Image Source Identification Based On Deep Convolution Neural Network

Since the 21 st century,with the rapid development of information technology,digital image/video images have rapidly integrated into all aspects of our lives,and images have been widely used as one of the efficient and intuitive description carriers.However,with the rapid development of computer image technology and the widespread emergence of various image APP software,advanced technology has brought a lot of security problems while bringing convenience to people's lives.the images synthesized by these softwares are very similar to natural images,it is difficult to distinguish between true and false by the naked eye,which makes the authenticity of the image and the reliability of the source suffer from people's doubts.Nowadays,forged images are widely used in news media,e-commerce and other fields.Forged images distorted the facts,conveyed the wrong information to people,misled people,and caused people's lives to be troubled.How to ensure and identify the authenticity of the images and the reliability of the source is an urgent problem to be solved.The image source identification in this paper contains two aspects: the identification of image sources and the identification of natural images and computer generated images.Firstly,the research background,significance and research status of image source identification are expounded.Then the related basic theory knowledge is introduced.Finally,based on the problems existing in the image source identification,the ResNet and DenseNet are proposed.Image source identification method.Compared with the previous method,the method of this paper not only improves the recognition accuracy of the image source,but also realizes the endto-end identification of the image.In addition,in order to quickly collect real and reliable computer-generated images,and in order to enhance the generalization ability of the model,the article uses the generated confrontation network to generate computer-generated images as part of the training data set.Finally,considering that the image source identification method can be applied to life quickly and conveniently,this paper proposes MobileNet image source identification method that can be used for mobile devices.