A novel approach of segmenting touching and kerned characters

摘要

Character segmentation is a critical step of OCR system. In this paper we discussed segmentation approaches of touching and kerned characters. A non-linear segmentation path-based algorithm for segmenting touching and kerned characters is put forward. First, touching and kerned characters are extracted and segregated with other characters by using character projections and recognition results. Then in order to find the non-linear segmentation path of touching and kerned characters, a heuristic method seeking minimal-penalty curved cut is used. Finally several regions belonging to the same character are merged. Experiment results show that the proposed method achieves high segmentation rate in documents written in both English and Japanese characters.