Vol.:(0123456789) SN Computer Science (2022) 3:46 https://doi.org/10.1007/s42979-021-00903-2 SN Computer Science ORIGINAL RESEARCH Copy-Move Image Forgery Detection Using Phase Adaptive Spatio-structured SIFT Algorithm Raimoni Hansda 1  · Rajashree Nayak 2  · Bunil Kumar Balabantaray 1  · Sonali Samal 1 Received: 11 May 2021 / Accepted: 24 September 2021 © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd 2021 Abstract This paper suggests a hybrid Copy-Move image forgery detection method using phase adaptive Spatio-structured SIFT algorithm which enables to localize forgery regions (FRs) in the presence of combination of intermediate and post-pro- cessing attacks. Moreover, the proposed method well localizes varied types and mixed-sized FRs at greater accuracy. Key contributions of the proposed method are: (i) detection of matched key-points via Spatio-structured SIFT (S-SIFT) algo- rithm; (ii) formation of larger blocks around the matched key-points followed by the division of larger blocks into several non-overlapping blocks; (iii) representation of non-overlapping blocks via feature descriptors based on the histogram of oriented phase congruency (HoPC); (iv) block matching via 2NN matching strategy to localize the fnal forged areas. S-SIFT algorithm integrates spatial and structural information additionally in the SIFT feature descriptor for detecting sufcient numbers of key-points at relatively smoothed and little structured FRs. Formation of larger blocks around the matched key- points enables to enhance the probability of detecting any sized FRs, whereas non-overlapping division of blocks generates a lesser number of blocks to be matched. Hence, the computational time of the matching process will be reduced. Feature descriptors based on HoPC are robust to noise, invariant to various geometric transforms, insensitive to change in image contrast and non-uniform illumination. Performance assessment of the proposed work in the presence of diferent attacks is validated both at image-level and pixel-level on Benchmark datasets, such as MICC-F220, GRIP, and CoMoFoD, and found to outperform the state-of-the-art methods. Keywords S-SIFT · HoPC · Intermediate attacks · Post-processing attacks Introduction In the modern era of digital India, digital images play vital role in almost all applications whether it is medical imag- ing, law enforcement, military, forensic applications, image and computer vision applications, or various commercial applications. Hence, it becomes indispensable to check the authenticity and originality of the input image before its further use. However, certain user friendly image manipu- lation tools, such as Adobe Illustrator, Adobe Photoshop, PaintShop, ACDSee, Polarr, and GIMP, enable to modify or tamper the integrity of the original content without pro- viding any mark or clue in the tampered images. Tamper- ing or image forgery (IF) is performed to hide important information or to mislead the situation by pasting some fake information in the same scene. By visual perception, one cannot assure the reliability or authenticity of digital images. Hence, detection and localization of forgery region become essential. It becomes more challenging to localize This article is part of the topical collection “Progresses in Image Processing” guest edited by P. Nagabhushan, Peter Peer, Partha Pratim Roy and Satish Kumar Singh. * Rajashree Nayak rajashreenayak17@gmail.com Raimoni Hansda raimoni.hansda1@gmail.com Bunil Kumar Balabantaray bunil@nitm.ac.in Sonali Samal ps20.cs.002@nitm.ac.in 1 Computer Science and Engineering, National Institute of Technology Meghalaya, Shillong, Meghalaya 793003, India 2 School of Engineering, MIT-ADT University, Pune, Maharashtra 412201, India