基于光流周期特性的视频帧率上转篡改检测

视频帧率上转是视频时域篡改的一种常见篡改手段,它通过周期性地在两个视频帧中间插入中间帧的方式,实现将视频由低帧率转换到高帧率的目标.本文提出了一种基于光流周期特性的视频帧率上转篡改检测算法,首先将视频转为帧图像序列,然后采用Horn-Schunck光流法计算每帧图像每个像素点的光流矢量,并计算相邻帧图像光流的变化率.最后利用快速傅里叶变换对光流变化率数据进行频谱分析,根据最高谱线的幅值与平均幅值的比值阈值来判别视频是否经过篡改.实验表明,算法不仅能够准确识别待测视频是否经过帧率上转篡改,并且提高了视频压缩的鲁棒性能,具有一定的实际应用价值.     

Video tampering dataset development in temporal domain for video forgery authentication

Videos are tampered by the forgers to modify or remove their content for malicious purpose. Many video authentication algorithms are developed to detect this tampering. At present, very few standard and diversified tampered video dataset is publicly available for reliable verification and authentication of forensic algorithms. In this paper, we propose the development of total 210 videos for Temporal Domain Tampered Video Dataset (TDTVD) using Frame Deletion, Frame Duplication and Frame Insertion. Out of total 210 videos, 120 videos are developed based on Event/Object/Person (EOP) removal or modification and remaining 90 videos are created based on Smart Tampering (ST) or Multiple Tampering. 16 original videos from SULFA and 24 original videos from YouTube (VTD Dataset) are used to develop different tampered videos. EOP based videos include 40 videos for each tampering type of frame deletion, frame insertion and frame duplication. ST based tampered video contains multiple tampering in a single video. Multiple tampering is developed in three categories (1) 10-frames tampered (frame deletion, frame duplication or frame insertion) at 3-different locations (2) 20-frames tampered at 3- different locations and (3) 30-frames tampered at 3-different locations in the video. Proposed TDTVD dataset includes all temporal domain tampering and also includes multiple tampering videos. The resultant tampered videos have video length ranging from 6 s to 18 s with resolution 320X240 or 640X360 pixels. The database is comprised of static and dynamic videos with various activities, like traffic, sports, news, a ball rolling, airport, garden, highways, zoom in zoom out etc. This entire dataset is publicly accessible for researchers, and this will be especially valuable to test their algorithms on this vast dataset. The detailed ground truth information like tampering type, frames tampered, location of tampering is also given for each developed tampered video to support verifying tampering detection algorithms. The dataset is compared with state of the art and validated with two video tampering detection methods.

     

使用PSNR周期特性检测视频帧率上转伪造

帧率上转（FRUC）是最常用的一种视频编辑技术,它在原始视频帧间周期性地插入新的帧,以便增加视频的帧率,这种技术经常用于两段不同帧率的视频拼接伪造中。为了减少视觉痕迹,高级的FRUC方法通常采用运动补偿的插值方式,这也带来了针对这种插值伪造检测的挑战。在本文,我们提出一种新的简单但有效的方法,可正确检测出这种伪造,并能估计出视频的原始帧率。该方法利用了FRUC算法生成的插值帧与相邻原始帧构成的视频序列再次插值重建得到的帧对在PSNR上的周期性差异。测试序列的实验结果表明本文方法检测准确率高,其中对有损压缩视频序列的测试结果进一步证实了该方法的实际使用价值。     

基于几何均值分解和结构相似度的同源视频时间域复制粘贴篡改快速检测及恢复方法

针对现有方法中篡改检测效率不高、定位不精确的问题,提出了一种基于几何均值分解(GMD)和结构相似度(SSIM)的同源视频复制-粘贴快速篡改检测及恢复的方法。首先,将视频转换为灰度图像序列。其次,将几何均值分解作为检测特征,提出了一个基于块的搜索策略来定位复制序列的起始帧。此外,算法首次将结构相似度用于度量视频两帧之间的相似度,并利用结构相似度对搜索策略得到的起始帧进行复检。由于复制视频序列对应两帧之间的相似度高于未篡改序列对应两帧之间的相似度,提出了一个基于结构相似度的从粗到精的方法来定位复制视频序列的末尾帧。最后,对视频进行恢复。与其他几种经典算法进行对比,实验结果表明,所提方法不仅能够检测经过复制-粘贴篡改操作的视频,而且能准确地定位复制-粘贴序列。此外,该方法在检测精度、召回率和检测时间上有较大提升。     

