Different Tampering Detection Algorithms of Digital Images
Digital images being the natural carriers of information are the most widely accepted and convenient way for expressing and transmitting information. As per the statistics, in 2013 on an average 350 million images and in 2014, 1.8 billion digital images were uploaded to Facebook every single day indicating that more than 20,000 images make their way on internet every single minute 1. Due to of this wide circulation of the images on the net and with the help of readily available image editing software’s, these digital images can easily be altered or manipulated in order to misguide the common masses. Thus, the tampering of a digital image means the intentional manipulation of the image for the purpose of modifying the actual meaning of the visual message included in it or any image manipulation becomes tampering, based upon the context in which it is used 2. This gradual takeover of original images by the tampered images may give rise to some serious problems such as image integrity and authenticity, completeness of image, image content security etc. These tampered images have great impact on our society and pose serious threat to the security and integrity of image content. The authenticity and integrity of the digital images can be achieved either by preventing it from being tampered or by detecting the tampered regions in it. There are many tampering detection techniques like Principal component Analysis (PCA), Discrete Cosine Transform (DCT), Discrete Wavelet Transforms (DWT), Singular Value Decomposition (SVD), etc. Our research is focused on detection of tampering in digital images. This research work presents a parametric non-overlapping block-based tampering detection technique for detecting different types of tampering like copy-paste, transformation based, feature based and noise based. The proposed model is computationally less complex and is less time consuming. On the basis of literature survey, it is observed that very less exhaustive research has been carried out in this area of tampering detection.
This research work is carried out in two phases. In first phase on the basis of literature survey, a comprehensive classification of tampering detection techniques is presented. In second phase, a parametric non-overlapping block-based tampering detection technique has been proposed. This proposed model is evaluated for detecting different types of tampering like copy-paste, transformation (rotation, resizing or rescaling, mirror and water reflections), feature and noise based tampering. This model is tested and applied for more than eight different image domains. In addition to this effect of various features have been analyzed using this parametric non-overlapping block-based model. Finally, all the results obtained have been validated with detailed and comprehensive experimental evaluation.
Objectives of Study:
I. To study different tampering detection techniques for digital images.
II. To develop a comprehensive classification of different tampering detection techniques on the basis of various image features.
III. To develop a parametric non-overlapping block-based tampering detection model for detecting different types of tampering with in the same image.
IV. Implementation and analysis of the proposed model for evaluating the impact of copy-paste, rotation, resizing, mirror reflection, water reflection, feature based, and noise based tampering on the proposed model.
V. Application of the proposed parametric non-overlapping block-based tampering detection model in different image domains.
VI. Inferences, recommendations and conclusions on the basis of analysis of the results obtained.
The entire experimentation including the data analysis, algorithm development, creation of models, applications, interfaces have been implemented in MATLAB environment. The hardware, software, image and block specifications for test cases are discussed below:
Hardware Specification (OS: Windows 7 ultimate; Processor: Intel Pentium CPU 2.80 GHz RAM 1.50 GB; System: 32 bit Operating system)
Software Specification (MATLAB – ver. – 18.104.22.1684 (R2008a) and 22.214.171.1244 (R20013a) with Image Processing and Mathswork Toolbox; IrfanView Image Analysis S/W, Tableau Public 9.3(32bit)).
The thesis has been organized into six chapters. Chapter 1 presents the introduction to the research topic, discusses about digital image tampering, its classification and some tampering detection techniques. Research significance, issues and objectives are also presented in this chapter. Chapter 2 presents a literature survey of the different techniques and methods developed in the area of digital image tampering detection. Chapter 3 deals with the classification of digital image tampering and tampering detection techniques based on different features identified during literature survey. Chapter 4 presents the methodology of proposed experimental model for image tampering detection based on parametric non-overlapping block based technique. The experimental results on the performance of the proposed algorithm are presented in Chapter 5. Finally, Chapter 6 concludes the thesis. The list of tables, figures and graphs are presented in Appendix A and B respectively. Some selected research papers referred for this research work are also presented. A list of publications related to research work in different national / international journals and conferences is also given. The description of each chapter is provided below:
? Chapter 1- Introduction
This chapter introduces research topic, digital image tampering and its classification. A brief description about content based, creation based and context based tampering is also given and is shown in Figure 1. Digital image tampering detection and its overview is also presented. Some detection techniques used to detect tampering are also discussed in general. Justification and motivation for this research work, its significance and various research issues are also presented in this chapter. Finally, this chapter presents the objectives, contributions and scope of this proposed research work.
