汉字手写笔迹鉴别预处理算法的研究

Analysis of pre-processing algorithms in computer handwriting identification

研究了笔迹鉴别中的各种预处理算法,对传统的方法进行了改进。与传统的预处理方法相比,增加了笔迹图像背景去除、标点去除、汉字过分合并等功能,省去了去除行间和字间的空档的步骤,实现了汉字的任意拼接。以Matlab7.0为开发平台进行了大量的实验,实验结果表明,提出的算法具有处理效果好、可行性强、速度快等优点。最后用20个人120幅样本作对比实验,新算法可以提高笔迹鉴别率5%左右。

Many pre-processing algorithms are researched thoroughly in handwriting identification, and some advanced methods instead of traditional ones are put forward, the traditional pretreatment method is compared with, and the handwriting picture background elimi- nation, punctuation elimination, the Chinese character merges excessively and so on are improved, and then neutral gear between elimi- nation between the lines and the character step is omitted, Chinese character random splicing is realized. The thesis also made a lot of experiments by the platform of Matlab7.0, it proved that the new methods had the advantages of better effects, Stronger feasibility and quicker speed the contrast experiments with 120 samples showed that the new method can increase the identification rate by about 5 percent.