In this study, the nonlinear control of image-moment based visual servoing has been considered. The combination of image-moments has been utilized as image features for tracking appropriate image of the target object. The control signal was calculated by comparison between the current and target image features. Visual servoing is utilized widely in industry and medicine. Therefore, increasing domain of attraction, robustness, and accuracy of the tracking are the main issues in this field. To achieve these purposes, combination of sliding mode control with image moments has been proposed. Based on the selected image features and modeling, proportional integral type sliding surface has been suggested, and due to nature of the system, the control signal type is velocity. The proposed method has been implemented on a five dof industrial robot, and the performance of this method has been compared to the nonlinear control of kernel based visual servoing. The results display an increase in the domain of attraction, velocity, and accuracy with respect to kernel based visual servoing. In situations, where some part of the object image is out of camera field of view, unmodeled distortion is imposed to the system. The conventional controller could not cope with distortion and failed to convergence. The nonlinear kernel based visual servoing controls each dof separately, as a result, it cannot converge the system to target point in this special case. However, the proposed method in this study is robust to uncertainty and controls four dof of camera motion simultaneously; therefore it can manage this situation. The stability of the controller is analyzed by the Lyapunov theory.