Multilevel Chinese Inscription Completion Based on Strokes and Radicals

Chinese calligraphy is the indispensable part of the traditional culture. Facing the present situation that ancient calligraphy existing in form of inscriptions is destroyed, we propose a complete technique for inscription corrupted by weathering or corrosion. Firstly, we use histogram projection to get single Chinese character. Then, Chinese characters of inscription should be analyzed to extract strokes and radicals as a set of template for complete procession. Finally, we use the highest similarity model found through Improving Shape Context and IS-Matching in template set as the information source to repair the damaged region of inscription. The experimental results show that this method can be applied to complete the damaged inscription on the basis of preserving original writing style.