BCSD: Bank Checks Segmentation Database

BCSD数据集由德国凯泽斯劳滕大学创建，包含158张银行支票的高分辨率图像，这些图像来自多种复杂背景，包含手写文本和签名。数据集提供了像素级和块级签名分割掩码，用于训练和测试从复杂背景中提取签名的网络。数据集的创建涉及从不同来源获取图像，并使用GIMP软件手动生成签名分割掩码，以及通过Python脚本生成块级掩码。该数据集主要应用于银行支票分析和文档签名分割领域，旨在解决自动签名验证中的提取难题。

The BCSD dataset was created by the University of Kaiserslautern in Germany. It contains 158 high-resolution bank check images sourced from diverse complex backgrounds, which incorporate handwritten text and signatures. The dataset provides pixel-level and block-level signature segmentation masks, which are used for training and testing networks that extract signatures from complex backgrounds. The development of this dataset involved acquiring images from various sources, manually generating signature segmentation masks using GIMP software, and producing block-level masks via Python scripts. This dataset is primarily applied in the fields of bank check analysis and document signature segmentation, aiming to address the extraction challenge in automatic signature verification.