KCPE-SMDI:Key Contour Points Extraction Method Based on Stroke Multi-Dimensional Information for Chinese Character Vectorization

The vectorization of Chinese characters is very important for the industrial design and manufacture, the digitization of characters, and the generation of handwritten font libraries. However, when most of the general vectorization methods and Chinese character vectorization methods are vectorizing Chinese characters, the vector Chinese characters will have edge bending problems due to the jitter noise at the edges of the Chinese characters. If we remove the jitter noise on the edge by reducing the overall threshold, some small features of the Chinese characters will be lost at the same time. And the key points of the generated Chinese character vector graphics will be redundant. Therefore, this paper proposes key contour points extraction method based on stroke multi-dimensional information for Chinese character vectorization (KCPE-SMDI). Through multi-dimensional information such as Chinese character stroke type, Chinese character key skeleton points and stroke width, the key contour points of Chinese characters are extracted, and the jitter noise contour points and redundant contour points are removed. Using multiple groups of Chinese character raster images, we compared the vectorization results of our method(KCPE-SMDI) with four different methods. The experiments showed that our method not only reduces the number of contour points by 8.0419% compared with the current method of generating the minimum contour points(Wang et al. method), but also solves the problem of edge bending of vectorized Chinese characters and maintains the feature of the original Chinese characters.