In many industry applications, industrial manipulators are frequently used to perform repetitive missions such as pick-and-place tasks for packing and placing objects. In general, the position information is essential when performing pick-and-place tasks. One way to obtain position information is to exploit computer vision techniques to retrieve 3D position information of objects through object images. Therefore, the relationship between object images and 3D positions of objects, and the relationship between object poses and object grasping methods are important research topics deserving more investigations. As a result, this paper focuses on research topics related to vision based object grasping of industrial manipulators, including object pose estimation, hand-eye calibration and coordinate transformation. In this paper, an eye-to-hand stereo camera system is used to retrieve object's 3D position information that is needed in pick-and-place tasks performed by an industrial manipulator. Experimental results indicate that an industrial manipulator equipped with the developed vision system can successfully perform automatic pick-and-place tasks.