Application research of image compression and wireless network traffic video streaming

https://doi.org/10.1016/j.jvcir.2018.12.042Get rights and content

Abstract

With the increasingly busy urban traffic and the development of modern communication technologies, traffic conditions need to be transmitted from major intersections to command and dispatch centers for analysis and processing, which raises a large number of problems of storing and transmitting static images of traffic conditions. Research on image compression of traffic conditions has also become a hot issue that people pay more and more attention to. In the process of traffic image research, due to the lack of essential attributes of the image, especially, the selection and use of compression methods has greater blindness. However, an overall analysis of the image prior to traffic image processing is a difficult task. This article selects the road traffic data, public transit data and orbital data first to compress the image. Then the streaming Media transmission System of DASH is introduced. In the specific application, the code of traffic data flow in this paper is converted into Real Media format through SDK. With the help of Helix Server, all traffic data flow files can be integrated into the synchronous Media integration language, based on the Internet of TCP/IP which is released in a stream through Real System. The experimental results show that the traffic conditions such as vehicle queuing, congestion and signal lights are directly mastered, the signal timing is timely adjusted or other means are adopted to ease the traffic, the distribution of traffic flow is changed, and ordinary terminal users are enabled to master the distribution of traffic flow through wireless network and choose the travel path actively.

Introduction

In the field of multimedia application, if the high-capacity multimedia information can be compressed effectively, it will bring great convenience to the automation office, digital TV and home multimedia digital entertainment. Internet fax and E-mail replace the letter exchanges of the postal system, which is not only convenient and swift, but also reduces the cost in terms of large capacity and easy storage. Electronic books and electronic newspapers with floppy disks and compact disks show great advantages. A single disc can store color images and compressed black and white images, and can hold multiple volumes of Chinese encyclopedia, while the storage capacity of the disc can be larger. With this capacity and volume, a few thin discs make up a small library, and the retrieval is simple and convenient, which is beyond the reach of the printing media.

In the military field, if there is a fast and efficient image compression algorithm, it will bring more efficiency and accuracy to satellite remote sensing, satellite reconnaissance and missile guidance.

In modern communication, image transmission has also become an important content. In addition to the requirement of reliable equipment and high fidelity at work, real-time performance will be one of the important indicators. Obviously, under the premise of constant channel bandwidth and communication link capacity, using coding compression technology to reduce data transmission is an important means to improve communication speed. Therefore, research on image compression algorithm has practical application significance in many aspects.

HAF Almurib et al. [1] proposed a new digital image processing framework; It relies on imprecise calculations to solve some of the challenges associated with Discrete Cosine Transform (DCT) compression. The proposed framework has three levels of processing; The first stage uses approximately DCT for image compression to eliminate all computationally intensive floating-point multiplication, performs DCT processing through integer addition, and performs logical right/left shift in some cases. The second level further reduces the amount of data that needs to be processed (from the first level) by filtering out frequencies that cannot be detected by human senses. Finally, to reduce power consumption and latency, the third stage introduces a circuit-level inexact adder to calculate the DCT. For evaluation, a set of standardized images is compressed using the proposed three-level framework. Different advantages (such as energy consumption, delay, power signal-to-noise ratio, average difference, and absolute maximum difference) are compared with existing compression methods; error analysis is also performed to confirm the simulation results. The results show that there are very good improvements in reducing energy and delay while maintaining an acceptable level of accuracy for image processing applications.

Multiple Description Coding (MDC) is capable of steadily transmitting signals in unreliable and non-prioritized networks, which has been extensively studied for decades. However, traditional MDC does not make good use of the context function of the image to generate multiple descriptions. An MDC standard framework for convolutional neural networks based on image context features is proposed. Secondly, a multiple description reconstruction network (MDRN) is proposed by L Zhao et al. [2], including edge reconstruction network (SRN) and central reconstruction network (CRN). When any of the two lossy descriptions is received at the decoder, the SRN network is used to improve the quality of the lossy description of the decoding by simultaneously removing the compression artifacts and upsampling. At the same time, if both lossy descriptions are available, they use CRN networks with two decoding descriptions as input for better reconstruction [4]. Thirdly, multi-description virtual coding and decoding network (MDVCN) is proposed to bridge the gap between MDGN network and MDRN network in order to train end-to-end MDC framework [5]. Here are two learning algorithms to train the whole framework. In addition to the loss function of structural similarity, the obtained description is used as the opposite label with multiple descriptions of distance loss function to regulate the training of MDGN network [7]. These loss assurances ensure that the generated descriptions are similar in structure but subtle in diversity. The experimental results show that a large number of objective and subjective measurements verify the effectiveness and flexibility of the method.

