K. Taya, N. Kuroki, Naoto Takeda, T. Hirose, M. Numa
{"title":"通过CNN检测JPEG图像中的篡改区域","authors":"K. Taya, N. Kuroki, Naoto Takeda, T. Hirose, M. Numa","doi":"10.1109/newcas49341.2020.9159761","DOIUrl":null,"url":null,"abstract":"Often, digital pictures are used as evidence in criminal investigations. Therefore, it is essential to check whether they have been tampered with or not. In this study, we propose a method for detecting the tampered region in a JPEG image by using a convolutional neural network (CNN). In the proposed method, DCT coefficients are input to the CNN. The output is a binary segmented image in which the tampered and non-tampered regions are represented using white and black pixels, respectively. In our experiment, 45 types of CNN models were created and compared with one another. The detection accuracy of the best model was 0.63 in terms of the F-measure, which is approximately 2.3 times that achieved using our preliminary method, which was based on a support vector machine.","PeriodicalId":135163,"journal":{"name":"2020 18th IEEE International New Circuits and Systems Conference (NEWCAS)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":"{\"title\":\"Detecting tampered regions in JPEG images via CNN\",\"authors\":\"K. Taya, N. Kuroki, Naoto Takeda, T. Hirose, M. Numa\",\"doi\":\"10.1109/newcas49341.2020.9159761\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Often, digital pictures are used as evidence in criminal investigations. Therefore, it is essential to check whether they have been tampered with or not. In this study, we propose a method for detecting the tampered region in a JPEG image by using a convolutional neural network (CNN). In the proposed method, DCT coefficients are input to the CNN. The output is a binary segmented image in which the tampered and non-tampered regions are represented using white and black pixels, respectively. In our experiment, 45 types of CNN models were created and compared with one another. The detection accuracy of the best model was 0.63 in terms of the F-measure, which is approximately 2.3 times that achieved using our preliminary method, which was based on a support vector machine.\",\"PeriodicalId\":135163,\"journal\":{\"name\":\"2020 18th IEEE International New Circuits and Systems Conference (NEWCAS)\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2020-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"3\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2020 18th IEEE International New Circuits and Systems Conference (NEWCAS)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/newcas49341.2020.9159761\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2020 18th IEEE International New Circuits and Systems Conference (NEWCAS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/newcas49341.2020.9159761","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Often, digital pictures are used as evidence in criminal investigations. Therefore, it is essential to check whether they have been tampered with or not. In this study, we propose a method for detecting the tampered region in a JPEG image by using a convolutional neural network (CNN). In the proposed method, DCT coefficients are input to the CNN. The output is a binary segmented image in which the tampered and non-tampered regions are represented using white and black pixels, respectively. In our experiment, 45 types of CNN models were created and compared with one another. The detection accuracy of the best model was 0.63 in terms of the F-measure, which is approximately 2.3 times that achieved using our preliminary method, which was based on a support vector machine.