数字视频区域篡改的检测与定位

目的 数字视频区域篡改是指视频帧图像的某个关键区域被覆盖或被替换,经过图像编辑和修补之后,该关键区域的修改痕迹很难通过肉眼来分辨。视频图像的关键区域承载了视频序列的关键语义信息。如果该篡改操作属于恶意的伪造行为,将产生非常严重的影响和后果。因此,视频区域篡改的检测与定位研究具有重要的研究价值和应用前景。方法 数字图像的复制粘贴篡改检测已经取得较大的研究进展,相关研究成果也很多。但是,数字视频区域篡改的检测与定位不能直接采用数字图像的复制—粘贴篡改取证算法。数字视频区域篡改检测与定位是数字视频被动取证研究领域中的一个新兴的研究方向,近年来越来越多的学者在该领域开展研究工作。目前,数字视频的区域篡改检测与定位研究还缺少完善的理论支撑和通用的检测与定位算法。在广泛调研最近几年的最新研究成果的基础上,对数字视频区域篡改的被动取证概念及重要性进行了介绍,将现有的数字视频区域篡改被动取证算法分为4类：基于噪声模式的算法、基于像素相关性的算法、基于视频内容特征的算法和基于抽象统计特征的算法。然后,对这些区域篡改检测与定位的算法进行对比分析,并介绍现有的视频区域篡改软件和算法,以及篡改检测算法的测试数据库。最后,对本研究领域存在的问题和挑战进行总结,并对未来的研究趋势进行展望。结果 选取了20篇文献中的18种算法,分别介绍每种算法的算法原理,并对这些算法进行对比分析。大部分的算法都宣称可以检测并定位出篡改可疑区域,但是检测和定位的精度、计算复杂度都各有差异。其中,基于时空域的像素相关性分析的算法具有较好的检测和定位效果,并且支持运动背景视频中的运动目标删除篡改检测和定位。基于光流平滑性异常的算法和基于运动目标检测的算法都是基于公开的视频篡改测试库进行比较测试的,两种算法都具有较好的检测和定位效果。基于隐写分析特征提取的集成分类算法虽然只能实现时域上的篡改定位,不能实现更精细的空域篡改定位,但是该算法为基于机器学习的大规模视频篡改取证研究提供了新思路和可能的发展方向,具有较大的指导意义。结论 由于视频编码压缩引入噪声,以及视频区域篡改软件工具和技术的改进,视频区域篡改检测和定位仍是一个极具挑战的课题。未来几年,基于视频内容特征和抽象统计特征的视频区域篡改检测和定位算法,有可能结合深度学习算法,得到进一步的研究和发展;相关的理论算法、系统模型和评价标准等研究成果将逐步完善。     

Malicious inter-frame video tampering detection in MPEG videos using time and spatial domain analysis of quantization effects

In this paper, a new algorithm is proposed for forgery detection in MPEG videos using spatial and time domain analysis of quantization effect on DCT coefficients of I and residual errors of P frames. The proposed algorithm consists of three modules, including double compression detection, malicious tampering detection and decision fusion. Double compression detection module employs spatial domain analysis using first significant digit distribution of DCT coefficients in I frames to detect single and double compressed videos using an SVM classifier. Double compression does not necessarily imply the existence of malignant tampering in the video. Therefore, malicious tampering detection module utilizes time domain analysis of quantization effect on residual errors of P frames to identify malicious inter-frame forgery comprising frame insertion or deletion. Finally, decision fusion module is used to classify input videos into three categories, including single compressed videos, double compressed videos without malicious tampering and double compressed videos with malicious tampering. The experimental results and the comparison of the results of the proposed method with those of other methods show the efficiency of the proposed algorithm.     

智能视频监控系统中的干扰检测及分类

针对智能视频监控系统中的干扰检测问题,提出了一种新的检测方法,并将干扰类型进行了分类.该方法对智能视频监控系统中的遮挡、失焦、亮度异常、偏色和噪声污染五种干扰分别提取检测特征,实现了对不同类型干扰的分类检测.同时,该方法采用了自适应更新阈值的方法,降低了检测方法的复杂度,提高了其实用性.实验结果表明:在能够满足监控系统实时性的要求下,与经典方法相比本文方法的检测性能较好,适用范围较广,分类正确率达到了92.2％.     

