Traitor tracing over YouTube video service—proof of concept

The development explained in this article proves that is possible to trace dishonest users who upload videos with sensitive content to the YouTube service. To achieve tracing these traitor users, fingerprint marks are embedded by a watermarking algorithm into each copy of the video before distributing it. Our experiments show that if the watermarking algorithm is carefully configured and the fingerprints are correctly chosen, the traitor, or a member of a set of traitors who have performed a collusion attack, can be found from a pirate video uploaded to the YouTube service.     

Video hashing with secondary frames and invariant moments

Video hashing is a useful technique of many multimedia systems, such as video copy detection, video authentication, tampering localization, video retrieval, and anti-privacy search. In this paper, we propose a novel video hashing with secondary frames and invariant moments. An important contribution is the secondary frame construction with 3D discrete wavelet transform, which can reach initial data compression and robustness against noise and compression. In addition, since invariant moments are robust and discriminative features, hash generation based on invariant moments extracted from secondary frames can ensure good classification of the proposed video hashing. Extensive experiments on 8300 videos are conducted to validate efficiency of the proposed video hashing. The results show that the proposed video hashing can resist many digital operations and has good discrimination. Performance comparisons with some state-of-the-art algorithms illustrate that the proposed video hashing outperforms the compared algorithms in classification in terms of receiver operating characteristic results.     

一种基于加速近端梯度法的视频散列算法研究

散列算法已经被广泛应用于视频数据的索引。然而,当前大多数视频散列方法将视频看成是多个独立帧的简单集合,通过综合帧的索引来对每个视频编制索引,在设计散列函数时往往忽略了视频的结构信息。首先将视频散列问题建模为结构正规化经验损失的最小化问题。然后提出一种有监管算法,通过利用结构学习方法来设计高效的散列函数。其中,结构正规化利用了出现于视频帧(与相同的语义类别存在关联)中的常见局部视觉模式,同时对来自同一视频的后续帧保持时域一致性。证明了通过使用加速近端梯度(APG)法可有效求解最小化目标问题。最后,基于两个大规模基准数据集展开全面实验(150 000个视频片断,1 200万帧),实验结果证明了该方法性能优于当前其他算法。     

基于稀疏自编码的无监督哈希算法

最近邻搜索在大规模图像检索中变得越来越重要。在最近邻搜索中,许多哈希方法因为快速查询和低内存被提出。然而,现有方法在哈希函数构造过程中对数据稀疏结构研究的不足,本文提出了一种无监督的稀疏自编码的图像哈希方法。基于稀疏自编码的图像哈希方法将稀疏构造过程引入哈希函数的学习过程中,即通过利用稀疏自编码器的KL距离对哈希码进行稀疏约束以增强局部保持映射过程中的判别性,同时利用L2范数来哈希编码的量化误差。实验中用两个公共图像检索数据集CIFAR-10和YouTube Faces验证了本文算法相比其他无监督哈希算法的优越性。     

Hash-Based Identification of Sparse Image Tampering

In the last decade, the increased possibility to produce, edit, and disseminate multimedia contents has not been adequately balanced by similar advances in protecting these contents from unauthorized diffusion of forged copies. When the goal is to detect whether or not a digital content has been tampered with in order to alter its semantics, the use of multimedia hashes turns out to be an effective solution to offer proof of legitimacy and to possibly identify the introduced tampering. We propose an image hashing algorithm based on compressive sensing principles, which solves both the authentication and the tampering identification problems. The original content producer generates a hash using a small bit budget by quantizing a limited number of random projections of the authentic image. The content user receives the (possibly altered) image and uses the hash to estimate the mean square error distortion between the original and the received image. In addition, if the introduced tampering is sparse in some orthonormal basis or redundant dictionary, an approximation is given in the pixel domain. We emphasize that the hash is universal, e.g., the same hash signature can be used to detect and identify different types of tampering. At the cost of additional complexity at the decoder, the proposed algorithm is robust to moderate content-preserving transformations including cropping, scaling, and rotation. In addition, in order to keep the size of the hash small, hash encoding/decoding takes advantage of distributed source codes.     

