Journal of Computer and Communications, 2014, 2, 5157 Published Online March 2014 in SciRes. http://www.scirp.org/journal/jcc http://dx.doi.org/10.4236/jcc.2014.24008 How to cite this paper: Wang, Q., Li, Z.H., Zhang, Z.Z. and Ma, Q.L. (2014) Video InterFrame Forgery Identification Based on Consistency of Correlation Coefficients of Gray Values. Journal of Computer and Communications, 2, 5157. http://dx.doi.org/10.4236/jcc.2014.24008 Video InterFrame Forgery Identification Based on Consistency of Correlation Coefficients of Gray Values Qi Wang, Zhaohong Li, Zhenzhen Zhang, Qinglong Ma School of Electronic and Information Engineering, Beijing Jiaotong University, Beijin g, China Email: 13120030@bjtu.edu.cn, zhhli2@bjtu.edu.cn, 11111053@bjtu.edu.cn, qlma@bjtu.edu.cn Received October 2013 Abstract Identifying interframe forgery is a hot topic in video forensics. In this paper, we propose a me thod based on the assumption that the correlation coefficients of gray values is consistent in an original video, while in forgeries the consistency will be destroyed. We first extract the consisten cy of correlation coefficients of gray values (CCCoGV for short) after normalization and quantiza tion as distinguishing feature to identify interframe forgeries. Then we test the CCCoGV in a large database with the help of SVM (Support Vector Machine). Experimental results show that the pro posed method is efficient in classifying original videos and forgeries. Furthermore, the proposed method performs also pretty well in classifying frame insertion and frame deletion forgeries. Keywords InterFrame Forgeries; Content Co nsis tency; Video Forensics 1. Introduction Nowadays with the ongoing development of video editing techniques, it becomes increasingly easy to modify the digital videos. How to identify the authenticity of videos has become an important field in information secu rity. Video forensics aims to look for features that can distinguish video forgeries from original videos. Thus people can identify the authenticity of a given video. A kind of distinguishing method which is based on video content and composed of copymove detection and interframe tampering detection becomes a hot topic in video forensics. ChihChung et al. [1] used correlation of noise residue to locate forged regions in a video. Leida et al. [2] presented a method to detect the removed object in video by employing the magnitude and orientation of motion vectors. A.V. et al. [3] detected the spatial and temporal copypaste tampering based on Histogram of Oriented Gradients (HOG) feature matching and video compression properties. For interframe forgeries, such as frame insertion and frame deletion forgeries, some effective methods were proposed [4,5]. Juan et al. [4] pre sented a model based optical flow consistency to detect video forgery as interframe forgery will disturb the optical flow consistency. Tianqiang et al. [5] proposed a method using gray values to represent the video content
Q. Wang et al. and detected the tampered video including frame deletion, frame insertion and frame substitution. In this paper, we propose a simple yet efficient method based on consistency of correlation coefficients of gray values (CCCoGV for short in the following text) to distinguish interframe forgeries from original videos. The basic idea of our method is that differences of correlation coefficients of gray values of original sequences are consistent while differences of interframe forgeries will have abnormal points. We first calculate the differ ences of correlation coefficients of gray values between sequential frames of videos, and then use Support Vec tor Machine (SVM) [6] to classify original videos and interframe forgeries. We test our method on a large da tabase, and results show that it is a method of high detection accuracy. 2. Review of Content Continuity In a video which has not been tampered, the correlation of the content between the adjacent frames is high, while the content correlation between the far apart frames is relatively low. Gray value of frame can be a good representative of video content, which reflect the distribution and characteristics of the color and brightness lev el of an image. Based on this idea, a method was proposed to find out which frames has been tampered in a vi deo by using gray value to depict the video content and the CCCoGV to describe whether video content is conti nuous or not [5]. The framework and details of the method [5] is as shown in Figure 1. The main steps are as follows: 1) Decompose the video into orderly image sequence. Video is actually the orderly sequence of the 2D image in the 1D time domain. To extract the content features of a video, we can extract the features of the orderly im age sequence. 