融合多特征的视频帧间篡改检测算法

传统的视频帧间被动取证往往依赖单一特征,而这些特征各自适用于某类视频,对其他视频的检测精度较低。针对这种情况,提出一种融合多特征的视频帧间篡改检测算法。该算法首先计算视频的空间信息和时间信息值并对视频进行分组,接着计算视频帧间连续性VQA特征,然后结合SVM–RFE特征递归消除算法对不同特征排序,最后利用顺序前向选择算法和Adaboost二元分类器对排序好的特征进行筛选与融合。实验结果表明,该算法提高了篡改检测精度。     

采用量化离散余弦变换系数检测视频单帧连续多次复制粘贴篡改

目前大多数时域视频帧复制粘贴篡改检测算法都是针对至少20帧以上的视频子序列的复制粘贴篡改,而对单帧复制粘贴篡改无法判断。而根据人眼视觉感知的特性,修改视频内容需要至少15帧以上的帧操作,因此篡改帧想通过单帧复制粘贴篡改来达到想要的效果,必须进行连续多次粘贴操作。为了检测这种篡改方式,针对性地提出了一种基于量化离散余弦变换(DCT)系数的视频单帧连续多次复制-粘贴篡改检测算法。首先,将视频转换为图像,采用量化后的DCT系数作为视频帧图像特征向量,并通过计算巴氏(Bhattacharyya)系数来衡量两相邻帧帧间相似度;再设定阈值来判断两相邻帧帧间相似度是否有异常,最后根据出现相似度异常的帧是否连续,以及连续出现的帧数来判断视频是否经过篡改,并定位篡改位置。实验结果表明,所提算法对不同场景的视频都能检测,不仅检测速度快,而且不受再压缩因素影响,算法的正确率高、漏检率低。     

视频帧内运动目标移除篡改检测算法

针对数字视频帧内对象被移除的篡改操作,提出了一种基于主成分分析（PCA）的篡改检测算法。首先对待测视频帧与基准帧相减得到的差异帧使用稀疏表示方法进行去噪,降低噪声对随后特征提取的干扰;其次将去噪后的视频帧进行非重叠分块,利用主成分分析提取像素点的特征并构造特征向量空间;然后使用k-means算法对特征向量空间进行分类,并将分类结果用二值矩阵表示;最后对二值矩阵进行图像形态学操作得到最终检测结果。实验结果表明所提算法的检测性能指标精确度达到91%、准确度达到100%、F1值达到95.3%,比基于压缩感知的视频篡改检测算法在性能指标上有一定程度的提高。实验证明,对于背景静止的视频,该算法能够检测出帧内运动目标被删除的篡改操作,而且对有损压缩视频具有很好的鲁棒性。     

基于运动矢量细化的帧率上变换与HEVC结合的视频压缩算法

将帧率变换技术与新型视频压缩编码标准HEVC相结合有利于提升视频的压缩效率。针对直接利用HEVC码流信息中的低帧率视频的运动矢量进行帧率上变换时效果不理想的问题,文中提出了一种基于运动矢量细化的帧率上变换与HEVC结合的视频压缩算法。首先,在编码端对原始视频进行抽帧,降低视频帧率;其次,对低帧率视频进行HEVC编解码;然后,在解码端与从HEVC码流中提取出的运动矢量相结合,利用前向-后向联合运动估计对其进行进一步的细化,使细化后的运动矢量更加接近于对象的真实运动;最后,利用基于运动补偿的帧率上变换技术将视频序列恢复至原始帧率。实验结果表明,与HEVC标准相比,所提算法在同等视频质量下可节省一定的码率。同时,与其他算法相比,在节省码率相同的情况下,所提算法重建视频的PSNR值平均可提升0.5 dB。     

A geometrically robust multi-bit video watermarking algorithm based on 2-D DFT

The development of the Internet, together with the progress of multimedia processing techniques, has led to the problems of data piracy, data tampering and illegal dissemination. Digital watermarking is an effective approach to data authentication and copyright protection. This paper proposes a geometrically robust multi-bit video watermarking algorithm based on 2-D DFT (two-dimensional discrete Fourier transform). While most of the existing video watermarking schemes require synchronization to extract the watermark from rotated or scaled videos, which is time-consuming and affects the accuracy, the proposed method can do direct extraction without performing synchronization for videos attacked by rotation, scaling or cropping. For embedding the watermark, circular templates in DFT domain are transformed into spatial masks and added to the video frames in spatial domain. A perceptual model based on local contrast is applied to keep the fidelity of the watermarked video. We also propose an accurate and efficient extraction method which is based on the cross-correlation between the Wiener-filtered DFT magnitude and the stretched template sequence in polar coordinates. Experimental results show that the proposed algorithm is robust against various kinds of attacks, such as compression, filtering, rotation, scaling, cropping, frame averaging and frame rate changing.

