Full HD Video Research Articles

Efficient Real time Zynq 7000 FPGA deployment of optimized YOLOv2 deep leaning model for target detection, based on HDL Coder Methodology

Field Programmable Gate Arrays (FPGAs) have garnered significant attention in the development and enhancement of target identification algorithms that employ YOLOv2 models and FPGAs, owing to their adaptability and user-friendliness. The Simulink HDL compiler was utilized to design, simulate, and implement our proposed design. In an effort to rectify this, this paper presents a comprehensive programming and design proposal. The implementation of the YOLOv2 algorithm for real-time vehicle detection on the Xilinx® Zynq-7000 System-on-a-chip is proposed in this work. Real-time testing of the synthesised hardware revealed that it can process Full HD video at a rate of 16 frames per second. On the Xilinx Zynq-7000 SOC, the estimated dynamic power consumption is less than 90 mW. When comparing the results of the proposed work to those of other simulations, it is observed that resource utilization is enhanced by around 204 k (75%) LUT, 305 (12%) DSP, and 224 k (41%) flip-flops at 200 MHz.

Effects of the Comprehensive and Technical Models of Sports Teaching in Secondary School Students

The aim of this study was to determine the effects of different teaching methodologies (comprehensive vs. technical), developed during a unit of ultimate frisbee, on the physical demands, technical–tactical variables, game performance and physical fitness of secondary school children. Forty-seven students were divided into two groups: Comprehensive Teaching Group (CTG; n = 25, Age: 14.5 ± 0.5 years) and Technical Teaching Group (TTG; n = 22, Age: 14.9 ± 0.8 years). The teaching models (CTG vs. TTG) were composed of 8 sessions of 55 min. Physical fitness was developed with the Alpha-Fitness method, while physical variables were analysed using GPS; the technical and tactical variables performed during the matches were recorded using Full-HD video cameras. The results showed that the CTG students had greater mean speed (p < 0.001), mean heart rate (p = 0.006) and covered more total distance/time (p < 0.001) than the TTG students during sessions. After the intervention, the CTG students showed greater values in good passes (p = 0.001), good catches (p = 0.001), good decisions (p = 0.002) and game performance (p < 0.001), while the TTG students only showed higher values in game performance (p = 0.039). In conclusion, the comprehensive teaching model is shown to be an effective pedagogical option to learn ultimate frisbee because the students achieved a higher overall development in the physical demands expressed in the sessions and in the management of the technical–tactical actions in the offensive and defensive phases of the ultimate frisbee game.

PD34-07 CAN CONVOLUTIONAL NEURAL NETWORK BE APPLIED TO THE PREDICTION OF URINARY STONE COMPOSITION IN SINGLE-USE FLEXIBLE URETEROSCOPIC IMAGES?

You have accessJournal of UrologyCME1 Apr 2023PD34-07 CAN CONVOLUTIONAL NEURAL NETWORK BE APPLIED TO THE PREDICTION OF URINARY STONE COMPOSITION IN SINGLE-USE FLEXIBLE URETEROSCOPIC IMAGES? Kyung Tak Oh, Dae Young Jun, Jae Young Choi, Dae Chul Jung, and Joo Yong Lee Kyung Tak OhKyung Tak Oh More articles by this author , Dae Young JunDae Young Jun More articles by this author , Jae Young ChoiJae Young Choi More articles by this author , Dae Chul JungDae Chul Jung More articles by this author , and Joo Yong LeeJoo Yong Lee More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003327.07AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: This study investigated whether Convolutional neural network (CNN) can show decent results in predicting urinary stone composition even in single-use flexible ureterorenoscopic (fURS) images with relatively low resolution. METHODS: This study retrospectively used surgical videos of fURS performed by a single surgeon using single-use scope (LithoVue) between 2018 and 2021. Digital images of the surface of the stone before laser fragmentation were captured in the surgical full HD video clips. Cases were divided into two groups according to whether they contained any calcium oxalate (Calcium group) or not (Non-calcium group). Among total 512 cases, 207 of stone surface images were finally included. Among total images, 25 were first designated as the test set, then augmentation for Non-calcium group was performed. The train set and validation set were divided by random split in an 8 to 2 ratio. In the CNN model, Resnet-18 model was used, and only endoscopic digital images and stone classification data were input to achieve minimal supervised learning (Fig 1). RESULTS: There were 175 cases in the Calcium group, and 32 cases in the Non-calcium group. After whole training, in validation set, the total accuracy was 82.0%, and recall, specificity and precision for Calcium group was 80.0%, 83.3% and 76.2%, and for Non-calcium group, it was 83.3%, 80.0% and 86.2%, respectively. Area under ROC curve of the model, which represents the classification performance of the model, was 0.863. After training and validation, the model was tested using the test set, and the total accuracy was 84.0%. Recall, specificity and precision of the test result was 80.0%, 100.0% and 100.0% in Calcium group, and 100.0%, 80.0% and 55.6% in Non-calcium group, respectively (Fig 2). CONCLUSIONS: As far as we know, this study is the first artificial intelligence study using a single-use fURS images. Source of Funding: None. © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e922 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Kyung Tak Oh More articles by this author Dae Young Jun More articles by this author Jae Young Choi More articles by this author Dae Chul Jung More articles by this author Joo Yong Lee More articles by this author Expand All Advertisement PDF downloadLoading ...

