Single image surface measurement for robotics application

In many robotics applications, a high-speed algorithm for determining object location is an essential part of the performance of the system. The class of algorithms based on projecting a pattern on the object surface to determine three-dimensional surface measurements is considered. The method consists of first projecting a grid pattern on the surface of the object. An image of the object with the grid pattern is taken from an arbitrary location to obtain an image of the object which has been distorted by the surface of the object. Using six selected control vertex points and the distance from the camera to the object surface, the perspective transformation matrix of both the acquired image and the projected grid can be computed. The final three-dimensional object coordinates at each vertex point of the distorted grid can be computed from the image vertices and their corresponding perspective transformation matrices. An important aspect of this algorithm is that only a single image of the object is required to obtain the three-dimensional coordinates of the object surface. The significance of this technique is that it demonstrates that surface measurement can be performed by a single image. This technique also applies to featureless curved surfaces objects. The technique has direct application to visually guided robotic systems, medical treatment, and the measurement of surfaces for stress.