Source code related to our scientific paper "Behavior Knowledge Space-Based Fusion for Copy-Move Forgery Detection", published by IEEE Transactions on Image Processing.

Author: Anselmo Ferreira

INSTRUCTIONS

The source code is divided in two parts:

1. Repeat Experiments: you can re-run our experiments and check the effectiveness of our proposed approach in a 5X2 cross validation scenario on the CPHPALL (uncompressed images) dataset. Results can vary a little bit depending on the precison your CPU uses to calculate the probabilities.

2. Do your own BKS: Here you can check how the multiscale BKS is built and also can change the code to include your images and/or compare your approach against ours.

In both cases you must download a folder called aux. Available to download at https://www.dropbox.com/sh/2sa1xbukql23bou/AACeJVHLEXucQKb-4-HwXvtGa?dl=0.

Please extract this folder in the same folder as the source code folders are in your PC.

If you use this code in your research don't forget to cite our work:

@ARTICLE{7517389, author={A. Ferreira and S. C. Felipussi and C. Alfaro and P. Fonseca and J. E. Vargas-Muñoz and J. A. dos Santos and A. Rocha}, journal={IEEE Transactions on Image Processing}, title={Behavior Knowledge Space-Based Fusion for Copy-Move Forgery Detection}, year={2016}, volume={25}, number={10}, pages={4729-4742}, keywords={Detectors;Feature extraction;Forgery;Lighting;Robustness;Training data;Transforms;Copy-move forgery detection;behaviour knowledge space;fusion;multi-direction data analysis;multi-scale data analysis}, doi={10.1109/TIP.2016.2593583}, ISSN={1057-7149}, month={Oct},}

You can also download aux folder directly from your browser at ftp://www.recod.ic.unicamp.br/2016-tip-copy-move-dataset

Any question, problems or suggestions can be reported to me. My e-mail is [email protected]

You can find more information about my research in my personal website: www.ic.unicamp.br/~anselmoferreira

Best Regards and have fun.