IMAGE TAMPERING DETECTION: A REVIEW OF MULTI-TECHNIQUE APPROACH FROM TRADITIONAL TO DEEP LEARNING
Vitthal B. Kamble1, Dr. Nilesh J. Uke2
1Research Scholar, Department of Computer Engineering, Vishwakarma Institute of Technology, Kondhwa Budruk, Pune, Maharashtra – 411048, India
(Affiliated to Savitribai Phule Pune University, Pune India)
2Professor, Department of Computer Engineering, Indira College of Engineering and Management, Parandvadi, Pune, Maharashtra – 410506, India
(Affiliated to Savitribai Phule Pune University, Pune, India)
Abstract: Digital images have applications in many different fields, including journalism, forensic analysis, police inquiry, smart systems, and radiologic diagnostics. Social media and online photographs are trustworthy sources of information. Photos can be changed or utilized for personal gain with the help of easily accessible software or editing programs like Photoshop, Corel Paint Shop, PhotoScape, PhotoPlus, GIMP, Pixelmator, etc. Photo-realistic photographs are now difficult to differentiate from actual photographs due to advancements in deep learning techniques like GAN (deep reinforcement learning) and other active, passive, and other techniques. Today, the goal of digital picture tampering detection is to guarantee the consistency and dependability of digital photographs. Maintaining the integrity of digital content is very important in different domains such as journalism, media, social media, forensics, and national security. This survey analyzes both active and passive image forgery techniques to identify the tampering signs and manipulation done in the image content. The forgeries in manipulated images are identified using camera source identification, JPEG compression tampering, illumination inconsistencies, and mathematical manipulation. These diverse approaches provide a clear overview of the image forensic field. To address the primary research challenges, normalized input sets, gauges, and analysis standards are used. In this paper, the numerous tampering techniques for recognition are summarized. In addition to comparing image criminological (forensic) approaches, this work also briefly discusses image datasets. Over the last little while the expanded deep neural network techniques and also its shortcomings have been examined. Here, in this paper, you will find fully examined photo tampering identification techniques using both standard and cutting-edge neural network techniques. The most recent methods were unable to reliably identify multiple attacks including whirligig, smearing, proportioning, JPEG compression, and illumination strength. Computational intensity, feature dimensionality, detection accuracy, and resistance against further creation are additional issues that must be resolved.
Keywords: Digital image forensics, GAN, Copy-move tampering identification, Deep Learning techniques