IEEE Trans Cybern. 2015 Nov;45(11):2498-511. doi: 10.1109/TCYB.2014.2375959. Epub 2014 Dec 10.
In this paper, a novel online signature verification technique based on discrete cosine transform (DCT) and sparse representation is proposed. We find a new property of DCT, which can be used to obtain a compact representation of an online signature using a fixed number of coefficients, leading to simple matching procedures and providing an effective alternative to deal with time series of different lengths. The property is also used to extract energy features. Furthermore, a new attempt to apply sparse representation to online signature verification is made, and a novel task-specific method for building overcomplete dictionaries is proposed, then sparsity features are extracted. Finally, energy features and sparsity features are concatenated to form a feature vector. Experiments are conducted on the Sabancı University's Signature Database (SUSIG)-Visual and SVC2004 databases, and the results show that our proposed method authenticates persons very reliably with a verification performance which is better than those of state-of-the-art methods on the same databases.
本文提出了一种基于离散余弦变换（DCT）和稀疏表示的新颖在线签名验证技术。我们发现 DCT 的一个新特性，该特性可用于使用固定数量的系数获得在线签名的紧凑表示，从而简化匹配过程，并提供处理不同长度时间序列的有效替代方法。该特性还用于提取能量特征。此外，我们尝试将稀疏表示应用于在线签名验证，并提出了一种新的特定于任务的构建过完备字典的方法，然后提取稀疏特征。最后，将能量特征和稀疏特征连接起来形成特征向量。我们在 Sabancı University 的 Signature Database (SUSIG)-Visual 和 SVC2004 数据库上进行了实验，结果表明，与同一数据库上的最新方法相比，我们提出的方法可以非常可靠地认证人员，并且具有更好的验证性能。
