Mezghani Neila, Mitiche Amar, Cheriet Mohamed
Laboratoire de Recherche en Imagerie et Orthopédie (LIO), Centre de Recherche du CHUM, pavillon J.A de Sève, Hôpital Notre-Dame, Montréal, Québec, Canada.
IEEE Trans Pattern Anal Mach Intell. 2008 Jul;30(7):1121-31. doi: 10.1109/TPAMI.2007.70753.
This study investigates Bayes classification of online Arabic characters represented by histograms of tangent differences and Gibbs modeling of the class-conditional probability density functions. The parameters of these Gibbs density functions are estimated following the Zhu, Wu, and Mumford constrained maximum entropy formalism, originally introduced for image and shape synthesis. We investigate two partition function estimation methods: one uses the training sample and the other draws from a reference distribution. The efficiency of the corresponding Bayes decision methods, and of a combination of these, is shown in experiments using a database of 9504 freely written samples by 22 scriptors. Comparisons to the nearest neighbor rule method and Kohonen neural network methods are provided.
本研究调查了由切线差直方图表示的在线阿拉伯字符的贝叶斯分类以及类条件概率密度函数的吉布斯建模。这些吉布斯密度函数的参数是按照朱、吴和芒福德的约束最大熵形式主义进行估计的，该形式主义最初是为图像和形状合成而引入的。我们研究了两种配分函数估计方法：一种使用训练样本，另一种从参考分布中抽取。在使用由22名书写者自由书写的9504个样本的数据库进行的实验中，展示了相应贝叶斯决策方法及其组合的效率。还提供了与最近邻规则方法和科霍宁神经网络方法的比较。
