The big data volume of 360-degree videos imposes high requirements for the video compression performance of the encoder. It is noteworthy that spherical rotation demonstrates potential for augmenting the Equirectangul...
详细信息
The big data volume of 360-degree videos imposes high requirements for the video compression performance of the encoder. It is noteworthy that spherical rotation demonstrates potential for augmenting the Equirectangular Projection (ERP) 360-degree video compression rate. However, the compression characteristics of projection under spherical rotation and the calculation of the optimal rotation angle have not been fully studied. To address these problems, we first conducted a comprehensive evaluation of the rotation encoding characteristics of ERP format videos, thereby revealing the impact of rotation angle changes on ERP encoding performance. Based on this, we propose a Basic ERP Rotation Coding Framework (B-ERPRCF). This framework contains two proposed angle prediction algorithms: an ERP Rotation Angle Prediction Algorithm for General Distribution Videos (GD-ERPAPA) and an ERP Rotation Angle Prediction Algorithm for Ideal Distribution Videos (ID-ERPAPA), which predict the optimal rotation angle for different video types. Where GD-ERPAPA considers the geometric distortion and combines motion and texture information to rotate relatively static and flat regions to the poles, and ID-ERPAPA focuses on boundary discontinuity, finding relatively static regions for the boundaries. Furthermore, to optimize the complexity of B-ERPRCF, we propose a High Efficiency ERP Rotation Coding Framework (HE-ERPRCF) that diminishes frequency of updating rotation angles. Experimental results show that B-ERPRCF and HE-ERPRCF can respectively save by 2.90% and 2.63% bitrates on average compared to the standard encoder. Moreover, compared to B-ERPRCF, HE-ERPRCF attains an average reduction in encoding time by 40.51%, achieving efficient ERP rotation coding.
Increased videos captured by widely deployed cameras are being analyzed by computer vision-based Deep Neural Networks (DNNs) on servers rather than being streamed for humans. Unfortunately, the conventional codecs (e....
详细信息
ISBN:
(纸本)9798350383515;9798350383508
Increased videos captured by widely deployed cameras are being analyzed by computer vision-based Deep Neural Networks (DNNs) on servers rather than being streamed for humans. Unfortunately, the conventional codecs (e.g., H.26x and MPEG-x) originally designed for video streaming lack content-aware feature extraction and hinder machine-centric video analytics, making it difficult to achieve the required high accuracy with tolerable delay. Neural codecs (e.g., autoencoder) now hold impressive compression performance and have been widely advocated in video streaming. While autoencoder shows transformative potential, the application in video analytics is hampered by low accuracy in detecting small objects of high-resolution videos and the serious challenges posed by multivideo streaming. To this end, we propose AdaStreamer with adaptive neural codecs to enable real machine-centric high-accuracy multi-video analytics. We also investigate how to achieve optimal accuracy under delay constraints via careful scheduling in compression Ratios (CRs, the ratio of the compressed size to the original data size) and bandwidth allocation, and further propose a Markov-based Adaptive compression and Bandwidth Allocation algorithm (MACBA). We have practically developed a prototype of AdaStreamer, based on which extensive experiments verify its accuracy improvement (up to 15%) compared to state-of-the-art coding and streaming solutions.
Recently, at the SPE Annual Technical Conference and Exhibition, I was asked a question about the ‘engineer of the future.’ As we take a pause to turn the page to next year, I believe now is a prescient time to cont...
详细信息
Recently, at the SPE Annual Technical Conference and Exhibition, I was asked a question about the ‘engineer of the future.’ As we take a pause to turn the page to next year, I believe now is a prescient time to contemplate what’s next for petroleum *** my career, I’ve been able to witness great change in our industry as the shale revolution took hold making horizontal drilling and hydraulic fracturing commonplace, diagnostic and modeling tools improved in access and cost, and data became more timely and readily available. This led to a tremendous period of US energy supply growth and decades-long productivity *** forward, I believe there are three trends that engineers should consider: 1) applying “new” tools to solve “old” problems, 2) connecting the dots via hybrid engineering, and 3) organizing, directing, and inserting data predictions into their workflow. These are not only my views but also the views of Devon’s talented *** problems will meet new solutions. The past decade has been dominated by unconventional horizontal development that was first pioneered in the Barnett Shale for natural gas and shortly thereafter transitioned into oil-focused development in the Bakken and Eagle Ford. Today, the Permian Basin continues to grow in total supply as the stacked pay is *** should take stock of the tool set we have acquired during this era with a mindset to deploy these methodologies to tier II unconventional extensions, previously developed or bypassed conventional formations, offshore and international opportunities, enhanced recovery projects such as EOR and refracturing, and new energy opportunities like geothermal. To make these future opportunities competitive and meet the global demand for energy, we will likely need a combination of price, technology, and cost improvements. Engineers are critical to the latter two. To better illustrate these points let’s utilize examples. Drilling rig specificati
暂无评论