Impact of Packet Loss Rate on Quality of Compressed High Resolution Videos.

Video delivered over IP networks in real-time applications, which utilize RTP protocol over unreliable UDP such as videotelephony or live-streaming, is often prone to degradation caused by multiple sources. The most significant is the combined effect of video compression and its transmission over the communication channel. This paper analyzes the adverse impact of packet loss on video quality encoded with various combinations of compression parameters and resolutions. For the purposes of the research, a dataset containing 11,200 full HD and ultra HD video sequences encoded to H.264 and H.265 formats at five bit rates was compiled with a simulated packet loss rate (PLR) ranging from 0 to 1%. Objective assessment was conducted by using peak signal to noise ratio (PSNR) and Structural Similarity Index (SSIM) metrics, whereas the well-known absolute category rating (ACR) was used for subjective evaluation. Analysis of the results confirmed the presumption that video quality decreases along with the rise of packet loss rate, regardless of compression parameters. The experiments further led to a finding that the quality of sequences affected by PLR declines with increasing bit rate. Additionally, the paper includes recommendations of compression parameters for use under various network conditions.

An Efficient Recurrent Adversarial Framework for Unsupervised Real-Time Video Enhancement

Video enhancement is a challenging problem, more than that of stills, mainly due to high computational cost, larger data volumes and the difficulty of achieving consistency in the spatio-temporal domain. In practice, these challenges are often coupled with the lack of example pairs, which inhibits the application of supervised learning strategies. To address these challenges, we propose an efficient adversarial video enhancement framework that learns directly from unpaired video examples. In particular, our framework introduces new recurrent cells that consist of interleaved local and global modules for implicit integration of spatial and temporal information. The proposed design allows our recurrent cells to efficiently propagate spatio-temporal information across frames and reduces the need for high complexity networks. Our setting enables learning from unpaired videos in a cyclic adversarial manner, where the proposed recurrent units are employed in all architectures. Efficient training is accomplished by introducing one single discriminator that learns the joint distribution of source and target domain simultaneously. The enhancement results demonstrate clear superiority of the proposed video enhancer over the state-of-the-art methods, in all terms of visual quality, quantitative metrics, and inference speed. Notably, our video enhancer is capable of enhancing over 35 frames per second of FullHD video (1080x1920).

High-Performance Image Acquisition and Processing for Stereoscopic Diagnostic Systems with the Application of Graphical Processing Units.