? Chapter 2- Review of Literature
Different detection techniques are proposed by different researchers for the detection of different types of digital image tampering. This chapter presents a literature survey of different techniques and methods presented by different researchers in the area of digital image tampering detection. Different tampering detection techniques related to the research area are based on pixel, format, physical environment, copy-paste, block-based, key-points, moments, frequency, intensity, overlapping, non-overlapping, parametric based and noise based techniques. During literature survey, some research issues are identified. It is observed
that there is a need to classify the various methods, techniques available or presented by different researchers. So, there exists a need to have a comprehensive classification available at one place for the benefit of other researchers working in the similar domain for the better understanding of the existing tampering detection techniques and their domains of working. This classification may be helpful to understand, further develop and optimize the different tampering detection models. Moreover, it is also observed that there is a need to further analyze and evaluate tampering detection models for non-overlapping block-based copy-paste, noise based and transformation based tampering detection using statistical features of an image and block. The value of the parameters is dependent on the size of the image blocks, so selection of an appropriate block size is very important. There is a need to have a framework for the identification, extent and localization of different types of tampering and to have optimal and appropriate block sizes for further experimentation and tampering detection within the same image. Different features have been identified during the literature survey. These features include number and size of blocks, parameters, threshold, image formats, type of tampering, orientation (rotation), image sizes (resizing / scaling), reflections / flips (mirror and water), noises and their intensities, image databases (more than eight) and number of images (more than 1200) have been identified for implementation and analysis.
? Chapter 3- Classification
Chapter 3 presents a comprehensive classification of digital image tampering detection techniques. In this chapter, seven different types of classifications are presented in detail. These include classifications based on different identified features, digital image tampering, digital forensics (image and video forensics), digital image tampering detection techniques based on pixel, format, camera, geometric, physical environment, copy-paste, block-based, key-point, hybrid, moment, frequency, intensity, dimensionality reduction, overlapping, non-overlapping, parametric and noise based techniques related to the research area. In addition different methods have been categorized on the basis of complexity, JPEG compression, image size, noise, robustness towards transformations i.e. rotation and scaling is also presented.
Fig.1 Different types of tampering induced in a digital image.
? Chapter 4- Methodology
The Chapter 4 presents the methodology of present research work. The methodology of the proposed research work is categorized into three sections i.e. Framework, Image databases-specifications and Experiments conducted. It also describes the execution environment used in all the experiments conducted in this thesis. All the experiments have been carried out in Matlab environment. Different imaging softwares have been utilized for performing the research analysis and to investigate the tampering status, its extent and location for tampered images using parametric non-overlapping block-based tampering detection model. The detailed description about the image domains, images and their blocks, software and hardware environment is also discussed in this chapter.
Initially, an overall research methodology framework for copy-paste, transformation based (rotation, resizing, mirror reflection / flip horizontal and water reflection / flip vertical), feature based and noise based tampering detection using non-overlapping block based technique for digital images is presented in Figure 2.
b) Image databases –specifications:
The entire experimentation is conducted using different image databases. More than 1200 different digital images having different image formats (bmp, png and jpg), dimensions (10 x 10 to 1000 x 1000), block sizes (5 x 5, 8 x 8, 16 x 16, 25 x 25, 32 x 32, 50 x 50, 64 x 64, 100 x 100) and tampering content are tested for copy-paste, transformation based, feature based and noise based digital image tampering detection using statistical parameters. The test images have been taken from more than eight different image domains that include- geometrical, facial, mixed, agricultural, medical, topographical, eyes or iris, astronomies, defence and images taken from digital camera or webcam.