基于局部线性嵌入的视频拷贝检测方法

该文提出了基于局部线性嵌入(LLE)的视频哈希方法,该方法首先利用一个图模型选取代表帧,然后以四阶累积量作为视频在高维空间的特征并利用LLE对视频进行降维,利用视频在3维空间中投影点的范数构造视频哈希序列来实现视频拷贝检测。实验证明该方法具有较好的鲁棒性和区分性。     

Deep multi-query video retrieval

Video retrieval methods have been developed for a single query. Multi-query video retrieval problem has not been investigated yet. In this study, an efficient and fast multi-query video retrieval framework is developed. Query videos are assumed to be related to more than one semantic. The framework supports an arbitrary number of video queries. The method is built upon using binary video hash codes. As a result, it is fast and requires a lower storage space. Database and query hash codes are generated by a deep hashing method that not only generates hash codes but also predicts query labels when they are chosen outside the database. The retrieval is based on the Pareto front multi-objective optimization method. Re-ranking performed on the retrieved videos by using non-binary deep features increases the retrieval accuracy considerably. Simulations carried out on two multi-label video databases show that the proposed method is efficient and fast in terms of retrieval accuracy and time.     

Graph-based video fingerprinting using double optimal projection

A double optimal projection method that involves projections for intra-cluster and inter-cluster dimensionality reduction are proposed for video fingerprinting. The video is initially set as a graph with frames as its vertices in a high-dimensional space. A similarity measure that can compute the weights of the edges is then proposed. Subsequently, the video frames are partitioned into different clusters based on the graph model. Double optimal projection is used to explore the optimal mapping points in a low-dimensional space to reduce the video dimensions. The statistics and geometrical fingerprints are generated to determine whether a query video is copied from one of the videos in the database. During matching, the video can be roughly matched by utilizing the statistics fingerprint. Further matching is thereafter performed in the corresponding group using geometrical fingerprints. Experimental results show the good performance of the proposed video fingerprinting method in robustness and discrimination.     

Analyzing the Video Popularity Characteristics of Large-Scale User Generated Content Systems

User generated content (UGC), now with millions of video producers and consumers, is reshaping the way people watch video and TV. In particular, UGC sites are creating new viewing patterns and social interactions, empowering users to be more creative, and generating new business opportunities. Compared to traditional video-on-demand (VoD) systems, UGC services allow users to request videos from a potentially unlimited selection in an asynchronous fashion. To better understand the impact of UGC services, we have analyzed the world's largest UGC VoD system, YouTube, and a popular similar system in Korea, Daum Videos. In this paper, we first empirically show how UGC services are fundamentally different from traditional VoD services. We then analyze the intrinsic statistical properties of UGC popularity distributions and discuss opportunities to leverage the latent demand for niche videos (or the so-called "the Long Tail" potential), which is not reached today due to information filtering or other system scarcity distortions. Based on traces collected across multiple days, we study the popularity lifetime of UGC videos and the relationship between requests and video age. Finally, we measure the level of content aliasing and illegal content in the system and show the problems aliasing creates in ranking the video popularity accurately. The results presented in this paper are crucial to understanding UGC VoD systems and may have major commercial and technical implications for site administrators and content owners.     

感知哈希综述

 感知哈希(Perceptual Hashing),是多媒体数据集到感知摘要集的一类单向映射,即将具有相同感知内容的多媒体数字表示唯一地映射为一段数字摘要,并满足感知鲁棒性和安全性.感知哈希为多媒体内容识别、检索、认证等信息服务方式提供安全可靠的技术支撑.本文在人类感知模型(Human Perceptual Model)的基础上,明确了感知哈希的定义、性质和一般性描述.并对目前感知哈希的典型算法、应用模式以及评测基准等进行了综述,指出了感知哈希未来的研究方向.     

