Biometric recognition system such as facial recognition system was widely developed over the past few years. Facial recognition system is commonly used in security system to allow user to protect their privilege. The normal security like key or password is no longer relevant as people prefer an easier and flexible way. Therefore, this paper presents a better and easier way of security system that can recognize the user successfully and give the matching percentage. By using Radial Basis Function Neural Network in MATLAB, a face recognition system can be created. The RBF system will be trained by data as reference, input image will undergo the same process and the data obtained will be used to match with the data in the RBF to obtain the matching percentage. A suitable matching percentage reference was chosen from this analysis as the minimum require matching to access the security system where error rate is one of the main concerns where it is the unwanted result that might occur. Different threshold number, spread value, and sizes of dimension also tested, the differences on the output matching result were observed. By using the microcontroller to control a relay to control the magnetic door lock, the system was able to successfully control the door lock.