Deep neural networks represent a powerful class of function approximators that can learn to compress and reconstruct images. Existing image compression algorithms based on neural networks learn a quantized representation of constant space bit rate on each image. Although entropy coding introduces some spatial variations, traditional codes have benefited from a clear adjustment of the bit rate based on local image complexity and visual saliency. D Minnen et al. [3] introduced an algorithm that combines deep neural networks with quality-sensitive bitrate adaptation using a tiled network. The importance of spatial background prediction is demonstrated and improved quantitative (PSNR) and qualitative (subjective evaluator's assessment) results compared with non-adaptive baselines and recently published image compression models based on fully convolutional neural networks [6].

Streaming media refers to the media format played on the Internet by means of streaming transmission. Streaming media, also known as streaming media, means merchants use a video transport server to send programs as packets over the network. After the user unzips the data through the decompression device, the program will be displayed as it was before sending.

In order to meet the high-speed access needs of mobile devices and focus on resource scheduling, J Wang et al. established a local cloud system on the Cloud Access Network (C-RAN) of the cloud access layer and proposed a layered cloud service system. Taking “utility function” as the intermediate bridge between the two, we define the benefit function and cost function, and successfully establish the comprehensive utility optimization model. During operation, c-ran is responsible for the allocation of wireless resources, the local micro-cloud server is responsible for coordinating the service demand of local users for virtual resources, and the data center is responsible for predicting user demand and caching data. By optimizing the combined centralized scheduling and distributed allocation, the services in the local micro-cloud system can obtain an optimal network resource combination allocation scheme, improve resource utilization, and expand system capacity.

In response to the live streaming online system for cooperation on the Internet, Y Wang et al. proposed an efficient synchronization and synthesis scheme for multi-audio video streams based on RTMP (Real Time Messaging Protocol). First, define an additional time domain for each data flow. Then, multiple data streams are synchronized based on thresholds. Multiple audio streams were synthesized into one by adaptive weighted average method based on time domain, and video frames were combined into one image based on pixel field. Finally, the audio and video streams are reused as a device that can be pushed to a streaming server for live broadcasting. The scheme does not require a common synchronization clock and feedback mechanism, and has low complexity. Also overcome the network delay, jitter and other obstacles. Clock skew, real-time broadcast data loss. Experiments show that it can effectively meet the time requirements of synchronous, integrated and real multi-channel live streaming system.

The scheme of multi-audio video streams can receive a multimedia content stream consisting of a series of fragments, each of which corresponds to a corresponding predictive search location for multiple predicted search locations in the multimedia content stream. The likelihood of receiving a search request from a user moving from the current broadcast location in the multimedia content stream to one of the predicted search locations in the multimedia content stream can be determined. The buffer size of each fragment in the multimedia content stream can be determined based on the possibility of receiving a search request.

With the development of wireless communication technologies such as 4G and 5G, wireless network bandwidth can meet the transmission needs of more and more services [8], [9], [10], [11], [12]. With the popularity of wireless terminals such as mobile phones and tablet computers, it has brought about the rapid growth of mobile services. Among these mobile services, video applications account for the largest proportion and the fastest growth rate. Various kinds of video applications emerge in an endless stream. Compared with audio and image services, video services require higher quality and transmission rate of transmission channels, especially ultra-high definition video and VR video, so the transmission pressure can not be underestimated. Therefore, it is essential to optimize the transmission of video services. The video transmission process starts from the source server of the video provider, passes through the content distribution network (CDN) in the middle, and then transmits to the user terminal through the base station. This paper considers video transmission optimization from two aspects: video popularity prediction and LTE resource allocation. Firstly, this paper briefly introduces some commonly used classification models and multi-linear regression theory, and then based on the correlation between video lifetime and video long-term popularity, that is, the longer the lifetime of video often means that the longer the popularity is, the greater the popularity is. We introduce a new video lifetime into the popularity prediction model. Based on these factors, a multi-linear prediction model EPBLP_ML is proposed, which is based on the video historical viewing number, the surge State of future viewing number and the video lifetime. Our EPBLP_ML model is validated by video data on Youku video [13], [14], [15], [16]. The experimental results show that the prediction error of EPBLP_ML model is 1.58% lower than that of the latest model. Secondly, focusing on the topic of wireless resource scheduling, this paper first describes the architecture of LTE system, introduces the related factors affecting wireless resource scheduling and resource scheduling process in LTE, and describes the classical resource scheduling algorithm. Then, the importance of different frames in video is analyzed, and an IFF wireless resource scheduling algorithm for optimizing video I frame transmission is proposed. The simulation results show that compared with PF algorithm and M_LWDF algorithm, IFF algorithm can effectively reduce the transmission delay of I frame, reduce the packet loss rate of I frame to a certain extent, and improve the throughput of I frame. It also improves the spectrum efficiency and throughput of the system.

