The problem of uncertain nonholonomic mobile robot tracking control based on visual servo is studied. Based on visual feedback and state input transformation, an uncertain model for a class of nonholonomic kinematic systems is proposed. In three different cases, adaptive dynamic feedback controllers are designed to track the desired trajectory of the uncertain system by two new transformations. Adaptive dynamic feedback controllers are designed to track the desire trajectory of the uncertain system for two cases. The convergence of the erroe system is proved rigorously using the extended barbalat theorem and the Lyapunov method. Simulation results demonstrate the effectiveness of the proposed method.