A new technique for geometry based visual depth estimation for uncalibrated camera

摘要

The most challenging issue in the field of visual control system is the accurate estimation of depth upon which all other parameters of pose estimation may depend. Here we have introduced a new approach for the depth estimation problem of a visual control system with a moving camera and stationary target. A unique visual pattern is designed to achieve the desired depth of the vision based system. In this paper the unknown depths of the special geometrical feature points are estimated from the pre-calculated depth calibrated equations, whose affects are then combined to acquire the desired goal. This specialized technique doesn't require any conventional camera calibration parameters, e.g., camera intrinsic parameters; rather it works on the specialized and accurate depth calculating calibration equations described in this paper. Simulation and experimental applications are presented which verify the performance of the proposed technique. The vision system, i.e., a wireless camera, pointing downwards, is mounted with a mechanical assembly which can move laterally as well as up and down. The required algorithm is designed in MATLAB which is capable of reading real-time images and processing it for proper lateral position and desired depths calculation at particular instants.