Distance Measurement from Single Image Based on Circles

Computing distance from captured images is a common task in image analysis and scene understanding. Distance serves as building block for calculating other geometry information, such as area and volume. Besides directly measuring distance on the spot, it can be computed using projective geometry from scene images. Existing work on distance measurement consider point or line constraints. Since circles are common in daily life, we propose two methods of measuring distance using circles in one captured image. The first method deals with two separate circles, which can be coplanar or in parallel planes. The second one handles two concentric circles on the same plane, which offers higher accuracy than directly solving equations of concentric circles. Both methods are verified by experiments with simulated data and real images.