Abstract: The result of segmented digital ink text in Chinese includes three levels of objects: characters, lines and paragraphs. Characters form a significant percentage in the result and the situations of them are always complex. Automatic methods hardly provide completely correct result of extracted characters. So the result needs to be modified by human-computer interactive operation. Optimized visualization can improve the efficiency of modification. With the modification of the errors of extracted characters, according to the adjacency and overlapping among characters, this paper proposes a self-adaptive visualization. The approach gets rectangular bounding box of characters; if they are overlapping, the approach changes to convex hull to visualize the segmented characters; if they are still overlapping, the approach uses different colors to perform the segmented characters. Tested on many sorts of extracted characters from digital ink text in Chinese, the approach is effective.

Key words: digital ink, visualization, character extraction

摘要： 中文数字墨水文本的分割结果包含单字、文本行和段落三个层次对象，单字在其中占有较大比例，情况复杂。使用自动的分割方法难以提供完全正确的单字提取结果，这时就需要进行人机交互校正单字提取结果。优化的可视化方法可以在人机交互时大大提高校正效率。面向交互校正错误的单字提取结果，针对单字结果间的邻近和重叠等情况，给出了一种自适应的可视化方法。该方法先生成单字的正放最小外接矩形，如果相邻矩形重叠，则改用凸包，仍然重叠，则给单字结果加上颜色。对多种数字墨水文本的单字提取结果进行可视化表示，取得了较好的效果。

关键词: 数字墨水, 可视化方法, 单字提取