Recent Advances in Image Forgery Detection: Handcrafted Feature Extraction and Deep Learning-Based Approaches

Images and videos are essential information sources in a variety of fields, such as social media, forensic investigations, and journalism, in the current digital era. However, digital media is now more vulnerable to manipulation due to the widespread use of sophisticated editing tools, more powerful computers, and high-resolution photography equipment. This presents serious questions about the veracity and validity of visual content because forgeries can help spread false information, commit fraud, and engage in other dishonest practices. With an emphasis on both handcrafted feature-based approaches, like keypoint-based techniques using SIFT and SURF, and deep learning-based approaches, like convolutional neural networks (CNNs) and transformer models, this paper provides an extensive review of recent developments in image forgery detection. Furthermore, we examine publicly accessible datasets that are frequently used to assess image forgery detection methods, offering insights into their traits and applicability to various forgery detection tasks. This review attempts to support the development of more efficient picture forensic tools by providing an overview of the most recent research trends and datasets, protecting the integrity of visual data in a digital environment that is becoming more and more corrupted.