To implement a character recognition system, a good algorithm has to be selected among the conventional ones. In this thesis, three conventional algorithms are implemented and compared. The algorithms are a method using moments, a method using outermost point as a structural feature, and a method using characteristic loci. The characters used in this experiment are Helvetica, Munjo, and Times font which are used widely. As a result, the algorithm using moments, and the algorithm using outermost point as structural feature are better in recognition rate. But, if there is a rotation of character or noise, the algorithm using outermost point has the best result. Therefore, the method using outermost point can be recommended in various applications.