Three-dimensional Coordinate Calibration of PTZ Camera

Abstract

Traditional image surveillance is limited by camera picture range and position of installation, often resulting in ineffective coverage of a monitored area. These blind areas create dead spaces that pose a surveillance security risk. PTZ cameras are installed with movable, multi-directional lenses, so they are able to track moving targets. Because of this characteristic, they have been heavily adopted over the past few years. To achieve this functionality, it is necessary to calibrate the camera and the monitoring system to achieve orientation. This paper presents image coordinates gathered via observation of elliptical distortion resulting in divergent space-plane and camera angles. In the calibration process, the first step is identifying camera parameters, directing them against the target subject ellipse, then using the least squares method to detect and measure elliptic-related data. Known intrinsic parameters and elliptic parameters can then be pushed through a transformational matrix that changes an image coordinate system to a world coordinate system, while camera and world divergent coordinates lay aside target subject ellipse. Topographical changes can be detected via these elliptic distorted angles. Research shows that this study's method of using a sphere to calibrate target subjects not only achieves space calibration but also can reveal space topographical undulations for more precise camera and space calibrations.