Design and statistical analysis of a hash-aided image watermarking system

This paper develops a joint hashing/watermarking scheme in which a short hash of the host signal is available to a detector. Potential applications include content tracking on public networks and forensic identification. The host data into which the watermark is embedded are selected from a secret subset of the full-frame discrete cosine transform of an image, and the watermark is inserted through multiplicative embedding. The hash is a binary version of selected original image coefficients. We propose a maximum likelihood watermark detector based on a statistical image model. The availability of a hash as side information to the detector modifies the posterior distribution of the marked coefficients. We derive Chernoff bounds on the receiver operating characteristic performance of the detector. We show that host-signal interference can be rejected if the hash function is suitably designed. The relative difficulty of an eavesdropper's detection problem is also determined; the eavesdropper does not know the secret key used. Monte Carlo simulations are performed using photographic test images. Finally, various attacks on the watermarked image are introduced to study the robustness of the derived detectors. The joint hashing/watermarking scheme outperforms the traditional "hashless" watermarking technique.     

Perceptual image hashing via feature points: performance evaluation and tradeoffs.

We propose an image hashing paradigm using visually significant feature points. The feature points should be largely invariant under perceptually insignificant distortions. To satisfy this, we propose an iterative feature detector to extract significant geometry preserving feature points. We apply probabilistic quantization on the derived features to introduce randomness, which, in turn, reduces vulnerability to adversarial attacks. The proposed hash algorithm withstands standard benchmark (e.g., Stirmark) attacks, including compression, geometric distortions of scaling and small-angle rotation, and common signal-processing operations. Content changing (malicious) manipulations of image data are also accurately detected. Detailed statistical analysis in the form of receiver operating characteristic (ROC) curves is presented and reveals the success of the proposed scheme in achieving perceptual robustness while avoiding misclassification.     

The (Non)Violent World of YouTube: Content Trends in Web Video

In this content analysis, we examined violence in Web‐based entertainment. YouTube videos (N = 2,520) were collected in 3 different categories: most viewed, top rated, and random, with additional comparisons between amateur and professional content. Frequencies of violent acts and the context of violence (e.g., characteristics of perpetrator and victim, justification, consequences) were compared both between these categories of YouTube videos and with existing research on television violence. The results showed far less violence as a percentage of programming on YouTube than there is on television. Moreover, the violence that was present showed more realistic consequences and more negative context than television violence. Post hoc comparisons illustrated several differences in the presentation of violence between make and category of video.     

Transparent hashing in the encrypted domain for privacy preserving image retrieval

Search through a database of encrypted images against a crumpled and encrypted query will remain privacy preserving only if comparisons between selective features derived from these images is executed in the encrypted domain itself. To facilitate this, the encryption process must remain transparent to specific image statistics computed in the spatial or transform domain. Consequently, the perceptual hash formed by quantizing the image statistics remains the same before and after the encryption process. In this paper, we propose a transparent privacy preserving hashing scheme tailored to preserve the DCT-AC coefficient distributions, despite a constrained inter-block shuffling operation. These DCT distributions can be mapped onto a generalized Gaussian model characterized by shape and scale parameters, which can be quantized and Gray-coded into a binary hash matrix. The encryption scheme has been shown to be perceptually secure and does not impair the search reliability and accuracy of the hashing procedure. Experimental results have been provided to verify the robustness of the hash to content-preserving transformations, while demonstrating adequate sensitivity to discriminate between different images.     

Double compression detection based on local motion vector field analysis in static-background videos

Videos captured by stationary cameras are widely used in video surveillance and video conference. This kind of video often has static or gradually changed background. By analyzing the properties of static-background videos, this work presents a novel approach to detect double MPEG-4 compression based on local motion vector field analysis in static-background videos. For a given suspicious video, the local motion vector field is used to segment background regions in each frame. According to the segmentation of backgrounds and the motion strength of foregrounds, the modified prediction residual sequence is calculated, which retains robust fingerprints of double compression. After post-processing, the detection and GOP estimation results are obtained by applying the temporal periodic analysis method to the final feature sequence. Experimental results have demonstrated better robustness and efficiency of the proposed method in comparison to several state-of-the-art methods. Besides, the proposed method is more robust to various rate control modes.     

