A novel calibration method is proposed for active vision systems. Projector projects colored concentric circles to a calibration plane

on which an array of concentric circles is printed. The centers of the circles in image frame are identified with geometry constraints

with which the homographies between the calibration plane

image plane and projector plane are estimated. The method uses color information to separate the image of projected circles and that of the circle patterns on calibration plane

which reduces the number of required images and simplifies the calibration procedure. To avoid manual intervention

cross-ratio invariance is used to match concentric circles automatically. Experimental results prove that the algorithm is simple and efficient.