Image Forgery Detection using FREAK Binary Descriptor and Level Set Segmentation [ READ ]
Bayumy A.B. Youssef and Essam H. Atta
This paper describes a novel approach for copy-move image forgery detection using FREAK binary descriptor and level set segmentation. Binary descriptors are fast to compute and compact to store, since they depend only on image intensity comparison. In the proposed approach image, key features are computed using FAST detector and feature matching is implemented using hamming distance. Clustering of matched features is achieved by using region-based Level set segmentation. Experimental results carried on several image data sets and comparison with SIFT based methods show that the proposed approach is accurate and effective in terms of copy move forgery identification and computational speed.
Bayumy A.B. Youssef and Essam H. Atta
This paper describes a novel approach for copy-move image forgery detection using FREAK binary descriptor and level set segmentation. Binary descriptors are fast to compute and compact to store, since they depend only on image intensity comparison. In the proposed approach image, key features are computed using FAST detector and feature matching is implemented using hamming distance. Clustering of matched features is achieved by using region-based Level set segmentation. Experimental results carried on several image data sets and comparison with SIFT based methods show that the proposed approach is accurate and effective in terms of copy move forgery identification and computational speed.