Robust image hash function using local color features

Conventional image hash functions only exploit luminance components of color images to generate robust hashes and then lead to limited discriminative capacities. In this paper, we propose a robust image hash function for color images, which takes all components of color images into account and achieves good discrimination. Firstly, the proposed hash function re-scales the input image to a fixed size. Secondly, it extracts local color features by converting the RGB color image into HSI and YCbCr color spaces and calculating the block mean and variance from each component of the HSI and YCbCr representations. Finally, it takes the Euclidian distances between the block features and a reference feature as hash values. Experiments are conducted to validate the efficiency of our hash function. Receiver operating characteristics (ROC) curve comparisons with two existing algorithms demonstrate that our hash function outperforms the assessed algorithms in classification performances between perceptual robustness and discriminative capability.     

Design and implementation of geo-tagged video search framework

User generated video content is experiencing significant growth which is expected to continue and further accelerate. As an example, users are currently uploading 20 h of video per minute to YouTube. Making such video archives effectively searchable is one of the most critical challenges of multimedia management. Current search techniques that utilize signal-level content extraction from video struggle to scale.Here we present a framework based on the complementary idea of acquiring sensor streams automatically in conjunction with video content. Of special interest are geographic properties of mobile videos. The meta-data from sensors can be used to model the coverage area of scenes as spatial objects such that videos can effectively, and on a large scale, be organized, indexed and searched based on their field-of-views. We present an overall framework that is augmented with our design and implementation ideas to illustrate the feasibility of this concept of managing geo-tagged video.     

Watching user generated videos with prefetching

Even though user generated video sharing sites are tremendously popular, the experience of the user watching videos is often unsatisfactory. Delays due to buffering before and during a video playback at a client are quite common. In this paper, we present a prefetching approach for user-generated video sharing sites like YouTube. We motivate the need for prefetching by performing a PlanetLab-based measurement demonstrating that video playback on YouTube is often unsatisfactory and introduce a series of prefetching schemes: (1) the conventional caching scheme, which caches all the videos that users have watched, (2) the search result-based prefetching scheme, which prefetches videos that are in the search results of users' search queries, and (3) the recommendation-aware prefetching scheme, which prefetches videos that are in the recommendation lists of the videos that users watch. We evaluate and compare the proposed schemes using user browsing pattern data collected from network measurement. We find that the recommendation-aware prefetching approach can achieve an overall hit ratio of up to 81%, while the hit ratio achieved by the caching scheme can only reach 40%. Thus, the recommendation-aware prefetching approach demonstrates strong potential for improving the playback quality at the client. In addition, we explore the trade-offs and feasibility of implementing recommendation-aware prefetching.     

SSIM-based Error Resilient Video Coding Over Packet-Switched Networks

The visual quality is a critical factor in prediction video coding over packet-switched networks. However, the traditional MSE-based error resilient video coding cannot correlate well with the perceptual characteristics of the human visual system (HVS). This paper proposes a structural similarity (SSIM) based error resilient video coding scheme to improve the visual quality of compressed videos over packet-switched networks. In the proposed scheme, a SSIM-based end-to-end distortion model is developed to estimate the perceptual distortion due to quantization, error concealment, and error propagation. Based on this model, an adaptive mode selection strategy is presented to enhance the communication robustness of compressed videos. Experiments show that the proposed scheme significantly improves the visual quality for H.264/AVC video coding over packet-switched networks.     

A penalty function semi-continuous thresholding methods for constraints of hashing problems

The Hashing process is an effective tool for handling large-scale data (for example, images, videos, or multi-model data) retrieval problems. To get better retrieval accuracy, hashing models usually are imposed with three rigorous constraints, i.e., discrete binary constraint, uncorrelated condition, and the balanced constraint, which will lead to being ‘NP-hard’. In this study, we divide the whole constraints set into the uncorrelated (orthogonality) constraint and the binary discrete balance constraint and propose a fast and accurate penalty function semi-continuous thresholding (PFSCT) hash coding algorithm based on forward–backward algorithms. In addition, we theoretically analyze the equivalence between the relaxed model and the original problems. Extensive numerical experiments on diverse large-scale benchmark datasets demonstrate comparable performance and effectiveness of the proposed method.