According to the problem of object manufacturing point can not be captured directly in binocular stereo visual monitoring system, a method was proposed to indirectly estimate space position and pose and 3-D motion parameters of a rigid object from image information of a truncated cone-shape target. Some essential features, such as longitude lines, latitude lines and their intersection points of the side surface of truncated cone, should be exacted from pairs of stereo images, and their 3-D coordinates were reconstructed at every sampling time to estimate position and pose of truncated cone and object by using the fitting method of a truncated cone. Then the corresponding longitude lines, which were extracted from image sequences at two continuous sampling times, were used to estimate motion parameters of rigid in a sampling period by Kalman filter. The experimental results show that this method is feasible, convergent, stable and anti-noise. It can be applied to some special scenes in which images of manufacturing point can not be directly obtained, such as robot carving and medical operation etc.
