ABSTRACT
Due to enhanced camera performance and easy using convenience, most videos are taken and distributed by smart phones. However, smart phone video file as an evidence needs to be verified. We analyzed digital videos with basic information (resolution, shooting time, frame rate, bit depth, sample rate, and so on) and compressed patterns with unique compression algorithm adjusted to certain models. We also classified them based on their original characters. For analysis, we first estimated the virtual data quantity of a real digital video EXIF information without compression. We then analyzed the correlation between real recorded data and virtual data to calculate the original unique compression pattern. We then proposed a technical standard to identify the origin by comparing original compression ratio and remade video's compression ratio through a video editing software.
ABSTRACT
This study designs a method of identifying the camera model used to take videos that are distributed through mobile phones and determines the original version of the mobile phone video for use as legal evidence. For this analysis, an experiment was conducted to find the unique characteristics of each mobile phone. The videos recorded by mobile phones were analyzed to establish the delay time of sound signals, and the differences between the delay times of sound signals for different mobile phones were traced by classifying their characteristics. Furthermore, the sound input signals for mobile phone videos used as legal evidence were analyzed to ascertain whether they have the unique characteristics of the original version. The objective of this study was to find a method for validating the use of mobile phone videos as legal evidence using mobile phones through differences in the delay times of sound input signals.
ABSTRACT
Because of the rapidly increasing use of digital composite images, recent studies have identified digital forgery and filtering regions. This research has shown that interpolation, which is used to edit digital images, is an effective way to analyze digital images for composite regions. Interpolation is widely used to adjust the size of the image of a composite target, making the composite image seem natural by rotating or deforming. As a result, many algorithms have been developed to identify composite regions by detecting a trace of interpolation. However, many limitations have been found in detection maps developed to identify composite regions. In this study, we analyze the pixel patterns of noninterpolation and interpolation regions. We propose a detection map algorithm to separate the two regions. To identify composite regions, we have developed an improved algorithm using minimum filer, Laplacian operation and maximum filters. Finally, filtering regions that used the interpolation operation are analyzed using the proposed algorithm.
ABSTRACT
Development of digital image-editing programs has enabled us to be widely exposed to forged digital images surrounding us. Such forged images have been dispersed through the Internet, newspaper articles, and magazines, and in particular, the information contained in these unverified images happened to be regarded as true. As a result, the forged images provided wrong information for individuals and society, thus sometimes creating social issues. In order to solve such problems, this study was aimed to suggest the methods of identifying the veracity of forged images. In this way, it suggested re-interpolation algorithm. Namely, the study re-interpolated in identical arbitrary values both the interpolated and un-interpolated regions based on the interpolation used a lot in forged, confirmed discrete fourier transform (DFT) characteristics of these two regions, and embodied a detection map for the final forged images, using the subtraction value between two regions in DFT characteristics.
