The Research And Application Of Skeleton Extraction Algorithm

Skeleton extraction is a fundamental problem with many applications in the field of computer graphics and visualization, which are widely used for many tasks including target detection, image processing, pattern recognition, virtual navigation and computer animation, etc. The3D or2D model used by most existing skeleton extraction methods makes use of a volumetric discrete representation or mesh surface representation. In recent years, the algorithm for skeleton extraction from point clouds has become a hot theme.Based on previous research, we present:(1) Improved algorithm of Chinese characters skeleton extraction. By learning the algorithm of curve skeleton extraction from incomplete point cloud, we use some algorithms of it in our skeleton extraction algorithm. In our algorithm, firstly, detect the edge of the Chinese character, the edge can be seen as a simple point clouds in2D.The point clouds are stored in the array, get the points by specific step, then calculate inscribed circle within the specific step iteratively, get the skeletons of the Chinese character after connecting every center of the inscribed circle.(2) Texture boundary extraction based on morphological skeleton extraction algorithm. We employ a standard deviation histogram as a statistic feature for texture analysis, and similarity among multiple histograms is measured using diagonal χ2-statistic method. According to the similarity among standard deviation histograms, we adopt skeleton extracting method based on mathematical morphology to obtain the final texture region boundary.The skeleton extraction algorithm of Chinese characters based on the prototype algorithm has been improved. The prototype of Chinese characters skeleton extraction algorithm can only use for TrueType fonts, and this method not only for TrueType fonts, but also for other fonts, such as Official script, bold, new Chinese Wei, etc. In the algorithm of texture boundary extraction based on morphological skeleton extraction, we proposed the concept of standard deviation histogram, according to the similarity among standard deviation histograms and morphological skeleton extraction, we constructed a new texture boundary extraction method, and it has high computational efficiency.