The addition of new words in handwritten documents such as bank cheques, bills, and notes is considered as common crime. Such immoral activities on handwritten documents have a bad effect on the victim in terms of mental and financial loss. For facilitating an impartial judicial process, it is important to differentiate between the used pen inks. Earlier, image processing and pattern recognition-based techniques for pen ink differentiation gained attention among researchers due to their non-destructive nature. Existing techniques use various colour models to represent the colour of pen inks. Thus, it is crucial to assess the capabilities of various colour models for differentiating pen inks in handwritten documents. In this paper, we propose a histogram distance based quantitative assessment technique for suitable colour model identification for differentiating pen inks. Seven blue and seven black pen ink samples are acquired on 112 cheque leaves. The k -means binarisation is used to identify pen ink pixels. colour model has been identified as the best colour model for this task. The statistical features of the ink colours in the identified colour model representation are extracted and a multi-layer perceptron (MLP) classifier validates the capability of the identified colour model for pen ink discrimination.

中文翻译：

---

色彩模型对手写文档中的笔墨鉴别能力的定量评估

在手写文件（例如银行支票，钞票和纸币）中添加新单词被视为常见犯罪。在手写文件上进行这种不道德的活动会给受害者带来精神和经济损失方面的不利影响。为了促进公正的司法程序，区分使用过的钢笔墨水很重要。早先，基于图像处理和基于模式识别的笔墨区分技术由于其无损的特性而受到研究人员的关注。现有技术使用各种颜色模型来表示笔墨水的颜色。因此，至关重要的是评估用于区分手写文档中的笔墨的各种颜色模型的功能。在本文中，我们提出了一种基于直方图距离的定量评估技术，以用于区分笔墨水的合适颜色模型识别。在112张支票上采集了七个蓝色和七个黑色钢笔墨水样品。的ķ -均值二值化用于识别笔墨水像素。 颜色模型已被确定为该任务的最佳颜色模型。提取所识别的颜色模型表示中的墨水颜色的统计特征，并使用多层感知器（MLP）分类器验证所识别的颜色模型用于笔墨水识别的能力。