3D shape of objects can provide richer information for detecting, tracking or identifying the objects than a single 2D image of them. We tackle the 3D shape and texture reconstruction of an object moving in a widespread space using multi-viewpoint active cameras. Considering 3D shape and texture reconstruction, the problem in existing tracking methods using active cameras is that they cannot calibrate the active cameras accurately. We propose a cell-based tracking method that can produce multi-viewpoint images and accurate camera parameters for every frame. Our idea is to divide the space into cells and perform active camera control and calibration based on the cells. We demonstrate the performance of our method by simulation. ©2009 IEEE.