Demonstrating a practical approach to coordinate transformation from two-dimensional plane to a pseudo three-dimensional plane for controlling vision-based systems

Abstract

A coordinate transformation image processing technique was devised to enable the interactive control of vision-based systems. This procedure transformed the two-dimensional coordinate plane of the video pixels to a pseudo three-dimensional plane that coincides with the actual coordinates of the system’s environment as seen through a camera. The coordinate transformation image processing technique of the system was evaluated on a mobile robot platform and the results showed a 0.36% error for the transformed y-axis and 2.05% error for the transformed x-axis.