Image Forgery Detection Based on the Convolutional Neural Network
Guorui Feng, Wu Jian · 2020
With the advent of the artificial intelligence era, a series of deep learning networks have shown great advantages in the field of image processing. The traditional method based on feature extraction has been replaced by the deep learning technology used for image forgery detection. And the latter is becoming the current research hotspot. This paper proposes an image forgery detection algorithm based on convolutional neural network. Compared with the traditional feature of extracting related features based on image content, this paper firstly uses SRM filtering and high-pass filtering to achieve image pre-processing for Columbia University image mosaic detection dataset. Then the processing of training and verification were realized by the convolutional neural network. The effects of pre-processing and the number of convolution layers on the classification results are fully compared. Experiments show that the convolutional neural network method mentioned in this paper has certain effectiveness and robustness for image forgery classification.