Dai-Kyung Hyun

Forgery Detection for Surveillance Video

In many courts, surveillance videos are used as important legal evidence. Nevertheless, little research is concerned with forgery of surveillance videos. In this paper, we present a forgery detection system for surveillance videos. We analyze the characteristic of surveillance videos. Subsequently, forgeries mainly occur to the surveillance videos are investigated. To identify both RGB and infrared video, Sensor Pattern Noise (SPN) for each video is transformed by Minimum Average Correlation Energy (MACE) filter. Manipulations on the given video are detected by estimating scaling factor and calculating correlation coefficient. Experimental results demonstrate that the proposed scheme is appropriate to identify forgeries of surveillance video.

Detection of upscale-crop and partial manipulation in surveillance video based on sensor pattern noise.

In many court cases, surveillance videos are used as significant court evidence. As these surveillance videos can easily be forged, it may cause serious social issues, such as convicting an innocent person. Nevertheless, there is little research being done on forgery of surveillance videos. This paper proposes a forensic technique to detect forgeries of surveillance video based on sensor pattern noise (SPN). We exploit the scaling invariance of the minimum average correlation energy Mellin radial harmonic (MACE-MRH) correlation filter to reliably unveil traces of upscaling in videos. By excluding the high-frequency components of the investigated video and adaptively choosing the size of the local search window, the proposed method effectively localizes partially manipulated regions. Empirical evidence from a large database of test videos, including RGB (Red, Green, Blue)/infrared video, dynamic-/static-scene video and compressed video, indicates the superior performance of the proposed method.

Camcorder Identification for Heavily Compressed Low Resolution Videos

In this paper, we propose a novel camcorder identification method based on photo-response non-uniformity (PRNU) to perform well with heavily compressed low-resolution videos. The existing methods calculate normalized-cross-correlation (NCC) to measure the similarity between two PRNUs. Since the NCC is very sensitive to noises, these methods show low accuracy for heavily compressed low resolution videos. The proposed method calculates the similarity by minimum average correlation energy (MACE) filter. Our method shows that it identifies source device more accurately than existing algorithm.

Source camcorder identification from cropped and scaled videos

In the field of imaging device identification, the unique property of Photo-Response Non-Uniformity (PRNU) is widely employed. One of disadvantages of the PRNU based methods is sensitive to de-synchronization attacks. In this paper, we propose an improved PRNU based camcorder identification method which performs well with simultaneously cropped and scaled videos. The proposed method solves the out-of-sync problems by achieving downscale-tolerance of Minimum Average Correlation Energy Mellin Radial Harmonic (MACE-MRH) filter. The experimental results demonstrate that the proposed method identifies source devices faster and more accurate than the existing method.

Detecting re-captured videos using shot-based photo response non-uniformity

With advances in digital camcorders, re-capturing commercial videos called camcorder theft is getting a big problem. In this paper, we propose an automatic detection method for re-captured videos based on the photo response non-uniformity (PRNU). To discern a re-captured video, a given video is divided into shots first. Several usable shots are selected and PRNU is estimated from each of the shots. Using peak-to-correlation energy (PCE), a connection matrix, which indicates which shots were recorded with a specific camcorder, is constructed. Then, false negative connections are corrected by using Warshalls algorithm. With the number of connections from connection matrix, the given video is determined whether it was the re-captured or not. The experimental results show that the proposed method performs well even with compressed and scaled re-captured videos.

Digital Imaging Source Identification Using Sensor Pattern Noises

With the advance of IT technology, contents from digital multimedia devices and softwares are widely used and distributed. However, novice uses them for illegal purpose and hence there are needs for protecting contents and blocking illegal usage through multimedia forensics. In this paper, we present a forensic technique for identifying digital imaging source using sensor pattern noise. First, the way to acquire the sensor pattern noise which comes from the imperfection of photon detector against light is presented. Then, the way to identify the similarity of digital imaging sources is explained after estimating the sensor pattern noises from the reference images and the unknown image. For the performance analysis of the proposed technique, 10 devices including DSLR camera, compact camera, smartphone and camcorder are tested and quantitatively analyzed. Based on the results, the proposed technique can achieve the 99.6% identification accuracy.

Fingerprint-PKI authentication using Zernike moments

Bio-cryptographic systems provide security by securing a cryptographic key with biometric data. To secure the cryptographic key, biometric data should be significantly consistent. However, the uncertainty of biometrics makes it complicated to achieve highly accurate authentication. In this paper, we propose a novel bio-cryptographic authentication scheme that improves the security of Public Key Infrastructure. To confirm authenticity of a public key and make authentication information using biometric features consistent, we created a one-way, bio-hash function by using a public-key bitstream and Zernike moments extracted from fingerprint images. Experimental results and analysis show that the proposed method prevents attackers from modifying public-key certificates and getting access to biometric data. Moreover, the results prove that the proposed method achieves very high accuracy.

USER-GENERATED PORNOGRAPHIC VIDEO DETECTION USING SHOT-BASED SENSOR PATTERN NOISE

Illegal distribution of user-generated pornography (UGP) videos raises lots of negative aspects associated with the digital contents. This paper proposes a first forensic technique to detect UGP video. To discriminate UGP videos, we exploit shot based sensor pattern noise (SPN) from an image under investigation. By thresholding average peakto-correlation (PCE) values from the shot based SPNs, the video is decided as a UGP video. Preliminary experiments, which consist of UGP videos, normal videos, and Japanese commercial pornographic videos, indicate adequate performance of the proposed technique.