Hough technique for 3D object recognition

Advances in imaging techniques make it efficient to extract information such as the length and orientation of a straight edge in 3D space. An algorithm using the Hough technique is proposed for the recognition of objects which can be represented or characterised by a set of vertices and straight lines. Assuming the scale is known a priori, the orientation and translation of an object are determined sequentially in two stages. The method is robust and can handle objects with curved or irregularly shaped surfaces, and surface reconstruction is not required.