     

<Emphasis Type="Bold">Inter-frame forgery detection in H.264 videos using motion and brightness gradients</Emphasis>

In the midst of low cost and easy-to-use multimedia editing software, which make it exceedingly simple to tamper with digital content, the domain of digital multimedia forensics has attained considerable significance. This research domain deals with production of tools and techniques that enable authentication of digital evidence prior to its use in various critical and consequential matters, such as politics, criminal investigations, defense planning. This paper presents a forensic scheme for detection of frame-based tampering in digital videos, especially those captured by surveillance cameras. Frame-based tampering, which involves insertion, removal or duplication of frames into or from video sequences, is usually very difficult to detect via simple visual inspection. Such forgeries, however, disturb the temporal correlation among successive frames of the tampered video. These disturbances, when analyzed in an appropriate manner, help reveal the evidence of forgery. The forensic technique presented in this paper relies on objective analysis of prediction residual and optical flow gradients for the detection of frame-based tampering in MPEG-2 and H.264 encoded videos. The proposed technique is also capable of determining the exact location of the forgery in the given video sequence. Results of extensive experimentation in diverse and realistic forensic set-ups show that the proposed technique can detect and locate tampering with an average accuracy of 83% and 80% respectively, regardless of the number of frames inserted, removed or duplicated.     

基于帧间差异的人脸篡改视频检测方法

近几年,随着计算机硬件设备的不断更新换代和深度学习技术的不断发展,新出现的多媒体篡改工具可以让人们更容易地对视频中的人脸进行篡改。使用这些新工具制作出的人脸篡改视频几乎无法被肉眼所察觉,因此我们急需有效的手段来对这些人脸篡改视频进行检测。目前流行的视频人脸篡改技术主要包括以自编码器为基础的Deepfake技术和以计算机图形学为基础的Face2face技术。我们注意到人脸篡改视频里人脸区域的帧间差异要明显大于未被篡改的视频中人脸区域的帧间差异,因此视频相邻帧中人脸图像的差异可以作为篡改检测的重要线索。在本文中,我们提出一种新的基于帧间差异的人脸篡改视频检测框架。我们首先使用一种基于传统手工设计特征的检测方法,即基于局部二值模式（Local binary pattern,LBP）/方向梯度直方图（Histogram of oriented gradient,HOG）特征的检测方法来验证该框架的有效性。然后,我们结合一种基于深度学习的检测方法,即基于孪生网络的检测方法进一步增强人脸图像特征表示来提升检测效果。在FaceForensics++数据集上,基于LBP/HOG特征的检测方法有较高的检测准确率,而基于孪生网络的方法可以达到更高的检测准确率,且该方法有较强的鲁棒性;在这里,鲁棒性指一种检测方法可以在三种不同情况下达到较高的检测准确率,这三种情况分别是：对视频相邻帧中人脸图像差异用两种不同方式进行表示、提取三种不同间隔的帧对来计算帧间差异以及训练集与测试集压缩率不同。     

Multi-granularity geometrically robust video hashing for tampering detection

The wide-spread video editing tools make it much easier to tamper a video, which raises a huge need for authentication techniques that can prove the originality of video content and locate the tampered regions on the video sequences. In this paper, a multi-granularity geometrically robust video hashing method is proposed for tampering detection and localization. In order to balance the robustness and sensitiveness, we describe a video from three levels of granularity: frame sequence level, block level and pixel level, and then hashes are generated at these three levels. Polar Complex Exponential Transform (PCET) moments are calculated on the low-pass sub-band of 3D Discrete Wavelet Transform (3D–DWT) on frame sequence to extract geometric invariant spatio-temporal hash, which is used for video authentication. Local PCET moments are calculated on annular and angular blocks, which are used for geometric correction and coarse tampering localization. Position information of salient objects is obtained from saliency map for fine tampering localization. Experimental results show that the proposed method is robust against temporal de-synchronization and geometrical transformation, and has high tampering localization accuracy even when the video is rotated. Compared with state-of-the-art methods, it is more robust against content-preserving operations and more sensitive to malicious manipulations.     

