Copy-move forgery detection using the segment gradient orientation histogram

Abstract

The ready availability of image-editing software makes ensuring the authenticity of images an important issue. The most common type of image tampering is cloning, or Copy-Move Forgery (CMF), in which part(s) of the image are copied and pasted back into the same image. One possible transformation is where an object is copied, rotated and pasted; this type of forgery is called Copy-Rotate-Move Forgery (CRMF). Applying post-processing can be used to produce more realistic doctored images and thus can increase the difficulty of forgery detection. This paper presents a novel segmentation-based Copy-Move forgery detection method. A new method has been developed to segment the Copy-Move objects in a consistent way that is more efficient than Simple Linear Iterative Clustering (SLIC) segmentation for CMF/CRMF. We propose a new method to describe irregular shaped blocks (segments). The Segment Gradient Orientation Histogram (SGOH), is used to describe the gradient distribution of each segment. The quality of initial matches is improved by applying hysteresis to grow the primary detection regions. We show that the proposed method can effectively detect forgery involving translation and rotation. Moreover, the proposed method can detect forgery in images with blurring, brightness change, colour reduction, JPEG compression, variations in contrast and added noise.