Online video services are now becoming popular because people are used to getting information in the form of video. While home users can easily retrieve a variety of content onto their laptops or smartphones, in-vehicle users are limited by the intermittent connection to the Road-Side Unit (RSU). Furthermore, as the load on the cellular infrastructure increases dramatically, it is contemplated that the cellular network can be offloaded by utilizing the RSU and the in-vehicle relay [17]. Y Sun et al. proposed a collaborative download mechanism in a heterogeneous in-vehicle network, including a Vehicle-mounted Ad hoc Network (VANET) and a cellular network. In this mechanism, the RSU acts as a traffic manager to obtain the appropriate data from the Internet and then distribute it to the vehicle in an approximately optimal manner. Specifically, based on vehicle mobility prediction and inter-node throughput estimation, a Storage Time Aggregation Graph (STAG) is constructed for planning a transmission scheme, and then an iterative greedy driving algorithm is designed for a suboptimal solution with polynomial computation time complexity [18], [19], [20], [21], [22]. Compared with the maximum throughput and Minimum Delay Cooperative Download (MMCD), the simulation results show that our method is better than MMCD 5–10% in terms of unloading score.

In order to obtain a better fine-grained video traffic classification result, L Yang et al. analyzed the relationship between feature change and video traffic classification in the transmission process. According to the properties that different types of video services contain different downlink transmission rate variation patterns, a new video stream feature - value probability distribution based on downlink byte rate variation, and video classification is realized through Support Vector Machine (SVM) and video classification is implemented by Support Vector Machine (SVM). The experimental results show that for the classification of six common network video applications, the probability distribution of M values is better than other common stream features.

TT NU et al proposed a content-aware public view multi-view video streaming video uploading scheme [23]. Their work considers relevant differential coding and multiple reference streams, monitored by packets to maintain better quality and reduce video traffic. Congestion events of video streams Many mobile cameras capture multi-view video to let the viewer experience different perspectives of the event. This stream of crowdsourced video is popular in many applications, such as entertainment, surveillance, and social sharing, for medical, educational, and military applications. However, due to the limited nature of wireless networks and large video traffic resources, video streams simultaneously limit the quality of video streams from multiple mobile cameras. Mehmood et al. proposed a content-aware video uploading scheme for crowdsourced multi-view video streaming [24], [25], [26]. Their work needs to take into account the relevant differential coding and multiple reference streams, and it is better to maintain and reduce the quality of video traffic through packet monitoring. First, the correlation between mobile cameras based on captured video content features is utilized. Secondly, the coding behavior of each camera is determined according to the degree of association between the cameras [27], [28], [29], [30]. Based on the behavior of the encoding, a differential encoding group eavesdropping camera transmission sequence can be implemented. The evaluation results show that in the strongly related mobile camera network, the traffic is reduced by 30% and the quality is improved by 2.5 db.

Road network is an important part of urban traffic network. The network with fractal characteristics and scale-free characteristics is often highly robust in the face of deliberate attacks against the network. Road network is an important part of urban traffic network. The network with fractal characteristics and scale-free characteristics is often highly robust in the face of deliberate attacks against the network. The fractal and scale-free characteristics of the road network in multiple scales indicate that the development and formation of the road network is a self-organizing process. In the process of urban planning and design, the structural characteristics of the road itself should be fully considered and its own laws should be respected. Therefore, this article further studies the transport of video.

Section snippets

Image compression based on spectral wavelet transform

Suppose ψ(t)L2(R), whose Fourier transform is ψ̂(ω¯), when ψ̂(ω) satisfies the allowable condition (complete reconstruction condition or identity resolution condition)Cψ=Rψ̂(ω)2ωdω<we call a ψ(t) basic wavelet or a mother wavelet. After generating function ψ(t) is stretched and translatedψa,b(t)=1aψ(t-ba)a,bR;a0call it a wavelet sequence. Where a is the scaling factor and b is the translation factor. The continuous wavelet transform for any function isWf(a,b)=<f,ψa,b>=a-1/2Rf(t)ψ(t-ba)dt

Data and preprocessing

With the rapid expansion of network bandwidth and the World Wide Web, multimedia content can be transmitted in large amounts and efficiently through a content distribution network based on the HTTP protocol. The advantage of applying HTTP to streaming media transmission is that the existing network architecture can effectively support HTTP transmission. For example, content distribution networks can provide local edge caches (EdgeCaches) to reduce long-distance transmission, and all firewall

Image compression effect evaluation

At present, the compression ratio or bit rate is usually used to measure the compression degree of the image, and the image quality evaluation is used to measure the compression effect of the image. therefore, in this article, two indicators of peak signal-to-noise ratio and compression ratio are mainly used to estimate the compression coding quality of image.

