Hand-eye coordination play vital role in object grasping task of humanoid robot. A method proposed for hand-eye coordination addresses four distinct problems i) Integration of vision control and motion control ii) hand-eye coordinate transformation iii) orientation of hand with target object iv) hand-eye localization. For hand localization and to limit the orientation problem a blue colored rectangle block was pasted on the robot hand. The blue colored rectangle block four sides were detected by applying karhunen loeve (K-L) transform, Hough transform and Least Square Error (LSE) methods. For hand-eye coordination a hand-eye model was developed to adjust the position and orientation error for object grasping task. In the hand-eye model world coordinate system was set on right shoulder joint. The position and orientation of the hand was obtained through coordinate transformation from left eye coordinate system to end-effector coordinate system. The results presented in this paper justify the effectiveness of the proposed method.