Improving Copy-Move Forgery Detection: An Investigation into Techniques Based on Blocks and Key Points

Digital image forgery, especially in the form of “Copy-Move Forgery” a prevalent form of image alteration that threatens the authenticity of digital visual content. To combat this challenge, this research conducts a thorough comparative analysis of three prominent techniques for detecting Copy-Move Forgery: “Principal Component Analysis (PCA)”, “Discrete Cosine Transform (DCT)”, and “Scale-Invariant Feature Transform” combined with “Dyadic Wavelet Transform” (SIFT-DyWT). The PCA-based method employs eigenvectors of image patches to identify copied regions, while DCT leverages frequency domain information to detect duplicated areas. SIFT-DyWT combines the powerful SIFT algorithm with Dyadic Wavelet Transform to extract and match invariant features for forgery detection. Each technique is implemented and evaluated on a diverse dataset of manipulated images, with performance metrics including precision, recall, and F1-score being assessed. Efficiency is a crucial factor, particularly for real-time applications. As a result, these strategies' computational complexity is also examined. This aspect is essential for aiding researchers and practitioners in selecting the most suitable forgery detection method based on their specific application requirements. In conclusion, this research contributes significantly to the field of image forensics by presenting a comprehensive comparison of PCA, DCT, and SIFT-DyWT for Copy-Move Forgery detection. These discoveries offer valuable understandings regarding the weaknesses and strengths of each method, facilitating the development of more robust and efficient forgery detection tools. Experimented result shows the DyWT and SIFT combination exhibit superior performance, achieving an accuracy of 89.56%. This outperforms both DCT, with an accuracy of 86.55%, and PCA, with an accuracy of 83.96%. Ultimately, this research enhances the security and reliability of digital visual content in an era where image manipulation and forgery are prevalent concerns.