A new solution to the problem of range identification in perspective vision systems

Abstract

A new solution to the problem of range identification for perspective vision systems is proposed. These systems arise in machine vision problems, where the position of an object moving in the three-dimensional space has to be identified through two-dimensional images obtained from a single camera. The proposed identifier yields asymptotic estimates of the object coordinates and is significantly simpler than existing designs. Moreover, it can be easily tuned to achieve the desired convergence rate. Simulations are provided demonstrating the enhanced performance of the proposed scheme and its robustness to measurement noise.