Environmental applications of camera images calibrated by means of the Levenberg–Marquardt method

Abstract

Even though different authors have presented procedures for camera calibration in environmental video monitoring, improvements in the robustness and accuracy of the calibration procedure are always desired and in this work the Levenberg–Marquardt method is included in the camera calibration process for environmental video monitoring images as a way to improve the robustness of the camera calibration when a low number of control points is available without using laboratory measurements. The Pinhole model and the Levenberg–Marquardt method are briefly described and a four step camera calibration procedure using them is presented. This procedure allows users to use ground control points to estimate all the Pinhole model parameters, including the lens distortion parameters and its implementation results with laboratory data are compared with the results presented by other authors. The procedure is also tested with field data obtained with cameras directed toward the beaches of the city of Cartagena, Colombia. The results show that the procedure is robust enough to be used when just a low number of control points are available, even though a large number of GCP is recommended to obtain high accuracy.