State observer for a class of nonlinear systems and its application to machine vision

In this note, we consider the state observer problem for a class of nonlinear systems which are usually encountered in the machine vision study. The formulation of the state observer is motivated by the sliding mode methods and adaptive control techniques. The proposed observer is applied to the identification problems of the motion parameters and space position of a moving object by using the perspective observation of a single point. It is clarified that the rotation parameters can be observed by using the observation of one camera, and the position and translation parameters cannot be observed by using one camera and must appeal to stereo vision. Simulation results show that the proposed algorithm is effective.