Progressive Chinese calligraphy stroke extraction

Calligraphy works are composed of Chinese characters, and Chinese characters are composed of strokes. To extract calligraphy strokes can help recognizing meanings and calligraphic style. However, calligraphy strokes have overlapped regions, even may exist distortion due to a hand tremor in write process. The above problems make the calligraphy strokes extraction harder. This paper proposes a progressive stroke extraction method by muti-turn algorithm. And it has solved: Segmenting single characters from calligraphy pages; Composing skeleton strokes based on trend of segmented skeleton strokes; Detecting and correcting the abnormal skeleton strokes; Predicting the writing trend. In terms of regular script. In the data set of 1250 characters, the number of correctly stroke extracted Chinese characters is 1216, and the accuracy rate is 97.3%.