Image measurement is required for mobile robots with vision. Stereo vision is widely used for this purpose. An image measurement using stereo vision can have a correspondence problem when there are similar objects in the environment. In order to avoid such a problem, a zooming measurement using monocular vision is applied. From this technique, it is possible to obtain the 3D position of an object by referring to two 2D images of it at distinct focus (view) points on the optical axis. The objects on the 2D images dynamically change with zooming. Image measurements on the digital images have uncertain regions depending on the disparity and the image resolution. The disparity is dynamically changed in the zooming measurement. Therefore, it is necessary to evaluate uncertain regions with every disparity.
This paper proposes an evaluation method for uncertain regions of the zooming measurement. In the proposed method, the relationship between the disparity and the uncertain regions is shown in the experiments by using the actual zooming camera. The zooming measurement based on the evaluation is achieved.