2) Transform the RGB color image into gray image. The required storage space of a RGB color image is three times larger than that of a gray image. To reduce the amount of calculation, we use gray image to represent the content of the real color image. In the YUV model, the physical meaning of Y component is brightness of point. The brightness value represents the gray value of image. According to the relationship between RGB model and YUV model, brightness Y can be represented by the R\G\B components as 0.29890.58700.1140 .Y RGB=×+×+× (1) 3) Calculate interframe correlation coefficient between adjacent frames. A gray image can be stored as a sin gle matrix. Each element of the matrix corresponds to one image pixel and represents a different gray level. Eq uation (2) defines the correlation coefficient between two frames. 11 2 2 11 ((, ))((, )), (((,)) )(((,)) ) 1, 2,,1 ijrara rara k kkk k ij ij ra rarara kkk k F ij FFij F rF ij FFij F kn ++ ++ −− ∑∑ = −− ∑∑ ∑∑ = − ( 2) where, is the correlation coefficient between frame k and frame k + 1. represents the gray value of the pixel which is in the row i and column j of the frame k. corresponds to the average value of the gray values . Therefore, from a continuous frame sequence of which the length is n, we can obtain a corre lation sequence of . 4) Calculate the difference between correlation coefficients of adjacent frames. Equation (3) defines the dif ference. Video input Video inputVideo into images Video into imagesColor image into gray image Color image into gray image Calculating interframe correlation coefficients between adjacent frames Calculating interframe correlation coefficients between adjacent frames Calculating the difference of correlation coefficients Calculating the difference of correlation coefficients Outlier detection Outlier detection Video was composited by a continuous image sequence. Video and images are equivalent in content. Video was composited by a continuous image sequence. Video and images are equivalent in content. Between adjacent frames the time interval is short and the correlation is high. While between the far apart frames the time interval is long and the correlation is low. Between adjacent frames the time interval is short and the correlation is high. While between the far apart frames the time interval is long and the correlation is low. The interframe correlation of the video without tampering must be continuous. The interframe correlation of the video without tampering must be continuous. Use the Chebyshev inequality twice. Use the Chebyshev inequality twice. Reduce the amount of computation. RGB color image(256×256×256) Gray image(256) Reduce the amount of computation. RGB color image(256×256×256) Gray image(256) Figure 1. The framework and details of the method [5].
Q. Wang et al. 1 ,2 0, 1 kk k rr k rk + −≥ ∆= = (3 ) By the equation, we can obtain a sequence 123432 1 {,,,,,,,} k nnnn r rrrrrrr −−−− ∆=∆∆∆…∆∆∆∆ to represents the variability of the content of a whole video. As we supposed, the variability of the video without tampering should be consistent, otherwise, the consistency will be destroyed by the deleted or inserted frames. 5) Finally, use the Chebyshev inequality twice to find out abnormal points which indicates where the tam pered frames are. Digital video forgeries detection based on content continuity is efficient. Based on this method, we can detect which frames were tampered in a video. But we cannot identify whether a video has been tampered by the me thod [5]. 3. Proposed Method 3.1. Analysis of CCCoGV for Original Videos and InterFrame Forged Videos We proposed a method based on CCCoGV to classify original videos and interframe forgeries. Gray value is efficient in depicting video content. For a frametampered video, the gray values would be changed a lot at the tampered point. In the original video database, every frame is nontampered. The continuous frames in the original video are as shown in Figure 2. These three frames are continuous in a video. We can see the motion of the people is small and the background almost has no difference. In fact, a whole video is composed by hundreds of frames. The fluctuation of differences between correlation coefficients of adjacent frames is shown in Figure 3. From Figure 3, we can see that the differences of sequential frames are almost stable with values which are a bit larg er than zero. That is to say, differences of correlation coefficients of original frames are consistent. The results are square with the consensus that for videos with fast motion, the correlation coefficient between continuous frames is low, but the change of correlation coefficients is stable. Videos in the 25frameinserted video database and 100fra meinserted video database are inserted 25 or 100 frames somewhere. Figure 4 shows four frames gotten from a 100frameinserted video. The first and forth im ages were the adjacent frames before frame insertion. We insert 100 frames between them then. And the inserted Figure 2. Three continuous frames from an original video. Figure 3. The CCCoGV of original video. 0100 200 300 400 500 0 0.005 0.01 0.015 0.02 0.025 0.03 The Difference Between Correlation Coefficients
