For the hand-eye calibration of a dual robot measurement system, a method based on machine vision to calculate the target robot flange pose and center coordinate was presented. By moving the target robot flange to a proper pose to take an image by the camera and extract the ellipse contour of the flange in the image to calculate the flange pose and its circle center data, the coordinate transform H1 between camera coordinate system and flange coordinate system was obtained by using the location of pinhole on the flange. Then,the coordinate transforms between flange coordinate system and robot base coordinate systems ,namely, H2 and H4,are gotten,respectivity, from the robot controllers. Furthermore,the coordinate system transform H3 between two robots were derived from single axis movements of robots, so that a hand-eye expression HCG was obtained to calculate the hand-eye coordinate transform. Finally,by moving the target flange to some coplanar poses to take their images, the calibration accuracy was improved by image fusion. Experimental results indicate that the calibration precisions of single image and coplanar poses using image fusion are 0.345° and 0.187°,re-spectively. It can satisfy the the requirements of dual robot systems for vision guiding measurement.