In recent years, cinematography and other digital content creators have been eagerly turning to Three-Dimensional (3D) imaging technology. The creators of movies, games, and augmented reality applications are aware of this technology’s advantages, possibilities, and new means of expression. The development of electronic and IT technologies enables the achievement of a better and better quality of the recorded 3D image and many possibilities for its correction and modification in post-production. However, preparing a correct 3D image that does not cause perception problems for the viewer is still a complex and demanding task. Therefore, planning and then ensuring the correct parameters and quality of the recorded 3D video is essential. Despite better post-production techniques, fixing errors in a captured image can be difficult, time consuming, and sometimes impossible. The detection of errors typical for stereo vision related to the depth of the image (e.g., depth budget violation, stereoscopic window violation) during the recording allows for their correction already on the film set, e.g., by different scene layouts and/or different camera configurations. The paper presents a prototype of an independent, non-invasive diagnostic system that supports the film crew in the process of calibrating stereoscopic cameras, as well as analysing the 3D depth while working on a film set. The system acquires full HD video streams from professional cameras using Serial Digital Interface (SDI), synchronises them, and estimates and analyses the disparity map. Objective depth analysis using computer tools while recording scenes allows stereographers to immediately spot errors in the 3D image, primarily related to the violation of the viewing comfort zone. The paper also describes an efficient method of analysing a 3D video using Graphics Processing Unit (GPU). The main steps of the proposed solution are uncalibrated rectification and disparity map estimation. The algorithms selected and implemented for the needs of this system do not require knowledge of intrinsic and extrinsic camera parameters. Thus, they can be used in non-cooperative environments, such as a film set, where the camera configuration often changes. Both of them are implemented with the use of a GPU to improve the data processing efficiency. The paper presents the evaluation results of the algorithms’ accuracy, as well as the comparison of the performance of two implementations—with and without the GPU acceleration. The application of the described GPU-based method makes the system efficient and easy to use. The system can process a video stream with full HD resolution at a speed of several frames per second.

RT-BVE—Real-Time no-reference Blocking Visibility Estimation in video frames

During lossy video compression, a blocking artifact is introduced into a video. It can significantly reduce the quality of a compressed video, and consequently end-user Quality of Experience. Hence, it is very important to measure and monitor the quality of compressed videos delivered to the end-user, which can be performed only by using the no-reference (NR) approach, without the original video signal available. Many NR video quality assessment (VQA) metrics are based on measurements of different artifacts visibility, which are then used as an input for an objective NR VQA metric. Thus, reliable artifacts detection and their visibility estimation are the first step in a reliable VQA process. In this paper, a new algorithm for Real-Time NR Blocking Visibility Estimation (RT-BVE) in video frames is proposed. The performance of the proposed RT-BVE algorithm is compared to the performance of two freely available algorithms, i.e., BBT-BDA and MSU Blocking, using videos from the well-known databases (156 video sequences in total and more than 48,000 frames). While RT-BVE achieves performance comparable to BBT-BDA and MSU Blocking algorithms for MPEG-2 videos, it outperforms both algorithms for H.264 videos. The results show that the proposed RT-BVE algorithm can precisely estimate blocking visibility in video frames for videos compressed according to different compression standards by using a fixed set of algorithm parameter values. Additionally, RT-BVE is suitable for usage in real-time applications since it is capable of processing approximately 100 Full HD video frames per second on the current midrange × 86–64 platform.

Fast background subtraction with adaptive block learning using expectation value suitable for real-time moving object detection

This paper presents a method of moving object detection through a fast background subtraction technique suitable for real-time performance in wide range of platforms. An intermittent background update using adaptive blocks individually calculates the learning rate through expected difference values. Then, coupled with a fast background subtraction process, the design achieves fast throughput with well-rounded performance. To compensate for the lagging effects of intermittent background update, an adaptation bias is devised to improve precision and recall metrics. Experiments show a versatile performance in varying scenes with overall results better than conventional techniques. The proposed method achieved a fast execution speed of up to 56 fps in PC using Full HD video. It also achieved 655 fps and 83 fps in PC and ARM core-embedded platform, respectively, using the minimum input resolution of 320 × 240. Overall, it is suitable for real-time performance applications.

Underwater Visual Census of Deeper Vertical Rocky Reefs

Diver-based underwater visual census (UVC), particularly transect-based survey, is a widely used method for the study of tropical and temperate fish assemblages. However, due to logistical constraints associated with conventional SCUBA diving, deeper habitats, such as vertical rocky reefs, are rarely studied and poorly known. This paper describes the Deep Vertical Transect (DVT) method as a safe and effective method for assessing fish in waters up to 50 m of depth. It is based on sampling of vertical transects browsing within it (S-type transect) by divers using Full HD video cameras. The diving profile includes the use of deep decompression stops and Nitrox 50 as a decompression gas. Hence, the study yields information on fish assemblages associated with deeper vertical coralligenous reefs. The results of 51 recorded species, yielded 41 considered as reef-associated and 10 as occasional. This suggests that underwater steep coralligenous reefs are marine biodiversity hotspots. They may be considered to represent a distinctive marine subecosystem, possessing its own food chain, with the depth, in relation to temperature, as the most important factor responsible for the diversity of fish assemblages within this habitat.

