Comparison of Matching Methods for Copy-Move Image Forgery Detection
Copy-Move is one of the most common image forgery types, where a region of an image is copied and pasted into another location of the same image. Such a forgery is simple to achieve but hard to be detected as the pasted region shares the same characterist
- PDF / 198,181 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 31 Downloads / 210 Views
Abstract Copy-Move is one of the most common image forgery types, where a region of an image is copied and pasted into another location of the same image. Such a forgery is simple to achieve but hard to be detected as the pasted region shares the same characteristics with the image. Although plenty of algorithms have been proposed to tackle the copy-move detection problem, those algorithms differ in two things; matching method and type of features. In this paper, we focus on analyzing and comparing four matching methods in terms of accuracy and robustness against different image processing operations. Such analysis and comparison provide indispensable information for the design of new accurate and reliable copy-move detection techniques. Keywords Copy-move ⋅ Digital image forensics ⋅ Image forgery
1 Introduction Digital images became an important source of information in our digital world. It has been said that a picture is worth a thousand words and seeing is believing. But those sayings seem not to be completely acceptable in the presence of photo editing software. Popular and simple computer software can be used by average computer users to manipulate digital images without leaving a noticeable trace. Although manipulated, or forged, images can be shared using social media for fun, they can be used in many serious situations such as journalism, criminal investigation, and surveillance systems [1]. Copy-move is one of the most popular methods for manipulating a semantics image [2]. O.M. Al-Qershi (✉) ⋅ B.E. Khoo (✉) School of Electrical and Electronic Engineering, Universiti Sains Malaysia, Penang, Malaysia e-mail: [email protected] B.E. Khoo e-mail: [email protected] © Springer Science+Business Media Singapore 2017 H. Ibrahim et al. (eds.), 9th International Conference on Robotic, Vision, Signal Processing and Power Applications, Lecture Notes in Electrical Engineering 398, DOI 10.1007/978-981-10-1721-6_23
209
210
O.M. Al-Qershi and B.E. Khoo
It can be achieved by copying a region from an image and pasting it into the same image with the intent of hiding undesired objects or replicating objects. In copymove forgery, the tampered region still shares most of its inherent characteristics, such as the color palette or pattern noise, with the remainder of the image. Most of the copy-move detection algorithms adhere to a common pipeline [3]. First, the image is optionally pre-processed (downscaling and/or conversion to greyscale). It is then subdivided into overlapping blocks of pixels. From each of these blocks, a feature vector is extracted. Highly similar feature vectors are matched as pairs. The similarity of two features can be determined by different similarity criteria, e.g. Euclidean distance. In the verification step, outliers are removed and holes are filled which may be achieved using a basic filtering such as morphological operations. The overall performance of copy-move detection methods depends mainly on two stages of that pipeline; the type of the features that are extracted from image block
Data Loading...
