As the stroke constitutional complexity of shapes and texts is different, this paper proposed a shape and text separation method based on the calculation of the entropy of the strokes. Because the entropy of a document may vary with stroke size, adaptive resampling was introduced to handle different writing stroke sizes. In addition, the paper employed a symmetrical judgement mechanism to handle the separation of texts and shapes with equivalent stroke constitutions. The experimental results demonstrate the effectiveness of the proposed method.%由于图形和文字的笔画构成复杂程度存在差异,通过计算组成图形和文字笔画的信息熵值度量该差异性,进行图文类型判别;自适应重采样解决了固定间距重采样机制可能导致的信息熵值随文档笔画尺寸变化而出现较大变化的问题,使方法能够适应不同用户的书写习惯;基于对称性检验的后处理可解决笔画构成等价导致信息熵值十分接近的图形和文字的判别.实验结果验证了所提方法的有效性.
展开▼
