基于图像识别与文本特征的作业查重系统研究  

Research on Homework Plagiarism Detection System Based on Image Recognition and Text Features

作　　者：黎斌 卫静婷 吴家隐 LI Bin;WEI Jingting;WU Jiayin(Guangdong Open University(Guangdong Polytechnic Institute),Guangzhou Guangdong 510630)

机构地区：[1]广东开放大学(广东理工职业学院),广东广州510630

出　　处：《软件》2024年第12期20-23,共4页Software

基　　金：广东开放大学(广东理工职业学院)2019年度科研课题重点项目“MIS结构InGaN基可见光探测器制作及其特性研究”(ZD1905);广东开放大学(广东理工职业学院)2021年度科研课题一般项目“5G关键技术在车联网中的应用”(YB2114);广东开放大学(广东理工职业学院)2024年教学建设与改革项目“面向‘现场工程师’岗位的课程改革探索与实践—以《低功耗广域网技术》课程为例”(2024JYJG002)。

摘　　要：在计算机类课程中,学生提交的作业通常包含大量的操作截图,传统的基于文本相似度的查重系统在处理这类作业时表现出明显的局限性。基于此,本文提出了一种结合图像识别与文本相似度计算的新型作业查重系统。该系统首先通过图像识别技术将作业中的截图转换为可读文本,然后利用文本特征算法对转换后的文本进行处理,提取关键特征,计算文档间的相似度,以此识别潜在的重复作业。实验结果显示,与传统的文件差异算法和文字识别算法相比,文本特征算法不仅在运行速度上具有优势,而且在检测准确性和效率之间达到了较好的平衡。该方法能够显著提高查重的准确性,营造良好的学习环境。In computer courses,the homework submitted by students usually contains a large number of operation screenshots,traditional text similarity based plagiarism detection systems exhibit significant limitations in handling such tasks.Based on this,this article proposes a novel homework plagiarism detection system that combines image recognition and text similarity calculation.The system first converts the screenshots in the homework into readable text through image recognition technology,then use text feature algorithms to process the converted text,extract key features,and calculate the similarity between documents,to identify potential repetitive tasks.The experimental results show that,compared with traditional file differentiation algorithms and text recognition algorithms,text feature algorithms not only have advantages in running speed,and a good balance has been achieved between detection accuracy and efficiency.This method can significantly improve the accuracy of plagiarism detection,create a good learning environment.

关 键 词：查重系统 文本识别 特征提取 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]