Fig.2 Framework of the proposed research study.
c) Experiments conducted:
The entire research study is divided into seven different experiments i.e. copy-paste, rotation, resizing, mirror reflection or flip horizontal, water reflection or flip vertical, feature based and noise based. During the detection process, the image is detected not only at image level but also at block level. In copy-paste tampering, the image is checked for identical or duplicated regions using parametric non-overlapping block based tampering detection technique. The tampering status, extent and location, loc is established using the proposed model. In case of
Image Database 1
Image Database 3
Image Database 2
Original Test Image, IOT
Generate Blocked Image, IOB / Blocks, BiO
Extracting ; Computing Statistical Parameters for IOT, IOB / BiO
Copy-Paste Tampering in IOT
Noise Tampering / Addition in BiO
Copy-Paste Detection Model
Transformation Based Detection Models
Rotation Based Detection Model
Resizing Based Detection Model
Flip Horizontal / Mirror Reflection Based Detection Model
Flip Vertical / Water Reflection Based Detection Model
Noise Tampering Based Detection Model
Comparison of Statistical Parameters for IOT, IFT, BiO ; BNT
Tampering Detection Status (DT/ND), Extent ; location, loc
Copy-Paste Tampered Image, IFT
Noise Tampered Blocks, BNT
rotation and resizing, the copy-paste tampered image IFT rotated at different degrees i.e. 0o, 90o, 180o and 270o or resized or rescaled to different dimensions i.e. 10, 50 and 100 is detected using the proposed parametric model. In case of reflections, the copy-paste tampered image is either mirror or water reflected. It is further tested for tampering detection. In feature based tampering detection, different image as well as block features are taken into consideration for the detection of tampering. In noise based tampering detection, the extent and location of the noise tampering within an image and its intensity at block level using the proposed non-overlapping block-based parametric tampering detection model is identified. Three different noises (i.e. salt and pepper, speckle and gaussian) of different intensities are taken into account for tampering detection at block as well as at image level. In noise based tampering detection, four different cases have been discussed related to different noises and their varied intensities, different images, their dimensions, formats, number and size of the blocks, different parameters, etc.
? Chapter 5- Experimental Results, Analysis and Discussion
The experimental results on the performance of the proposed detection algorithm are presented in Chapter 5. First the parameter settings and image database selected are described. This chapter discusses the details of different experiments are conducted to test different aspects of the proposed algorithm. In the first experiment, the results are obtained after the implementation of the proposed parametric non-overlapping block-based model for copy-paste tampering detection. In the second and third experiment, the proposed model is tested for observing the impact of transformations i.e. rotation and resizing respectively using larger domain of images having different types, formats and dimensions and for tampering within an image. In the second experiment, the image or an image region is rotated at different degrees i.e. 0o, 90o, 180o and 270o for the selected block size S=32 and considered for further experimentation. In third experiment, the image or an image region is resized to selected different dimensions. Later on, a comparative analysis of the generated results for the selected statistical parameters will be done for tampering detection. In the fourth and fifth experiment, the model is tested and analyzed for observing the impact of reflections / flips (i.e. mirror / flip-horizontal or water / flip-vertical) made to the copy-paste tampered images. In the sixth experiment, the model is tested for three different formats bmp, png and jpg and for different selected block sizes. In order to verify that the proposed model identifies the tampered area for all the given images having different types, formats and dimensions, the statistical parameters of the input images of three different formats are computed, analyzed and compared with those of their tampered counterparts using specific threshold values. In seventh experiment, the model is implemented to ascertain the impact of different noises and their varied intensities at block and image level for four different experiments conducted using different images having different formats, dimensions, block sizes and using different parameters.
The experimental evaluation of the proposed framework shows that the parametric block-based non-overlapping tampering detection model generates satisfactory to good results for identification of the tampering in given digital images. This model works well with a large domain of different image features. Variation of different thresholds and parametric values results in generation of different desired outputs. In addition to identification of the presence of tampering, the proposed framework also identifies the location of the tampering in most of the cases. In addition, the proposed frame work and detection model is not resource intensive. This framework may further be integrated as the initial verification phase of any larger tampering detection system to enhance the tampering detection process by identifying the most likely cases of possible image tampering.
