Taylor‐RNet: An approach for image forgery detection using Taylor‐adaptive rag‐bull rider‐based deep convolutional neural network

Abstract

Due to the use of powerful computers and advanced software for photo editing, image manipulation in digital images simply degrades the trust in digital images. Image forensic analysis focuses on image authenticity and image content. To process forensic research, different methods are introduced, which effectively differentiate fake images from the original image. A technique named image splicing is commonly used for image tampering, and the tampered image may be used in photography contents, news reports, and so forth, which brings negative influences among the society. Thus, for detecting spliced images, this paper proposed an automatic forgery detection approach named Taylor‐adaptive rag‐bull rider (RR) optimization algorithm‐based deep convolutional neural network (Taylor‐RNet). At first, the face of a human is detected from the spliced image using the Viola Jones algorithm, and later, to estimate light coefficients, the three‐dimensional (3D) shape of the face is determined by using a landmark‐based 3D morphable model (L3DMM). Then, distance measures, like, Bhattacharya, Euclidean, Seuclidean, Chebyshev, correlation coefficients, and Hamming, are determined from the light coefficients that form the feature vector to the proposed Taylor‐RNet, which identifies the spliced image. Taylor‐adaptive RR is the integration of the Taylor series with the adaptive RR optimization algorithm. Finally, the experimental analysis is performed using four data sets, such as DSO‐1, DSI‐1, real data set, and hybrid data set. The analysis result of the proposed method obtained a maximum accuracy of 96.921%, true positive rate of 99.981%, and true negative rate of 99.783%.

PR‐NET: Progressively‐refined neural network for image manipulation localization

“…With the superior performance of deep learning in a wide range of computer vision tasks, 8–14 some recent work based on convolutional neural networks (CNNs) is continuing to explore the potential applications in image manipulation detection 15,16 . The existing methods can be mainly divided into the following two categories: (1) Binary classification detection.…”

MC‐Net: Learning mutually‐complementary features for image manipulation localization

“…Pun et al 30 took the contextual super‐pixel blocks to assist the classification of the blocks. Furthermore, a line of research focuses on image manipulation localization , which aims to localize the tampered regions 31–33 …”

“…Furthermore, a line of research focuses on image manipulation localization, which aims to localize the tampered regions. [31][32][33] The core hypothesis of the image manipulation localization methods is that any of the manipulation operations would leave some abnormal traces. [34][35][36][37] Some localization methods capture the manipulation anomalies by using manually constructed features, such as resampling, 37 noise, 38 edge, 39,40 and so forth.…”

Document images forgery localization using a two‐stream network

“…10 Some works have been proposed for forgery detection. 2,[11][12][13][14][15][16][17][18][19] Chen et al 16 detect median filtering by Convolutional Neural Networks (CNNs). Feng et al 17 utilized the normalized energy density in the frequency domain to detect resampled image.…”

“…Considering the excellent performance of deep learning technologies, 30,31 the deep learning is utilized to capture forgery traces. Relying on the deep analysis of forgery methods, different traces have been utilized for forgery localization, such as noise level, 3,5,32,33 unnatural boundaries, 34 contrast/brightness inconsistencies, [35][36][37] pixels periodic correlation caused by interpolation, 1,3,19 and so on. Both copy-move and splicing contain pasting image regions to the target image, which will cause unnatural boundaries.…”