Objects Positioning in Water Surface From a Single Image

Abstract

[Abstract] Boats positioning is an important task required for autonomous navigation and control. This paper proposes a new method for estimating real positions (x-y coordinates) of small boats in the 3D scene from a single image. These positions are obtained with respect to a world reference system by the only means of using rows and columns of pixels where the boat is located in the image. A single camera is needed without requiring knowledge of its intrinsic and extrinsic parameters. The proposed approach avoids the use of more complex and sophisticated systems such as GPS. The method is valid for any object on a planar surface with the only constraint that the camera must be on a fixed position. It has been proven with a differential GPS of high precision. The main contribution is made on the computation of both x and y coordinates from a reference system. This method is also valid for objects in other planar surfaces, such as flat fields, crops, indoor floors or horizontal roads.