作者: C. Lawrence Zitnick
关键词: Beautification 、 Matching (statistics) 、 Stylus 、 Speech recognition 、 Artificial intelligence 、 Handwriting 、 Sketch 、 Computer science 、 Handwriting recognition 、 Security token 、 Natural language processing
摘要: In this paper, we propose a general purpose approach to handwriting beautification using online input from stylus. Given sample of writings, drawings, or sketches the same user, our method improves user's strokes in real-time as they are drawn. Our relies on one main insight. The appearance average multiple instances written word shape is better than most individual instances. We utilize observation two-stage approach. First, an efficient for finding matching sets stroke samples called tokens potentially large database writings user. Second, refine recently by averaging them with tokens. works without recognition, and does not require predefined letters, words, shapes. results show improved wide range writing styles drawings.