Hardware implementation of real-time pedestrian detection system

Histogram of oriented gradients (HOG) descriptor and support vector machine (SVM) classifier have been considered as a very efficient and robust tools used in the development of pedestrian detection systems. However, the enormous calculations make it difficult to meet real-time requirement specifically for advanced driver assistance system applications. The latter requires higher frame rate and resolution. In the present work, efficient pipelined architecture founded on cell-based scanning and simultaneous linear SVM classification is highlighted. The proposed real-time pedestrian detection architecture performs without external memories, and only 3 × 3 pixels are processed using two line buffers. Indeed, intermediate results will contribute directly to build the cell histograms. Consequently, it allows to reduce the required memory and to speed up the HOG feature extraction procedure. The implementation results on a heterogeneous platform show that the proposed architecture achieves an efficient pedestrian detection for full-HD video recordings (1080 × 1920 pixels) at 60 fps with reduced hardware resource consumption. Using the same hardware resources at 444 MHz, the proposed system turns out to handle a HD video at 180 fps.

An Approximate HEVC Intra Angular Prediction Hardware

In this paper, an approximate High Efficiency Video Coding (HEVC) intra angular prediction technique is proposed for reducing area of HEVC intra prediction hardware. The proposed approximation technique uses closer neighboring pixels instead of distant neighboring pixels in intra angular prediction equations. It causes 0.0569% bit rate increase and 0.0028 dB PSNR loss on average. In this paper, an approximate HEVC intra angular prediction hardware implementing the proposed approximation technique is also proposed. FPGA and ASIC implementations of the proposed approximate hardware can process 24 and 40 quad full HD (3840 × 2160) video frames per second, respectively. The proposed approximate HEVC intra angular prediction hardware is the smallest and the second fastest HEVC intra prediction hardware in the literature. It is ten times smaller and 20% slower than the fastest HEVC intra prediction hardware in the literature.

AUTOMATIC LIP-READING OF HEARING IMPAIRED PEOPLE

Abstract. Inability to use speech interfaces greatly limits the deaf and hearing impaired people in the possibility of human-machine interaction. To solve this problem and to increase the accuracy and reliability of the automatic Russian sign language recognition system it is proposed to use lip-reading in addition to hand gestures recognition. Deaf and hearing impaired people use sign language as the main way of communication in everyday life. Sign language is a structured form of hand gestures and lips movements involving visual motions and signs, which is used as a communication system. Since sign language includes not only hand gestures, but also lip movements that mimic vocalized pronunciation, it is of interest to investigate how accurately such a visual speech can be recognized by a lip-reading system, especially considering the fact that the visual speech of hearing impaired people is often characterized with hyper-articulation, which should potentially facilitate its recognition. For this purpose, thesaurus of Russian sign language (TheRusLan) collected in SPIIRAS in 2018–19 was used. The database consists of color optical FullHD video recordings of 13 native Russian sign language signers (11 females and 2 males) from “Pavlovsk boarding school for the hearing impaired”. Each of the signers demonstrated 164 phrases for 5 times. This work covers the initial stages of this research, including data collection, data labeling, region-of-interest detection and methods for informative features extraction. The results of this study can later be used to create assistive technologies for deaf or hearing impaired people.

Encoding Shaky Videos by Integrating Efficient Video Stabilization

This paper presents a novel video coding method by integrating video stabilization for shaky videos. By reusing the stabilized motion of feature points and geometric transformations, a better predictor can be established to replace the original motion vectors in the motion estimation stage of video coding. Then, these motion vectors are optimized based on statistics of the residuals between the stabilization predictor and the standard one to improve the efficiency of motion search. As a result, our method brings much less computational cost to motion estimation than encoding the stabilized frames separately (e.g., 24% for the enhanced predictive zonal search algorithm, 17% for the unsymmetrical-cross multi-hexagon-grid search algorithm), while the Bjontegaard distortion (BD) bit rate and BD-peak signal-to-noise ratio still have the comparable performance with multiple quantization parameters. Specially, the implementation of our integrative system based on $\times 264$ is very fast and of low latency, by which full HD videos can be simultaneously stabilized and encoded with more than 30 frames per second in the fastest mode, even on the mobile platform. The experiments on a variety of shaky videos demonstrate the potential of our method in terms of effectiveness and efficiency.

