Video Image Perception Technology in Smart Cities

    Wuxi is one of the first batch of national smart city pilot cities. Through high-standard network infrastructure construction, coordinated construction of urban public information service platform and urban public basic database, it has gradually completed the demonstration application of smart city technology in urban construction, real estate, water, gas and lighting, pipelines, transportation, home and community. In recent years, with the support and promotion of relevant policies, Wuxi's smart city construction has made outstanding progress in top-level design, overall planning and demonstration application. In particular, in the basic fields of smart government affairs, smart transportation, smart medical care, etc., it has achieved leading results in the country. In 2013, it won the first place in the domestic smart city development assessment. On July 30 this year, Wuxi stood out from hundreds of candidate cities around the world and became the only Chinese city selected for the "IEEE Smart City Pilot Program". Wuxi's leading position in the construction of smart cities in the country is evident. This article mainly discusses the current situation of Wuxi's urban construction from the perspective of smart video perception system in smart city construction.

　　Intelligent Video Perception

　　Smart cities cannot be separated from mature infrastructure construction. Wuxi started to build infrastructure relatively early, such as basic network facilities, wireless network systems, urban information data systems, etc. Among them, the construction and application of urban audio and video information systems is also a very important one. From the perspective of the components of smart cities, the smart video perception application system based on video surveillance technology and Internet of Things technology is an important part of smart city construction, and is also the core support platform for urban management public services and emergency command platforms. It is also a key part of the entire smart city perception system and an important source of information.

　　The construction of the smart video perception application system further realizes computer intelligent management, making the layout of transportation, security, environmental protection, etc. more reasonable. The system makes the comprehensive management of the city orderly, effectively enhances the control ability of key parts of the city, is conducive to timely discovering various situations, and serves the leadership's command and decision-making.

　　At the same time, the construction of the system also makes prevention work safer, more thoughtful and more convenient. Through a strict audio and video perception monitoring network, across geographical time and space, real-time tracking of each monitoring scene, greatly improving the ability to respond quickly, more effectively combating crime, improving the security level and comprehensive management level of Wuxi City, and enhancing the confidence of foreign investors in capital input to Wuxi City, which is conducive to the rapid development of the economy of all walks of life in Wuxi.

　　Secondly, based on the Wuxi Internet of Things platform, under the premise of unified information exchange, based on the city alarm and video surveillance system, urban intelligent traffic management, government portal, digital city management, and urban emergency command system that have been built by multiple government business departments, the city's sensor information resources are integrated and gathered to achieve alarm linkage for multiple business departments and multiple business systems, effectively avoiding inconsistent data and information island effects in the business system, and also avoiding the waste of funds and energy caused by single-line contact and exclusive information when solving the same problem in various departments, which will have a value-added effect on Wuxi City and generate huge added value. By realizing resource sharing, it can provide information support for more government functional departments, thereby giving the system new functions and roles, and effectively improving the comprehensive ability to solve urban security.

　　From a national perspective, after nearly 10 years of video surveillance system construction and networking integration, the infrastructure of video perception has been further improved. The goal of the current stage and future construction is to integrate and expand the entire audio and video resources on the basis of fully integrating the existing video resources of the public security, to achieve the networking integration of all social resources, and to make them intelligent for smart city management. The goal of smart city audio and video applications is to give full play to the characteristics of high audio and video information density, to extract key information from video data efficiently and accurately, and to describe it in the form of information that can be recognized by both machines and humans. In fact, it requires the audio and video perception system of smart cities to have the ability of intelligent analysis, which is equivalent to the human eyes and brain, and to give full play to the role of urban visual senses. This goal is quite grand and arduous, and must rely on more advanced technical means and system architecture to achieve it. From the current situation, thanks to the development of current information technology and the progress of audio and video processing technology, we can rely on cloud computing, big data and video intelligent analysis technology to provide high-tech means to gradually move towards this goal.

　　Smart cities must rely on intelligent analysis and prediction capabilities based on video big data to give full play to the role of urban video sensory systems. In order to accurately and efficiently process large amounts of audio and video data, the system needs to have high-performance, high-accuracy intelligent video analysis algorithms on the one hand, and a system architecture that can support the large-scale application of these analysis algorithms on the other hand, such as cloud computing systems, big data analysis systems, etc.

　　Cloud computing architecture technology for video image systems

　　Cloud computing is the product of the integration of traditional computer technologies and network technologies, such as parallel computing, distributed computing, network storage, and virtualization. Cloud computing uses the network to build and share various types of computing physical and business resources on a large scale, providing users with a new service model for on-demand use. Users can directly obtain a variety of flexible IT and business services through the network, including computing services, storage services, and other application services such as video applications.

