Determining the kinematic parameters of a moving imaging sensor by processing spatial and temporal intensity changes

A method is presented that determines the motion parameters of an imaging sensor by evaluating the spatial and temporal intensity changes in the image plane. It is not necessary to compute the optical flow. If the environment in the field of view is described parametrically, constraint equations between the kinematic parameters and the intensity changes can be derived. Linear solutions of these equations are the so-called pure parameters, which are nonlinear functions of the motion parameters. These quantities are the roll, pitch, and yaw rate, the angle of attack and the angle of yaw, and orientation parameters. They can be obtained directly by solving a 3 X 3 eigenvalue problem. The algorithm was implemented and applied to artificial and natural image sequences. An error and covariances analysis shows that the method is robust and accurate enough for autonomous navigation.