Abstract:

[Purposes] We use different methods to detect image manipulations and thereby put forward clear image detection specifications and standards in English academic journals. [Methods] We summarized the characteristics of typical cases of image manipulations in recent years and used computer-aided identification tools, such as Photoshop, Droplets, and Forensically, to detect image manipulations at different stages of article processing, aiming to improve the detection method of image manipulations. [Findings] The detection method and judging criteria for image manipulations were developed and improved, with which image manipulation was detected in 3.2% and 1.3% of the papers before and after publication, respectively. [Conclusions] Journals should emphasize the image manipulation and guide the correct processing of images, and the competent department of journals should strengthen the ethical construction of journals.

Key words: Scientific journal, Academic misconduct, Image manipulation, Misconduct, Detection

摘要：

【目的】 尝试通过针对图片不当处理检测方法的实践,提出明确的图片检测规范和具体的检测标准。【方法】 通过梳理近年来图片学术不端典型案例的特征,使用Photoshop、Droplets及Forensically三种计算机辅助识别工具,在期刊日常工作的不同阶段进行图片不当处理检测,以完善图片不当处理的检测方法。【结果】 构建并完善图片不当处理的检测方法和判断分级标准,并在论文发表前和发表后,分别检测到有3.2%和1.3%的论文存在图片不当处理。【结论】 期刊编辑部应提高对图片不当处理的重视,并加强对作者关于正确处理图片的引导,期刊主管部门则应注重对期刊学术伦理的建设。

关键词: 科技期刊, 学术不端, 图片篡改, 不当处理, 检测