　　Cloud computing can meet the sustainable development needs of users in the process of informatization. It has the characteristics of reducing costs, improving IT resource utilization, convenient capacity expansion, and flexible use. Its core concept is to concentrate resources and effectively manage and allocate them. By continuously improving the processing power of the "cloud" and reducing the processing burden of user terminals, users can enjoy the powerful computing and processing capabilities of the "cloud" in a self-service manner on demand. The resources in the "cloud" are infinitely expandable in the eyes of users, and can be obtained at any time, used and expanded on demand. Cloud computing is provided to individual and corporate users in the form of services.

　　The smart video perception platform will be designed based on the cloud platform architecture solution, and various image and video business applications will be built on the cloud platform infrastructure. According to the general technical solution for cloud computing platform design and deployment, for the safe city video surveillance cloud platform, the overall deployment architecture is as shown in the following figure, see Figure 1.

　　Video surveillance cloud platform architecture

　　As can be seen from Figure 1, corresponding to the service mode of cloud computing, the video surveillance cloud platform also provides three levels of services for video surveillance business.

　　At the IaaS level: The system provides management of physical devices and resources, a unified virtualization mechanism for physical devices, and management and maintenance of related virtual resources, such as virtual machine management services, unified cloud storage management services, cloud resource monitoring, and other video infrastructure services.

　　At the PaaS level: This level is a platform that provides the most basic services and capability support for video surveillance services, including streaming media services, video storage services, video gateway access services, video summary services, video intelligent analysis services, big data analysis services, operation and maintenance management services and other video surveillance support services.

　　At the SaaS level: This level is oriented towards end users, providing a rich video surveillance-related business usage environment, such as public security monitoring patrols, command and dispatch, image detection, intelligent transportation, such as urban emergency command, safe cities, smart communities, smart tourism and other application business services.

　　Big data processing technology for video image system

　　Big data and smart cities are related. In terms of people's livelihood and government decision-making assistance, they need to be captured, mined and responded to quickly, accurately and efficiently in smart cities. In fact, there are already many big data application scenarios in smart cities. For smart video perception, it is also inseparable from the application of big data. At present, the amount of data in video surveillance systems is constantly expanding. The emergence of high-definition video, the continuous advancement of video surveillance system networking and integration, and the improvement of storage technology and capacity have led to huge data volumes, and the characteristics of big data are prominent. The construction goal of large-scale video surveillance systems is to meet the needs of public security management, urban management, traffic management, emergency command, etc., and often take into account the needs of disaster accident warning, production safety monitoring, etc. for image monitoring, and consider the integration of supporting systems such as alarm and access control and the linkage with the broadcasting system. Such a video surveillance system is destined to be a large data collection, and its accurate analysis and efficient use are crucial.

　　In addition, in video image monitoring applications, many types of data information are involved. From the perspective of data structure type, it includes various types of unstructured, structured and semi-structured information.

　　Unstructured data mainly includes video recordings and picture records, such as surveillance video recordings, alarm recordings, summary recordings, vehicle checkpoint pictures, face capture pictures, alarm capture pictures, etc.

　　Structured data includes alarm records, system log records, operation and maintenance data records, summary analysis structured description records, and various related information databases, such as population information, geographic data information, vehicle and driver management information, etc.

　　Semi-structured data includes face modeling data, fingerprint records, etc.

　　All these data as a whole, from an industry perspective, constitute the big data foundation of video surveillance systems or video image surveillance. In smart cities, the use of big data analysis and processing technology can improve the analysis and processing capabilities of urban audio and video information and provide more accurate and efficient visual sensory applications.

　　With the emergence of large data storage systems such as cloud storage, the amount of data available for analysis is larger, and the analysis results and predicted information based on this more information will be more accurate. In addition, with the application of high-definition technology and the rapid development of IT architecture and analysis technology, relying on big data analysis technology, valuable information can be extracted from a large amount of unstructured video data, improving the ability to discover and utilize data information, thus making "early warning" a reality. The big data processing technology in the video surveillance industry, especially in real-time intelligent analysis and data mining, has gradually enabled video surveillance to progress from manual sampling to efficient early warning and post-event analysis, and further realize the intelligent information analysis and prediction of the system. Through the accurate extraction of video image information and the integration of other video surveillance and information systems, for an information point, more relevant information can be associated and comprehensively presented, forming a variety of effective technical and tactical applications, and enhancing the value of video surveillance information.