基于模式特征的H.264/AVC可逆视频水印

目的 针对当前可逆视频水印隐蔽性和篡改定位能力不足问题,提出一种新颖的基于帧内预测模式的可逆视频水印算法。方法 首先,深入分析I帧亮度分量的预测模式对不同类型篡改的敏感性,提取每个帧内编码宏块的预测模式,通过预测模式生成特征码。然后,结合H.264/AVC编码特性和误差补偿算法,在每个亮度4×4残差块中筛选出误差最小系数。最后,运用差值扩展的方法将特征信息作为水印可逆的嵌入到所选系数。结果 在含水印视频未受到篡改时,解码端提取水印后可对原始视频进行无损恢复。当视频受到篡改时,算法能精确定位篡改区域并且篡改定位精度达到4×4子块级。由于水印嵌入在误差最小的系数中,能够有效地降低水印嵌入对于视频质量的影响,嵌入水印后图像的PSNR值比现有的基于H.264/AVC可逆水印方案平均提高10%,测试序列的码率增量平均降低了22%左右。结论 本文算法较现有算法具有更好的嵌入率、隐蔽性、篡改检测精度, 适用于医学、军事、卫星等领域。     

基于时空融合的多帧压缩视频增强方法

为了减少视频的存储和传输开销,通常对视频进行有损压缩处理以减小体积,往往会在视频中引入各类不自然效应,造成主观质量的严重下降。基于单帧的压缩图像复原方法仅利用当前帧有限的空间信息,效果有限。而现有的多帧方法则大多采用帧间对齐或时序结构来利用相邻帧信息以加强重建,但在对齐性能上仍有较大的提升空间。针对上述问题,提出一种基于多帧时空融合的压缩视频复原方法,通过设计的深度特征提取块和自适应对齐网络实现更优的对齐融合,充分地利用多帧时空信息以重建高质量视频。该方法在公开测试集上(HEVC HM16.5低延时P配置)优于所有对比方法,并在客观指标上(峰值信噪比PSNR)相比于目前最先进的方法 STDF取得了平均0.13 dB的提升。同时,在主观比较上,该方法也取得了领先的效果,重建出更干净的画面,实现了良好的压缩不自然效应去除效果。     

Reducing blocking artifacts in JPEG-compressed images using an adaptive neural network-based algorithm

It is well known that at low bit rates, a block-based discrete cosine transform compressed image or video can exhibit visually annoying blocking and ringing artifacts. Low-pass filters are very effective in reducing the blocking artifacts in smooth areas. However, it is difficult to achieve a satisfactory result for ringing artifact removal using only an adaptive filtering scheme. This paper presents a neural network-based deblocking method that is effective on various types of images. The first step of this scheme is block classification that identifies each 8 × 8 block as one of the three types: PLAIN, EDGE or TEXTURE, based on its statistical characteristics. The next step is the reduction in the blocking and ringing artifacts by applying three trained layered neural networks to three different types of image areas. Comparing this method with other algorithms, the simulation results clearly show that the proposed algorithm is very powerful in effectively reducing both blocking and ringing artifacts while preserving the true edge and textural information and thus significantly improving the visual quality of the blocking images or videos.     

基于压缩感知的视频异源篡改检测

考虑到视频序列固有噪声特征的特点,提出一种基于压缩感知的视频异源篡改检测算法。提取视频中每帧图像的噪声信息并建立噪声矩阵,通过引入压缩感知理论对噪声矩阵进行压缩,极大地降低每帧图像噪声信息的冗余度,对压缩噪声矩阵使用[cos]相似性衡量,得到帧图像间的相似度矩阵,并构造篡改度量,利用参数模型对视频的异源篡改图像进行检测。实验表明提出算法能以较小的压缩比对视频序列中的异源篡改位置进行有效检测,并得到比现有两种算法更高的篡改检测准确率。     

Exposing speech tampering via spectral phase analysis

Audio recordings serve as important evidence in law enforcement context. The most crucial problem in practical scenarios is to determine whether the audio recording is an authentic one or not. For this task, blind audio tampering detection is typically performed based on electric network frequency (ENF) artifacts. In case there is a high level of noise, ENF analysis would become invalid. In this paper, we present a novel approach to detect and locate tampering in uncompressed audio tracks by analyzing the spectral phase across the Short Time Fourier Transform (STFT) sub-bands. Spectral phase reconstruction is employed to counteract the impact of noise. Also, a new feature based on higher order statistics of the spectral phase residual and the spectral baseband phase correlation between two adjacent voiced segments is proposed to allow for an automated authentication. Experimental results show that a significant increase in detection accuracy can be achieved compared to the conventional ENF-based method when the audio recording is exposed to a high level of noise. We also testify that the proposed method remains robust under various noisy conditions.