A new method for baseline extraction of Manchu word

Abstract

The baseline of Manchu word extracted using project profile histogram often shifts from its accuracy position, and leads to segmentation errors. This paper proposes a new method for baseline extraction of Manchu word after studying and fusing morphology, Hough transform and max-run-length-proportion method. Experiments on 400 Manchu word images including 4 different Manchu fonts with 6 font sizes show that the extraction method proposed in this paper is effective and has strong adaptability. The extraction accuracy reaches 100% which is higher than other 2 compared methods. The new baseline extraction method is also useful for Mongolian, Arabic, etc. characters which also have baselines.

Keywords:

• Manchu
• Baseline
• Morphology
• Thinning
• Hough transform
• Max-run-lengthproportion method