This paper presents an omni-directional vision-based control scheme for the car-like mobile robot (CLMR), which possesses autonomous parking capability. At first, we apply the image processing technique to the omni-directional images to filter out the noise and find the characteristic points of the parking lot. From the image information, one can estimate the position of the CLMR in the parking space and figure out a feasible reference path. Then, we propose a fuzzy logical control to manipulate the steering wheel such that it can execute parallel-parking missions. Finally, the real-time experiment demonstrates the feasibility of the developed schemes.