发明授权
US09384409B1 Word segmentation for document image using recursive segmentation 有权
使用递归分割的文档图像的分词

Word segmentation for document image using recursive segmentation
摘要:
A word segmentation method uses a recursive technique to segment a text line image into word segments. Spacing segments of the line are obtained; an initial word segmentation is performed to classify the spacing segments based on their lengths into candidate character spacing segments and candidate word spacing segments. The initial segmentation result is evaluated to determine whether the candidate character spacing segments still have a bimodal or multi-modal distribution or a large spread in the distribution, or whether the line contains long words and too few words. If the conditions indicate that the initial segmentation is inadequate, another classification step is performed for the candidate character spacing segments to further classify them into new candidate character spacing segments and new candidate word spacing segments. The process is repeated until the word segmentation is deemed adequate based on the evaluation.
信息查询
0/0