? Chapter 6- Conclusion and Future work
This chapter concludes the entire research work presented in the thesis. It also summarizes the results and analysis of this research work. The limitations of the work and the possible directions for future research are also presented.
Appendix A. Provides the list of different tables presented in the thesis.
Appendix B. Provides the list of figures and graphs generated.
1. Online source http://tech.firstpost.com/news-analysis/now-upload-share-1-8-billion-photos-everyday-meeker-report-224688.html.
2. B.L. Shivakumar, and Dr. S.S. Baboo, “Detecting Copy-Move Forgery in Digital Images: A Survey and Analysis of Current Methods”, IEEE Global J. Computer Science and Technology, vol. 10, no.7, pp. 61-65, IEEE, 2010.
3. M.K. Johnson and H. Farid. Exposing digital forgeries through chromatic aberration. ACM Multimedia and Security Workshop, 0:48–55, 2006.
4. L. Kang, and X. Cheng, “Copy-move Forgery Detection in Digital Image”, in Proc. of 3rd Int. Congress on Image and Signal Processing,pp. 1-5, IEEE, 2010.
5. A. C. Popescu and H. Farid, “Exposing digital forgeries by detecting traces of re-sampling,” IEEE Trans. Signal Processing, vol. 53, no. 2, pp. 758–767, 2005.
6. G. Li, Q. Wu, D. Tu, and S. Sun, “A Sorted Neighborhood Approach for Detecting Duplicated Regions in Image Forgeries Based on DWT and SVD,” in IEEE International Conference on Multimedia and Expo, Jul. 2007, pp. 1750–1753.
7. E. Ardizzone, A. Bruno, and G. Mazzola, “Copy-move forgery detection via texture description,” in Proc. of the 2nd ACM workshop on Multimedia in Forensics, Security and Intelligence (MiFor ’10), pp. 59–64, ACM, October 2010.
8. S. Kumar, J. Desai and S. Mukherjee, “A fast DCT based method for copy move forgery detection, “IEEE Second International Conference on Image Information Processing (ICIIP), pp. 649 – 654, Dec. 2013.
9. L. Kang and X. Cheng, “Copy-move forgery detection in digital image”. IEEE 3rd International Congress on Image and Signal Processing, vol.5, pp. 2419-2421, 2010.. 10. X. Pan, X. Zhang and S. Lyu, “Exposing Image Forgery with Blind Noise Estimation”, ACM, Proc. of the thirteenth ACM multimedia workshop on Multimedia and security, pp. 15-20, 2011.
11. X. Wang, X. Zhang, Z. Li and S. Wang, “A DWT-DCT Based Passive Forensics Method for Copy-Move Attacks”, IEEE Third International Conference on Multimedia Information Networking and Security, 2011.
12. Dattatherya, S.V. Chalam and M.K. Singh, “A generalized image authentication based on statistical moments of color histogram”, ACEEE International Journal on Recent Trends in Engineering and Technology, vol. 8, no. 1, Jan 2013.
13. Y. Gan and J. Zhong, “Image Copy-Move Forgery Blind Detection Algorithm Based on the Normalized Histogram Multi-Feature Vectors”, Journal of software Engineering, vol. 9, no. 2, pp. 254-264, 2015.
14. B. Mahdian and S. Saic, “Using noise inconsistencies for blind image forensics”, Elsevier, Image and Vision Computing, vol. 27, issue 10, pp. 1497-1503, Sept. 2009.
15. D. Zoran and Y. Weiss, “Scale Invariance and Noise in Natural Images”, 12th International Conference on Computer Vision, IEEE, pp. 2209-2216, 2009.
16. E. Kee and H. Farid. Exposing digital forgeries from 3-d lighting environments. IEEE International Workshop on Information Forensics and Security, 0:1–6, 2010.
17. I. Amerini, L. Ballan, R. Caldelli, A. D. Bimbo, and G. Serra, “A SIFT-based Forensic Method for Copy-Move Attack Detection and Transformation Recovery,” IEEE Transactions on Information Forensics and Security, vol. 6, no. 3, pp. 1099–1110, Sep. 2011.
