采集能表征签名者潜在手写习惯的签名特征,利用人工免疫模型的自学习和自适应实现在较少训练样本的条件下获得具有更高区分度的手写签名模板.实验结果表明,文中方法识别具有良好的训练效果,能获得较好的验证率和鉴别率.
This paper presents an approach for online signature recognition which extracts the most commonly used signature features, and utilizes the self-learning and self-adaptation of artificial immune theory to obtain new models with higher distinguishability when the training samples are limited. Experiments show that this approach performs well in sample training and results in satisfactory verification rate and identification rate.