Q. Wang et al. 100 frames are continuous. The second and third images are the first and last frames of the video which was in serted to the original video. In a whole 100frameinserted video, the fluctuation of differences between correla tion coefficients of adjacent frames is shown in Figure 5. From Figure 5 we can see two pairs of peaks, which represent two tampered points. At every tampered point, a pair of peaks appears. The front pair of peaks denotes the start point of the insertion, while the latter pair of peaks represents the end point. The fluctuations in these two places are much larger than that of other points. Similar to the frameinserted video, in the 25framedeleted video database and 100framedeleted video da tabase, every video is deleted 25 or 100 frames somewhere. Figure 6 shows three frames gotten from a 100 framedeleted video. The first two images are the frames before frame deletion and the third image is the frame after deleting 100 frames from the original video. In fact, we cannot find much difference between the last two images. In a whole 100framedeleted video, the fluctuation of differences between correlation coefficients of adjacent frames is shown in Figure 7. From Figure 7 we can see one pair of peaks, which represent a tampered point. The pair of peaks denotes the deletion point, where the fluctuation of differences is much larger than that of other points. 3.2. Framework of InterFrame Forgery Identification Scheme In this paper, a CCCoGV based interframe forgery identification scheme is proposed. The framework of our method is as shown in Figure 8. Figure 4. Four continuous frames from a 100frameinserted video. Figure 5. The CCCoGV of 100frameinserted video. Figure 6. Three continuous frames from a 100frame deleted video. 0100200 300400 0 0.2 0.4 0.6 0.8 The Difference Between Correlation Coefficients
Q. Wang et al. Figure 7. The CCCoGV of 100framedeleted video. Original video Original video Video forgery Video forgery The differences of correlation coefficients of gray values The differences of correlation coefficients of gray values The differences of correlation coefficients of gray values The differences of correlation coefficients of gray values Normalization and Quantization Normalization and Quantization Normalization and Quantization Normalization and Quantization Statistical hist distribution Statistical hist distribution Statistical hist distribution Statistical hist distribution SVM SVM Figure 8. The framework of video interframe forgery identification. The detailed scheme of our method is: 1) Calculate the differences between correlation coefficients gray values of adjacent frames. This step is same to the previous work [5]. Equation (13) is used to calculate gray value Y, correlation coeffi cients and difference of correlation coefficients . For a video with M frames, get a vector to store the differences of the video. 2) Normalize and quantize the vector. For each vector of the array, find the maximum m and divide every element in the matrix by m. Then quantify all the elements into D quantization levels. As all of them are distributed from 0 to 1, the quantization interval is 1/D. 3) Count the statistical distribution of differences. After the normalization and quantization step, the elements are all discrete values from 1/D to 1 (1/D, 2/D,…,1). Count the distribution of the discrete values into a vector. For a video database with N videos, build an array contains N vectors of the statistical distribution to represent the CCCoGV features of all the vi deos in the database. 4) Distinguish original videos and video forgeries with Support Vector Machine (SVM) [6]. Put original video array and a video forgery array to the SVM, and distinguish them. Repeat 20 times and get the average classification accuracy. The difference between our method and previous works [5] is the purpose. Their method is used for detecting where was tampered in a video. While our method is for the purpose to identify whether a video was tampered and classify original videos and video forgeries. As their method cannot distinguish the original and forgeries, we improve their method by using normaliza tion and quantization to obtain distinguishing feature of a fix length, and then use SVM to classify original vi deos and interframe forgeries. 4. Experimental Results 4.1. Experimental Object In experiments, we use five video databases. One is original video database. The other four are tampered video databases, of which each separately contains the videos inserted 25 frames, inserted 100 frames, deleted 25 frames and deleted 100 frames. There are 598 videos with still background and a little camera shake in each video database (N = 598). 0100 200 300 400 0 0.1 0.2 0.3 0.4 0.5 The Difference Between Correlation Coefficients