18. B. L. Shivakumar and S. Baboo,”Detection of Region Duplication Forgery in Digital Images Using SURF,” International Journal of Computer Science Issues, vol. 8, no. 4, pp. 199–205, 2011.
19. A.N. Myna, M.G. Venkateshmurthy, C.G. Patil, “Detection of region duplication forgery in digital images using wavelets and log-polar mapping”, in: ICCIMA ’07: Proc. of the International Conference on Computational Intelligence and Multimedia Applications (ICCIMA 2007), IEEE Computer Society, Washington, DC, USA, pp. 371–377, 2007.
20. Q. Wu, S. Wang, and X. Zhang, “Detection of Image Region-Duplication with Rotation and Scaling Tolerance”, Proc. of Second International Conference, ICCI, Part-I, Springer, vol. 6421, pp. 100-108, 2010.
21. S. B. Solorio and A. K. Nandi, “Exposing Duplicated Regions Affected by Reflection, Rotation and Scaling,” in International Conference on Acoustics, Speech and Signal Processing, pp. 1880–1883, May 2011.
List of publications
1. K. Sharma P. Abrol and Devanad, “Blind Image Tampering Detection using Noise Inconsistencies”. (Submission in progress)
2. K. Sharma and P. Abrol, “Duplicate Object Detection for Digital Image Authentication”, International Journal of Scientific Research Engineering ; Technology (IJSRET), vol. 5, no. 4, pp. 270-276, April 2016.
3. K. Sharma and P. Abrol, “Transformation Based Parametric Analysis for Copy-Paste Tampering Detection”, International Journal of Scientific and Technical Advancements, vol. 2, no. 1, pp. 247-254, March 2016.
4. K. Sharma and P. Abrol, “Non-Overlapping Block-Based Parametric Forgery Detection Model”, International Journal of Computer Applications, vol. 133, no. 3, pp. 17-24, January 2016.
5. K. Pawar, M. Ishrat and P. Abrol, “Detection and Response to External Stimuli”, International Journal of Scientific and Technical Advancements, vol. 1, no.3, pp. 97-99, September 2015.
6. Kusam, P. Abrol and Devanad, “Digital Tampering Detection Techniques: A Review”, BVICAM’s International Journal of Information Technology, vol. 1, no. 2, pp. 125-132, 2009.
7. Kusam, P. Abrol and Devanad, “Detecting Forgery in Digital Images: A Review”, Researcher, A Multidisciplinary Journal of the University of Jammu, vol. 1, no. 1, pp. 170-180, 2008. (ISSN 2278-9022)
1. K. Sharma, P. Abrol and Devanad, “Block Based Tampering Detection Model for Mirror and Water Reflections”, 12th JK Science Congress, Science and Technology: Emerging Trends and Innovations, University of Jammu, pp. 235, March 2-4, 2017.
2. Kusam, P. Abrol and Devanad, “Experimental Analysis of Copy-Paste Tampering Detection”, 10th JK Science Congress, Science and Technology for Inclusive Development: A Way Forward, University of Jammu, pp. 343-344, March 14-16, 2015.
3. Kusam, P. Abrol and Devanad, “Detecting Duplicated Regions in Digital Images: A Survey of Current Algorithms”, 7th JK Science Congress, Science & Technology: Inspiring Innovation, University of Jammu, pp. 184, October 13-15, 2011.
4. Kusam, P. Abrol and Devanad, “Digital Tampering Detection Techniques: A Review”, Proceedings of the 3rd National Conference on Computing For Nation Development, vol. 1, no. 2, pp. 125-132, February 26 – 27, INDIACom-2009.
5. Kusam, P. Abrol and Devanad, “Detection of Image Manipulation: Techniques and Algorithms”, 2nd National Conference on Next Generation Computing and Information Systems, MIET, Kot Bhalwal, February 14-15, 2009.
6. Kusam, P. Abrol and Devanad, “Investigating Digital Tampering Detection Techniques for Web Applications”, AICTE sponsored National Seminar on Current Trends in Mobile Computing, Baba Ghulam Shah Badshah University, Rajouri, November 28-29, 2008.
Different Tampering Detection Algorithms of Digital Images