensate the distortion contribute. In this paper we first describe the pinhole camera model that considers cameras as central projection systems. After a brief introduction to the camera calibration process and in particular the Zhang's method, we give a description of the different types of lens distortions and the techniques used for the distortion compensation. At the end some numerical example are shown in order to demonstrate the importance of the distortion compensation to obtain accurate measurements.