Locating destination address block in Korean mail images

We propose a method of locating destination address blocks in both machine-printed and handwritten Korean mail images. The proposed method extracts connected components from a binarized mail image, generates rough text lines by merging them, and then groups the text lines into nine clusters. A destination address block is determined by selecting some clusters. Considering the geometric characteristics of information on Korean mails, we segment a mail image into nine areas with an equal size. The nine clusters are initialized with the center coordinate of each area. A modified Manhattan distance is used to calculate the distance between a text line and a cluster. We modified the distance function on which the aspect ratio of mail image could be reflected. The experiment done with live Korean mail images has demonstrated the superiority of the proposed method. The success rate for the 1,988 testing images was about 93.56%.