High-Level Synthesis of Online K-Means Clustering Hardware for a Real-Time Image Processing Pipeline.

The growing need for smart surveillance solutions requires that modern video capturing devices to be equipped with advance features, such as object detection, scene characterization, and event detection, etc. Image segmentation into various connected regions is a vital pre-processing step in these and other advanced computer vision algorithms. Thus, the inclusion of a hardware accelerator for this task in the conventional image processing pipeline inevitably reduces the workload for more advanced operations downstream. Moreover, design entry by using high-level synthesis tools is gaining popularity for the facilitation of system development under a rapid prototyping paradigm. To address these design requirements, we have developed a hardware accelerator for image segmentation, based on an online K-Means algorithm using a Simulink high-level synthesis tool. The developed hardware uses a standard pixel streaming protocol, and it can be readily inserted into any image processing pipeline as an Intellectual Property (IP) core on a Field Programmable Gate Array (FPGA). Furthermore, the proposed design reduces the hardware complexity of the conventional architectures by employing a weighted instead of a moving average to update the clusters. Experimental evidence has also been provided to demonstrate that the proposed weighted average-based approach yields better results than the conventional moving average on test video sequences. The synthesized hardware has been tested in real-time environment to process Full HD video at 26.5 fps, while the estimated dynamic power consumption is less than 90 mW on the Xilinx Zynq-7000 SOC.

Efficient Computation Offloading in Mobile Cloud Computing for Video Streaming Over 5G

In this paper, we investigate video quality enhancement using computation offloading to the mobile cloud computing (MCC) environment. Our objective is to reduce the computational complexity required to covert a low-resolution video to high-resolution video while minimizing computation at the mobile client and additional communication costs. To do so, we propose an energy-efficient computation offloading framework for video streaming services in a MCC over the fifth generation (5G) cellular networks. In the proposed framework, the mobile client offloads the computational burden for the video enhancement to the cloud, which renders the side information needed to enhance video without requiring much computation by the client. The cloud detects edges from the upsampled ultra-high-resolution video (UHD) and then compresses and transmits them as side information with the original low-resolution video (e.g., full HD). Finally, the mobile client decodes the received content and integrates the SI and original content, which produces a high-quality video. In our extensive simulation experiments, we observed that the amount of computation needed to construct a UHD video in the client is 50%-60% lower than that required to decode UHD video compressed by legacy video encoding algorithms. Moreover, the bandwidth required to transmit a full HD video and its side information is around 70% lower than that required for a normal UHD video. The subjective quality of the enhanced UHD is similar to that of the original UHD video even though the client pays lower communication costs with reduced computing power.

Multi-modal Asian Conversation Mobile Video Dataset for Recognition Task

Images, audio, and videos have been used by researchers for a long time to develop several tasks regarding human facial recognition and emotion detection. Most of the available datasets usually focus on either static expression, a short video of changing emotion from neutral to peak emotion, or difference in sounds to detect the current emotion of a person. Moreover, the common datasets were collected and processed in the United States (US) or Europe, and only several datasets were originated from Asia. In this paper, we present our effort to create a unique dataset that can fill in the gap by currently available datasets. At the time of writing, our datasets contain 10 full HD (1920 1080) video clips with annotated JSON file, which is in total 100 minutes of duration and the total size of 13 GB. We believe this dataset will be useful as a training and benchmark data for a variety of research topics regarding human facial and emotion recognition.

