Mehdi Semsarzadeh

Tampering Detection in Compressed Digital Video Using Watermarking

This paper presents a method to detect video tampering and distinguish it from common video processing operations, such as recompression, noise, and brightness increase, using a practical watermarking scheme for real-time authentication of digital video. In our method, the watermark signals represent the macroblocks and frames indices, and are embedded into the nonzero quantized discrete cosine transform value of blocks, mostly the last nonzero values, enabling our method to detect spatial, temporal, and spatiotemporal tampering. Our method can be easily configured to adjust transparency, robustness, and capacity of the system according to the specific application at hand. In addition, our method takes advantage of content-based cryptography and increases the security of the system. While our method can be applied to any modern video codec, including the recently released high-efficiency video coding standard, we have implemented and evaluated it using the H.264/AVC codec, and we have shown that compared with the existing similar methods, which also embed extra bits inside video frames, our method causes significantly smaller video distortion, leading to a PSNR degradation of about 0.88 dB and structural similarity index decrease of 0.0090 with only 0.05% increase in bitrate, and with the bit correct rate of 0.71 to 0.88 after H.264/AVC recompression.

Complexity Modeling of the Motion Compensation Process of the H.264/AVC Video Coding Standard

With recent advances in computing and communication technologies, ubiquitous access to high quality multimedia content such as high definition video using smart phones, Net books, or tablets is a fact of our daily life. However, power is still a major concern for any mobile device, and requires optimization of power consumption using a power model for each multimedia application, such as a video decoder. In this paper, a generic decoding complexity model for the motion compensation (MC) process, which constitutes up to 25% of the computational complexity and hence power consumption of an H.264/AVC decoder, has been proposed. For the model to remain independent from a specific implementation or platform, it has been developed by analysing the MC algorithm as described in the standard. Simulation results indicate that the proposed model estimates MC complexity with an average accuracy of 95.63%, for a wide range of test sequences using both JM and x.264 software implementations of H.264/AVC. For a dedicated hardware implementation of the MC module the modeling accuracy is around 89.61%, according to our simulation results. It should be noted that in addition to power consumption control, the proposed model can be used for designing a receiver-aware H.264/AVC encoder, where the complexity constraints of the receiver side are taken into account during compression.

A video encoding speed-up architecture for cloud gaming

In cloud-based video gaming systems, game engines are hosted in the cloud, and rendered gaming scenes are streamed to players over the Internet. In such systems, the tasks of rendering graphics and video encoding impose huge computational complexity on cloud servers. Therefore, speeding up the encoding process to meet the stringent requirements of the game becomes a critical issue in cloud-based video gaming systems. In this paper, we analyze the feasibility of developing a mechanism to accelerate the power-intensive process of video encoding, by using available game objects information in game engines. Specifically, we utilize the game engines information about the motion of the objects within the scene in order to bypass the time-consuming procedure of Motion Estimation (ME) in conventional video encoders like H.264/AVC. Based on our analysis, the game engines information could be usable inside a video encoder if an interface is involved to re-shape object information and make them compatible for the video encoder. Our experiments show that our approach accelerates the motion estimation process by 14.32% on average for two specific games, when objects information is taken into account during the encoding phase.

A fine-grain distortion and complexity aware parameter tuning model for the H.264/AVC encoder

Most existing video encoders currently used in mobile applications are unable to gracefully degrade their output quality as the battery life nears its end. In other words, they cannot manage power consumption to efficiently utilize the available power resources. To be able to effectively adapt to changes in the encoders software and hardware platforms, especially due to the power limitations of mobile devices, the effect of encoder parameters on the encoding quality and power consumption has to be represented using a Rate-Distortion-Complexity (R-D-C) model. Most existing R-D-C models only consider macroblock level parameters, and overlook other higher level parameters that may have a more significant impact on complexity. In this paper, the distortion and complexity of the H.264/AVC encoder is controlled considering a subset of higher level encoding parameters consisting of search range, number of reference frames, and motion vector resolution. First, the complexity of full and fast motion estimation methods is modeled in an implementation and platform independent manner. Then, using this complexity model, a common encoding parameter setting table is derived, which leads to the least amount of distortion for each complexity condition. Finally, a complexity control mechanism is proposed which tunes the encoding parameters in a real-time manner. The proposed model can be combined with other existing macroblock level models in order to design a two-phase fine grain complexity controller. Simulation results indicate that when our method is integrated with the direct resource allocation (DRA) approach, performance increases by an average of 1.02dB and 1.06dB for full and fast motion estimation approaches, respectively.

