基于支持向量机的文本无关字迹鉴别系统

Writer Verification System Based on Support Vector Machines

介绍一个脱机方式的文本无关的字迹鉴别系统 .系统由把手写的文字作为图像通过扫描仪输入计算机 ,预处理 ,提取特征参数 ,训练模型 ,最后进行识别等阶段组成 .在预处理阶段 ,提出了局部分割字符图像中的行 ,行的标准化 ,通过行的投影纠正行的水平位置 ,最后消除水平方向文字之间的空白的方法 .在参数提取阶段 ,采用 Gabor变换计算 4个方向每个方向 4个频率的笔迹特征 .在训练和识别阶段 ,使用支持向量机建立模型和进行识别 .通过对 17个人任意书写的笔迹进行实验 。

The text independent writer recognition system based on support vector machines is proposed in this paper. The system includes the following procedures: scan handwritings, preprocess the character images, extract features, train and verify the handwriting. In the stage of preprocessing, we divide images into several blocks in vertical direction, then set the standardize lines, use the horizontal projection to rectify the lines in level position and remove the blank between two characters. The main method employed in the stage of feature extraction is based on the Gabor transform. It analyses every parameter from four directions and every direction has its own four frequencies. We apply the support vector machines to writer verification. In the experiment, we tested 17 writers random handwritings and achieved a better result.