Analysis of image sequences to determine rotational and translational parameters

An approach for analyzing image sequences for motion parameter estimation is presented. A sequence of an arbitrary number of image frames is utilized to determine rotational and translational parameters. A dynamic scene model is developed in which image sequences are processed as a temporally correlated complex. The object motion is represented as a discrete-time time-varying system. The measurement consists of a sequence of image coordinates of three or more feature points in each frame. Using this model, measurement of the position of the object in a set of consecutive frames permits the estimation of motion as a function of time. An iterative parameter estimation technique is used to minimize the projection error. The technique is based on results from optimal control theory. Motion parameters are estimated from the sequences of image correspondences by modeling the motion dynamics using motion transformation and viewing projection. This methodology is suitable for processing a long sequence in situations where a high rate of imagery is available. Results are presented for general rigid-body motion in the context of synthesized images and real robot images