Q. Wang et al. Table 1. Classification accuracy. Forgeries Experiment Video Databases(With original video database) 25frameinsertion 100frameinsertion 25framedeletion 100framedeletion Accuracy 99.22% 99.34% 94.19% 97.27% Table 2. Classifying accuracy of two kinds of forgeries. Forgeries Experiment Video Databases 25frameinsertion and 25framedeletion 100frameinsertion and 100 framedeletion Accuracy 96.21% 95.83% 4.2. Experimental Setting In our experiments, we use SVM of polynomial kernel [6] as the classifier. To train the SVM classifier, about 4/5 of the videos are randomly selected as the training set (480 originals and 480 forgeries). The rest 1/5 form testing set (118 originals and 118 forgeries). The experiments are repeated for 20 times to secure reliable classification results. 4.3. Results and Discussion We extract CCCoGV features of the five video databases, and classify original video database and each of the four tampered video databases separately. The accuracy was as shown in Table 1. Obviously, this method is efficient to identify whether a video was tampered. As expected, the classifying accuracy between originals and frameinserted forgeries is higher than that of framedeleted forgeries. But even the classifying accuracy between originals and 25f r a me deleted forgeries is 94.19%, which is the lowest level of the four experiments. We then try to mix the four kinds of forgeries and classify them from originals. The result is efficient, too. The accuracy is 96.75%. Inspired by this, we further try to classify frameinserted forgeries and framedeleted forgeries. With the same method, we classify the 25fra me inserted videos and 25framedeleted videos, as well as 100frame inserted videos and 100framedeleted videos. The result is as shown in Table 2. We can see that the accuracy still stays high. The method is effective in classifying two kinds of forgeries. Then we mix the 25frameinserted videos and 100frameinserted videos, as well as 25framedeleted videos and 100framedeleted videos. And classify the two kinds of mixed forgeries. The accuracy is 98.79%. To sum up, our method is efficient in classifying original videos and forgeries as well as different kinds of forgeries. 5. Conclusions In this paper, we extract the consistency of correlation coefficients of gray values as distinguishing features to classify original videos and forgeries. The classifying accuracies are high in all of the experiments indicate that the proposed method is efficient in classifying original videos and forgeries yet different kinds of video forge ries. In future work, we will find some new distinguishing features to identify the forgeries. And in view of the fact that the videos we use in the experiments are under still background, we will build a new video database with moved background. References [1] Hsu, C.C., Hung, T.Y., Lin, C.W. and Hsu, C.T. (2008) Video Forgery Detection Using Correlation of Noise Resi due . 2008 IEEE 10th Workshop on Multimedia Signal Processing. [2] Li, L.D., Wang, X.W., Zhang, W., Yang, G.B. and Hu, G.Z. (2013) Detecting Removed Object from Video with Sta
Q. Wang et al. tionary Background. Digital Forensics and Watermarking. Lecture Notes in Computer Science, 7809, 242252. [3] Subramanyam, A.V. and Emmanuel, S. (2012) Video Forgery Detection Using HOG Features and Compression Prop erties. 2012 IEEE 14th International Workshop on Multimedia Signal Processing (MMSP), 8994. [4] Chao, J., Jiang, X.H. and Sun, T.F. (2013) A Novel Video InterFrame Forgery Model Detection Scheme Based on Optical Flow Consistency. Digital Forensics and Watermarking, Springer, Berli n, Heidelberg, 267281. [5] Huang, TQ., Chen, Z.W., Su, L.C., Zheng, Z. and Yuan, X.J. (2011) Digital Video Forgeries Detection Based on Con tent Continuity. Journal of Nanjing University (Natural Science), 47, 493503. [6] Chang, C.C. and Lin, C.J. (2011) LIBSVM: A Library for Support Vector Machines. ACM Transactions on Intelli gent Systems and Technology (TIST), 27, 127. http://www.csie.ntu.edu.tw/~cjlin/libsvm/