Continuous One-Way Detection of Available Bandwidth Changes for Video Streaming Over Best-Effort Networks

Video streaming over best-effort networks, such as the Internet, is now a significant application used by most Internet users. However, best-effort networks are characterized by dynamic and unpredictable changes in the available bandwidth, which adversely affect the quality of video. As such, it is important to have real-time detection mechanisms of bandwidth changes to ensure that video is adapted to the available bandwidth and transmitted at the highest quality. In this paper, we propose a Bayesian instantaneous end-to-end bandwidth change prediction model and method to detect and predict one-way bandwidth changes at the receiver. Unlike existing congestion detection mechanisms, which use network parameters such as packet loss probability, round trip time (RTT), or jitter, our approach uses weighted interarrival time of video packets at the receiver side. Furthermore, our approach is continuous, since it measures available bandwidth changes with each incoming video packet, and therefore detects congestion occurrence in <;200 ms, on average, which is significantly faster than existing approaches. In addition, it is a one-way scheme, since it only takes into account the characteristics of the incoming path and not the outgoing path, as opposed to other approaches, which use RTT and are hence less accurate. In this paper, we provide extensive experimental simulations and real-world network implementation. Our results indicate that the proposed detection method is superior to existing solutions.

An adaptive rate control for faster bitrate shaping in x264 based video conferencing

In most video streaming applications the available bandwidth may vary during transmission due to the shared nature of IP networks. For efficient bandwidth utilization and in order to manage network congestions, the outgoing bitrate of a streaming application should adapt to changes in network traffic. Current encoders have no strategy for acting in response to bitrate variations that occur as a result of changes in network condition. The time it takes to adapt has a significant impact on the received video quality. A slower response means that more packets are dropped by the network, which in turn results in a lower video quality. Furthermore, a slower response makes the congestion to persist for a longer period of time. In this paper an improved adaptive rate control algorithm for the x264 encoder has been proposed. The proposed scheme is able to adapt the bitrate, by considering the available network capacity, much faster than the conventional method. Simulation results indicate that the proposed algorithm is able to adjust to a bitrate in less than 15 frames, where as it takes more than 42 frames for the conventional method to reach the same bitrate. Consequently, the proposed scheme has no quality degradation or packet loss in %83 of test cases where the bitstream has been exposed to various network conditions, while the conventional rate control reduces quality to 24.02 dB in average in all test cases but one. This improved performance has been achieved with minimum added computational complexity.

A realtime spatio-temporal watermarking scheme for H.264/AVC

This paper introduces a digital video watermarking scheme for real-time authentication of H.264/AVC compressed videos. The watermark signals, which represent the macroblocks (MB) and frames indices, are embedded into the last nonzero quantized DCT value of blocks. Also, analyzing the detection error helps to distinguish malicious attacks from some common processing such as H.264/AVC recompression. Experimental results show that the distortion caused by our watermarking system is negligible, leading to a PSNR of about -0.78 dB and increasing the bitrate by just 3.6%. Furthermore, average bit correct rate (BCR) after H.264/AVC recompression is 0.725.

Continuous one-way available bandwidth change detection in high definition video conferencing

In order to deal, as fast as possible, with available bandwidth variations in High Definition Video Conferencing over best-effort networks, we propose a Bayesian instantaneous end-to-end bandwidth change discovery mechanism. Most other congestion detection mechanisms use network parameters such as packet loss probability, round trip time or jitter. On the other hand, our approach uses weighted inter-arrival time of video packets at the receiver in order to detect bandwidth variations more accurately and more quickly. Our approach is continuous, since it monitors available bandwidth with each incoming video packet, and therefore detects congestion occurrence in less than 200 ms, on average, which is significantly faster than existing RTCP-based approaches. It is also one-way, because it only takes into account the characteristics of the incoming path and not the outgoing path, as opposed to other approaches which use round trip time.

Video Encoding Acceleration in Cloud Gaming

Cloud computing provides reliable, affordable, and flexible resources for many applications and users with constrained computing resources and capabilities. The cloud computing concept is becoming an appealing paradigm for many industries including the gaming industry, leading to the introduction of cloud gaming architectures. Despite its advantages, cloud gaming suffers from unguaranteed end-to-end delay as well as server sides computational complexity. In this paper, a novel algorithm for reducing the computational complexity and hence speeding up the video encoding speed is proposed. Specifically, by performing minimum modifications in the game engine and the video codec, some information from the game engine is fed into the video encoder to bypass the motion estimation (ME) process. Our results show that the proposed method achieves up to 39% speedup in the ME process, leading to a 24% acceleration in the total encoding process.

An enhanced Mean-Shift and LBP-based face tracking method

Face tracking is widely used in many applications. Mean-Shift method is one of the most popular face detection algorithms. In this article, we propose an enhanced version of Mean-Shift face tracking using Local Binary Pattern (LBP) histogram. Simulation results demonstrate that the proposed joint method outperforms both Mean-Shift and LBP histogram methods.