中文摘要：

基于弹性网格方法能够在一定程度上吸收手写汉字形变因而在汉字识别中取得较理想的效果，在传统弹性网格的基础上，提出一种新的伪二维弹性网格，通过对手写汉字图像的局部线密度进行模糊化与均衡，得到非四方形的曲线网格，相比由全局密度投影均衡生成的传统弹性网格有更好的吸收局部形变的能力．将多种特征提取方法与伪二维动态网格结合进行实验，与结合传统弹性网格比较，取得更为理想的识别率．由于避免了非线性归一化方法产生的形变和锯齿效应，因而识别结果也优于传统非线性归一化方法．

英文摘要：

The elastic mesh method which can absorb character variation has shown its superior performance in offline hand-written character recognition. A novel elastic mesh produced by local blurred line density equalization as pseudo two-dimension elastic mesh is proposed, which is shown as curve grids and can trace local character variation more tightly. Experiments of hand-written Chinese character databases show that compared with the traditional elastic mesh method, the pseudo two-dimension elastic mesh increases the recognition accuracy rates when it is cooperated to many feature extraction approaches. Pseudo two-dimension elastic mesh avoids the effect of distortion or zigzag caused by shape normalization. Thus, its recognition accuracy rates is higher than that of normalization methods.