摘要
提出一种基于遗传算法的离线签名鉴定方法。将签名图像分成多个子区域,提取各区域的分形维数矢量来描述笔迹的抖动程度,在此基础上,以形状特征、伪动态特征、分形维数作为签名的特征;运用加权欧式距离法构建分类器,并运用遗传算法对不同人的签名样本进行最优阈值选择。实验结果表明该方法能够取得较高的鉴别率。
A method of off-line signature verification based on genetic algorithm is proposed.An image is divided into many small regions to descript the jitter level and then fractal features of each region are calculated.The shape features,pseudo dynamic features and fractal features have been extracted in feature extraction.In particular,we utilize weighted euclidean distance for classification and genetic algorithm is used to find out the optimal threshold for different writers.The experiment result shows that the approach is effective and has encouraged the performance of the verification.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第35期171-173,178,共4页
Computer Engineering and Applications
基金
国家自然科学基金No.79816101~~
关键词
签名鉴定
特征提取
遗传算法
最优阈值
signature verification
feature extraction
genetic algorithm
optimal threshold
