FGRC-Net：一种用于识别墨水光谱信息的高信息量交互式卷积神经网络

摘要：篡改文件中的关键词或数字以改变原意是一种违法行为。在某些诉讼案件，尤其是经济案件中，常常需要对文件 上的墨水类型进行鉴定。本文提出了一种无损且精确的墨水检测方法。本研究提出了一种高信息交互的全群残差卷积网 络（（ FGRC-Net），并结合光谱信息对六种不同品牌钢笔的墨水进行分类。首先，FGRC-Net通过多组卷积和多路径级联实 现了通道间的信息交互，有效提升了特征提取能力。同时，为避免特征退化，引入了残差连接。其次，基于高光谱系统 获取了六种钢笔的墨水光谱信息。最后，通过对比FGRC-Net与其他深度学习方法，FGRC-Net表现出最佳的分类性能， 准确率达到 98.38%，精确度为 98.53%，召回率为 98.40%。结果表明，将 FGRC-Net 与高光谱系统相结合是一种有效的 手写墨水分类检测方法，同时也为手写文件墨水鉴定提供了一种有效的检测手段。此外，通过多个数据集的对比，验证 了FGRC-Net在处理光谱信息方面的普适性和有效性。

关键词：墨水分类；光谱信息；卷积神经网络；信息交互

期刊：Expert Systems With Applications

DOI：https://doi.org/10.1016/j.eswa.2023.121167

设备：GaiaSorter

FGRC-Net: A high-information interactive convolutional neural network for identifying ink spectral information

Abstract：Modifying some keywords or numbers on documents to change the original intention is illegal. In some litigation cases, especially economic cases, there is often a need to examine the type of ink on documents. This paper proposes a nondestructive and accurate detection method for ink inspection. In this work, a high-information interactive, full group residual convolution network (FGRC-Net) is proposed and combined with spectral information to classify the inks of six different brands of pens. First, FGRC- Net realizes the information interaction among channels to effectively improve the feature extraction ability through multigroup convolution and multipath cascade. Meanwhile, to avoid feature degradation, the residual connection is introduced. Second, the ink spectral information of six pens based on the hyperspectral system is obtained. Finally, in comparing FGRCNet and other deep learning methods, FGRC-Net shows the best classiffcation performance with 98.38% accuracy, 98.53% precision, and 98.40% recall. The results show that combining FGRC-Net with a hyperspectral system is an effective detection method for handwritten ink classiffcation. It also provides an effective detection method for handwritten document ink inspection. At the same time, the universality and validity of FGRC-Net in processing spectral information are veriffed by the comparison of multiple datasets.

Keywords：Ink Classiffcation; Spectral Information; Convolutional Neural Networks; Information Interaction