Abstract: The 3D tracking problem based on multi visual sensor systems is dealt with , of which one sensor that has low resolution and large field of view is responsible for the control of the global tracking and the other sensor, which controls the local tracking, has low resolution and small field of view .A visual Jacobian mapping from task space to vision space is developed to establish the visual servoing control equation that is used to predict the future location of the imaged object. By doing that the region of image to be processed is reduced and the processing speed is improved . Meanwhile a HCMAC neural network is fabricated to replace the complex calculation of the inverse Jacobian mapping. Simulation results show that the speed of image processing and the precision of tracking are improved by these measures.

Key words: Prediction HCMAC neural network, Servoing, Global-local 3D tracking Visual Jacobian mapping