Image-based control is one of robot control methods using a camera and it can control a robot indirectly by controlling only image information on an image plane without using the robot states. In this paper, an image-based trajectory tracking method is described for a nonholonomic mobile robot. Especially, a fuzzy controller is adopted to regard the robot speed, usually assumed to be constant, as time varying one. The controller was tested on simulation experiments. The results of the experiments show the usability of the designed fuzzy controller, giving the improvement of the control performance, such as the reduction of overshoot, the shortening of convergence time, etc. Additionally, the impact of control parameters on the robot behavior is proved in the simulations.