In this paper, we propose an efficient algorithm which recognizes the Korean characters. The recognition procedure which is proposed in this paper is as following. First, binary image is obtained from input image and its boundary is represented by the chain codes and then vertices are extracted from the chain codes. Input patterns are recognized using the relation among the vertices. If input patterns are connected, recognition procedure is done after elementary patterns are separated from input patterns. Finally, we rearrange elementary patterns as the writting sequence. The experiment is carried out for 600 characters used daily, and average recognition rate of 95 % is obtained. Since the proposed method do not use the thinning process, recognition time is reduced. The experimental results show the effectiveness of the proposed method.