JJEM: Volume 3 Issue 1 (Jan - Jun 2018)
JJEM: Volume 3 Issue 1 (Jan - Jun 2018)
2019-06-30
Image Forgery Detection Using Adaptive Oversegmentation and Feature Point Matching
H R Arpita., Shwetha B., S. V. Sathyanarayana
From the early days, images are generally accepted as a proof of occurrence of the past events. Digital image is a part of the real world which is generated after many processes of image generation. The availability of low cost hardware and software tools makes easy to create and manipulate digital images with no traces. This has led to the situation where one can no longer take the integrity and authenticity of the digital images for granted. So it is important to have an image forgery detection tools to know whether the digital image is authentic or not. In this regard, adaptive oversegmentation and feature point matching is used to detect copy move forgery. First, adaptive oversegmentation algorithm segments the source image into nonoverlapping and irregular blocks. Then, block feature extraction algorithm extracts the feature points from each image block as block features using SIFT(Scale Invariant Feature Transform) and SURF (Speed Up Robust Features) methods and find the matched feature points by matching the feature points with one another by using block feature matching algorithm. After that, forgery region extraction algorithm is used, which replaces the feature points with small superpixels and merges the neighboring blocks which is having similar local color features to generate the merged regions. Lastly, morphological operation is applied to generate the forged regions. Based on precision, recall, F1measure and accuracy parameters obtained from different detection methods, the comparative analysis of copy-move forgery detection method is performed.
Copy-Move Forgery Detection, Adaptive Oversegmentation, SIFT, SURF, Block Feature Matching, Local Color Feature, Forgery Region Extraction.