Word segmentation of printed text lines based on gap clustering and special symbol detection

This paper proposes a word segmentation method for machine-printed text lines. It utilizes gaps and special symbols as delimiters between words. A gap clustering technique is used to identify the gaps between words regardless of the gap-size variations among different document images. Next a special symbol detection technique is applied to find two types of special symbols lying between words. An experiment with 1,675 text lines in 100 different English and Korean documents shows that the proposed method achieves a high accuracy of word segmentation.