  • (1)

    Peak signal to noise ratio

The Peak Signal-to-Noise Ratio (PSNR) is a commonly used objective evaluation index for images, indicating the

Conclusions

As one of the most important artificial geographical entities on the surface, roads are closely related to the social and economic life of human beings. In the early days, the research on road network mainly focused on the geometric level, analyzing and modeling the geometric properties of the road network and the shape of the plane structure. The vigorous development of complex networks has also expanded the research direction of road networks. The study of abstracting road networks into

Conflict of interest

There is no conflict of interest.

Acknowledgement

This work was supported by NSFC (No. 61802114, 61802113), Scientific Research Foundation of the Higher Education Institutions of Henan Province(18A520021), Henan University Foundation (2016YBZR018).

Ge Zhang received the B.S. and M.S. degrees in computer science from the Zhongnan University of Economics and Law, Wuhan, China, in 2005 and 2007, respectively, and received the Ph.D. degree from Huazhong University of Science and Technology, Wuhan, China, in 2015. Now he is a assistant professor in School of Computer and Information Engineering, Henan University, Kaifeng, China. His research interests include image processing, streaming media, network modeling, and measurement.

References (30)

  • M. Kumar et al.

    Fuzzy theoretic approach to signals and systems: Static systems

    Inform. Sci.

    (2017)
  • W.P. Zhang et al.

    Analytical fuzzy approach to biological data analysis

    Saudi J. Biol. Sci.

    (2017)
  • E. Agustsson et al.

    Generative adversarial networks for extreme

    Learn. Image Compression

    (2018)
  • H.A.F. Almurib et al.

    Approximate DCT image compression using inexact computing

    IEEE Trans. Comput.

    (2018)
  • R. Torfason, F. Mentzer, E. Agustsson, et al. Towards image understanding from deep compression without decoding,...
  • L. Zhao et al.

    Multiple description convolutional neural networks for image compression

    IEEE Trans. Circ. Syst. Video Technol.

    (2018)
  • A.H.H. Alnuaimi, S.S. Aljuboori, R.J. Mohammed, Image compression using proposed enhanced run length encoding...
  • D. Minnen et al.
  • J. Begaint et al.

    Region-based prediction for image compression in the cloud

    IEEE Trans. Image Process.

    (2018)
  • Z.H.A.N. Wei et al.

    Research on image compression coding method based on wavelet transform

    Comput. Technol. Dev.

    (2018)
  • L. Chen et al.

    Comparison of image compression methods used in electronic shelf labels

    J. Suzhou Vocational Univ.

    (2018)
  • W.U. Cong-Rui et al.

    Design of SDRAM controller for image compression system

    Electronic Des. Eng.

    (2018)
  • R.J. Cheng

    High spectral image compression method based on segmental adjustable ship

    Ship Sci. Technol.

    (2018)
  • W.U. Chun-Ming et al.

    A image compression algorithm based on multi-node cooperation in wireless multimedia sensor networks

    Trans. Beijing Inst. Technol.

    (2018)
  • R. Herrero

    Reliable transmission of stream transported media in wireless real time communications

    Wirel. Netw.

    (2018)
  • Cited by (9)

    View all citing articles on Scopus

    Ge Zhang received the B.S. and M.S. degrees in computer science from the Zhongnan University of Economics and Law, Wuhan, China, in 2005 and 2007, respectively, and received the Ph.D. degree from Huazhong University of Science and Technology, Wuhan, China, in 2015. Now he is a assistant professor in School of Computer and Information Engineering, Henan University, Kaifeng, China. His research interests include image processing, streaming media, network modeling, and measurement.

    Jianlin Wang received the bachelor’s degree in mathematics from Jiangxi Normal University, in 2001, and Master’s and PhD degrees in computer science from East China Normal University, in 2006 and 2012 respectively. He is currently a lecturer in School of Computer and Information Engineering Henan University. His research focuses on symbolic computation, parallel computation and automated reasoning.

    Chaokun Yan was born in Kaifeng, Henan, P.R. China, in 1978. He received the PhD degree from Central South University in 2013. Now, he is an associate professor in School of Computer and Information Engineering, Henan University, Kaifeng, China. His research interests include data mining and bioinformatics.

    Sheng Wang received his Ph.D. degree at Nanjing University of Science and Technology (NUST) at April 2016. Now, he is an He is an assistant professor at Henan University. His research interests include pattern recognition, machine learning and medical image processing and analysis.

    This article is part of the Special Issue on TIUSM.

    View full text