Segmentation for Handwritten Numerals Based on Background Analysis
基于背景分析的手写数字切分方法

In this paper a scheme is presented for segmentation of handwritten numerals based on analysis of the specified background space. In the scheme, some interesting background regions called water reservoirs are first obtained by considering accumulation of water poured from the top or from the bottom of the numerals to illustrate where numerals touch each other. Then, the touching numerals are classified into six types according to the features extracted from the reservoirs. For numerals segmentation, a set of Drop Fall algorithms is developed and employed to generate the cutting paths with different strategy to separate each touching type. Experimental results show that the proposed scheme is effective for handwritten numerals segmentation.