Visual Positioning Using Four-Point Planar Patterns

De Xu, Youfu Li, Min Tan

Abstract: The existence of multi-resolutions for the perspective-n-point (PnP) problem makes the position-based visual control based on PnP positioning method difficult for practical applications. In this paper, two simple four-point patterns are designed to ensure unique solution for the PnP problem. One is a triangle planar pattern, which consists of three points and their centroid. Another is quadrangle planar pattern. A recursive least square algorithm with a forgetting factor is introduced to improve the solution accuracy. The positioning accuracy based on PnP positioning method is investigated for the two patterns. The method presented in this paper has the advantages of minimum set of known points providing unique solution and linear pose estimation. Experiments are performed to verify the effectiveness of the proposed patterns and method.