Oh Seok Kwon , Byeong Yong Kim

The Transactions of the Korea Information Processing Society (1994 ~ 2000), Vol. 7, No. 10, pp. 3086-3097, Oct. 2000
10.3745/KIPSTE.2000.7.10.3086,   PDF Download:

Abstract

In this paper, we proposed a method of the text area segmentation and layout vectorization of off-line handwritten forms. We applied DRC algorithm to the scanned image to protect data loss during binarization and thinning of the image. To detect the skew angle of the image, we applied the Hough transform to the image and estimated the angle of the skew. After correcting the skew angle, we extracted the line components of the image, which constitute the frame of the form. The character areas of the image are calculated based on white-pixel connected components extraction method and the vectors of the extracted line components are estimated by sorting, merging and refinements. In order to show the abilities of the proposed method, experiments with two kinds of forms that were written by 25 people are performed. One was drawn with a ruler and the other was drawn without a ruler and all of them were written with freely chosen writing tools. As a result, we got 666 vectors without preprocessing and 746 vectors with preprocessing among 750 vectors respectively, which showed the effectiveness of the method.

Statistics
Cite this article