Abstract: In order to optimize off-line handwritten signature verification, an efficient off-line signature verification method based on interval notation and fuzzy similarity measure is presented. In the feature extraction step, a set of features based on an improved Local Binary Pattern（LBP） and a Gray Level Co-occurrence Matrix（GLCM） are computed from the signature image and its undersampled bitmap. Then obtaining the interval value symbol data for each feature in each signature class. A signature model consisting of a set of interval values（corresponding to the number of features） is created for each person’s handwritten signature class. In order to verify the test sample, a new fuzzy similarity measure is also proposed to calculate the similarity between the test sample signature and the corresponding interval-valued sign model. In order to evaluate the proposed verification method, different types of Chinese handwritten signature handwriting pictures are used for testing and comparison. The recognition rate can reach 92.75%. The experimental results show that when the number of training samples is 10 or more, the recognition rate is effectively improved. The advantage of the proposed method is that the proposed method does not need to be retrained when adding new classes to the system. The proposed method is cheaper than neural networks in terms of memory usage and computation time.

Key words: off-line handwritten signature, feature extraction, feature fusion, interval notation, intra-class variability, fuzzy similarity

摘要： 为优化离线手写签名验证，提出了一种基于区间符号表示和模糊相似性度量的高效离线签名验证方法。在特征提取步骤中，从签名图像及其欠采样位图计算一组基于改进局部二值模式（LBP）与灰度共生矩阵（GLCM）相融合的特征。然后获得每个签名类中每个要素的区间值符号数据。为每个人的手写签名类创建由一组间隔值（对应于特征的数量）组成的签名模型。为了验证测试样本，还提出了一种新的模糊相似性度量来计算测试样本签名和相应的区间值符号模型之间的相似度。为了评估所提出的验证方法，使用了不同类型的中文手写签名笔迹图片进行测试与比对，识别率可以达到92.75%。实验结果表明当训练样本的数目是10或更多时，有效提高了识别率，所提出的方法优点在于当向系统添加新类时不需要被重新训练，并在内存使用和计算时间方面与神经网络比较是廉价的。

关键词: 离线手写签名, 特征提取, 特征融合, 区间符号表示, 类内变异性, 模糊相似性