Quad-tree–structured feature extraction for enhanced forgery detection in digital images

Copy-move forgery is a common image tampering technique where a part of an image is copied and pasted elsewhere within the same image to conceal or duplicate objects. Detecting such manipulations is critical for ensuring the authenticity of digital images and has become a major focus in image forensics. This study proposes a novel copy-move forgery detection algorithm based on Quadtree feature extraction and Stationary Wavelet Transform (SWT). The input image is divided into overlapping grayscale blocks of size 8×8. SWT is applied to the LL sub-band, and features are extracted using Quadtree decomposition. Lexicographical sorting is used to organize feature vectors, and Euclidean distance helps identify similar blocks. Forged areas are detected and localized using a binary map. Experimental results show that the method effectively detects forgeries, even after post-processing such as blurring, brightness/contrast adjustments, and color reduction, and is robust against small and multiple duplicated regions.