　　Video intelligent analysis technology

　　Video intelligent analysis technology is extremely important for smart video perception in smart cities. It is the basis for realizing system intelligence, provides the system with the ability to identify image data information, and completes the conversion of photoelectric data signals into information.

　　This technology uses computer image vision analysis technology to separate the background and target in the scene, and then analyze and track the target of interest that appears in the camera scene. At the same time, through computer modeling and other technologies, it can identify the type, color, characteristics, speed, size and other related details of the target, thereby extracting the information data contained in the image video, supporting rich intelligent video analysis applications and comprehensive analysis applications of big data.

　　On the one hand, the intelligent video perception system can extract important information from video images, store and associate it with business, and provide a data basis for big data applications; on the other hand, it can also realize intelligent automatic monitoring and alarm functions based on the detected information. By presetting different alarm rules in the scenes of different cameras, once the target is detected to have violated the predefined rules in the scene, the system will automatically issue an alarm, and the monitoring workstation will automatically pop up the alarm information and issue a warning sound. Users can click on the alarm information to realize the alarm scene reorganization and take relevant measures. For smart cities, this application allows the effectiveness of visual senses to be reflected most directly.

　　In fact, with the continuous improvement of video analysis algorithms, the accuracy of the algorithms is constantly improving, the scope of application is gradually expanding, and the market demand for such applications is growing. The main function of these applications is to assist the security departments of the government or other institutions to improve the security protection of large outdoor public environments. Such applications mainly include: advanced video motion detection, intrusion detection, object tracking, abandoned object detection, removed object detection, illegal detention, etc. In addition, there are also a large number of applications for urban traffic management, environmental management, government services, retail services and other industries: crowd counting, congestion detection, tailgating prevention, traffic flow control, vehicle control, illegal parking detection and alarm.

　　The connection and integration of various technologies

　　The cloud computing system architecture technology, big data analysis technology, and intelligent analysis technology mentioned above are core technology systems related to smart video perception in smart cities. They are closely related and are not built in isolation. They need to be designed and built as a complete system.

　　Cloud computing architecture technology provides the overall architecture of the video perception system. Only through this architecture can we fully utilize and flexibly schedule computing, network and storage resources, and realize efficient distributed computing and efficient audio and video storage. Based on this architecture, big data analysis technology can play its most efficient role. Distributed computing and efficient storage can ensure that the big data analysis process is faster, the amount of data analyzed is larger, and the analysis results are more accurate. Video intelligent analysis technology can also be greatly assisted by cloud computing architecture. Cloud computing supports high-performance distributed computing, which can ensure that intelligent video analysis is faster in parallel. It can distribute and execute intelligent algorithms according to the data distribution characteristics of cloud storage. For video clips of the same length, the analysis results can be output faster. The high-performance and high-elasticity computing power provided by the cloud computing system also enables sufficient resources to be called in the video intelligent analysis process, improving the integrity and accuracy of the analysis results.

　　Video big data analysis technology is the data analysis foundation of business applications. The audio and video resources of the city provide a huge amount of data information. In order to form effective information materials and then achieve trend prediction results, it is necessary to conduct efficient correlation collision and comprehensive analysis on these data. For the massive video data that is updated very quickly, the traditional data organization and management architecture analysis process can no longer effectively cope with it. It is necessary to use big data analysis technology to support efficient data analysis process, ensure the timeliness and integrity of information processing, and ultimately provide more accurate results. Without big data analysis technology, a large amount of valuable and implicit information will not be fully utilized, and the smart video perception ability of smart cities will not be reflected.

　　As for video intelligent analysis technology, on the one hand, it uses the computing power of the cloud computing platform to provide analysis and recognition of video images, improving the system's ability to respond to real-time conditions; on the other hand, by analyzing video images, it extracts key information data that can be recognized by computers or humans, accumulating original and identifiable information foundations for the big data application of the smart video application system, so that big data applications can collide, analyze and utilize the results of these analyses at a higher level. In the stage of video research and judgment, the combination of video intelligent analysis functions and big data analysis technology can ensure that the data used for analysis and judgment is richer, and the final research and judgment results are more accurate.

　　Conclusion

　　Smart safe cities are the current development theme and the advanced stage of urban informatization. Smart video perception is an extremely important part of smart city construction, providing smart cities with visual perception capabilities, video recognition, analysis and processing capabilities, and reliable and intuitive management decision-making information. In the fields of cloud computing, big data, and video intelligent analysis, Keda explores the frontier of the industry to make full use of video information technology to give things intelligence, and further provide smart cities with intelligent information technology.

　　Develop the power of technology. Based on Wuxi's current smart city construction, Keda will provide technological support for Wuxi's urban governance and environmental construction.