No-reference real-time video transmission artifact detection for video signals

Video signals are a very important part of multimedia applications. Due to limited network bandwidth, video signals are subjected to the compression process, which introduces different compression artifacts. During network transmission, additional artifacts are introduced in video signals due to random bit errors and packet loss (PL). Both mentioned artifact types degrade visual quality of the video signal and thus, it has to be continuously monitored to ensure the required quality of service (QoS) provided to end users. An important component of the video quality monitoring system deals with video transmission artifact detection. In this paper, a no-reference (NR) pixel-based video transmission artifact detection algorithm is proposed, called the packet loss area measure (PLAM) algorithm. When detecting video transmission artifacts, the PLAM algorithm takes into account spatial and temporal information of a video signal. The performance of the proposed PLAM algorithm has been compared to those of the three existing different PL detection algorithms on a broad set of significantly different video signals from two publicly available video databases. One of these databases, called the Referent Packet Loss (RPL) database, has been created within this research and is presented in this paper. The algorithm performance testing results show that PLAM achieves high performance and overcomes other tested algorithms. Furthermore, the results show that the PLAM algorithm is very robust when detecting video transmission artifacts in video signals of different contents, with distinct degradation levels and PL error-concealment methods used in decoder post-processing. Due to its low computational complexity, the PLAM algorithm is capable of processing Full HD and Ultra HD video signals with the frame rate up to 100 and 25 frames per second (fps), respectively, in real time, in the case when high-end CPU is used.

High definition video loggers provide new insights into behaviour, physiology, and the oceanic habitat of a marine predator, the yellow-eyed penguin.

Camera loggers are increasingly used to examine behavioural aspects of free-ranging animals. However, often video loggers are deployed with a focus on specific behavioural traits utilizing small cameras with a limited field of view, poor light performance and video quality. Yet rapid developments in consumer electronics provide new devices with much improved visual data allowing a wider scope for studies employing this novel methodology. We developed a camera logger that records full HD video through a wide-angle lens, providing high resolution footage with a greater field of view than other camera loggers. The main goal was to assess the suitability of this type of camera for the analysis of various aspects of the foraging ecology of a marine predator, the yellow-eyed penguin in New Zealand. Frame-by-frame analysis allowed accurate timing of prey pursuits and time spent over certain seafloor types. The recorded video footage showed that prey species were associated with certain seafloor types, revealed different predator evasion strategies by benthic fishes, and highlighted varying energetic consequences for penguins pursuing certain types of prey. Other aspects that could be analysed were the timing of breathing intervals between dives and observe exhalation events during prey pursuits, a previously undescribed behaviour. Screen overlays facilitated analysis of flipper angles and beat frequencies throughout various stages of the dive cycle. Flipper movement analysis confirmed decreasing effort during descent phases as the bird gained depth, and that ascent was principally passive. Breathing episodes between dives were short (<1 s) while the majority of the time was devoted to subsurface scanning with a submerged head. Video data recorded on free-ranging animals not only provide a wealth of information recorded from a single deployment but also necessitate new approaches with regards to analysis of visual data. Here, we demonstrate the diversity of information that can be gleaned from video logger data, if devices with high video resolution and wide field of view are utilized.

Comprehensive Data Set for Automatic Single Camera Visual Speed Measurement

In this paper, we focus on traffic camera calibration and a visual speed measurement from a single monocular camera, which is an important task of visual traffic surveillance. Existing methods addressing this problem are difficult to compare due to a lack of a common data set with reliable ground truth. Therefore, it is not clear how the methods compare in various aspects and what factors are affecting their performance. We captured a new data set of 18 full-HD videos, each around 1 hr long, captured at six different locations. Vehicles in the videos (20865 instances in total) are annotated with the precise speed measurements from optical gates using LiDAR and verified with several reference GPS tracks. We made the data set available for download and it contains the videos and metadata (calibration, lengths of features in image, annotations, and so on) for future comparison and evaluation. Camera calibration is the most crucial part of the speed measurement; therefore, we provide a brief overview of the methods and analyze a recently published method for fully automatic camera calibration and vehicle speed measurement and report the results on this data set in detail.

Power-Efficient Sum of Absolute Differences Hardware Architecture Using Adder Compressors for Integer Motion Estimation Design

Sum of absolute differences (SAD) calculation is one of the most time-consuming operations of video encoders compatible with the high efficiency video coding standard. SAD hardware architectures employ an adder tree to accumulate the coefficients from absolute difference between two video blocks. This paper exploits different adder compressors structures into the SAD hardware architecture. The architectures were synthesized to 45-nm CMOS standard cells. Synthesis results show that SAD architecture using 8–2 compressor composed with 4–2 compressors and Kogge–Stone adder in the recombination line reduces power dissipation by 25.5% on average when compared with the SAD architecture using conventional adders from a state-of-the-art synthesis tool. Our throughput analysis shows that the designed SAD units are capable of encoding full HD ( $1920\times 1080$ ) videos in real time at 30 frames/s.