Big Data Application in Intelligent Transportation

First, application background

Big data is the most popular technology concept recently. Especially after the Big Data Expo held in Guiyang, Big Data has become the next major development direction of the country and cities in the IT basic field. Regardless of whether the government, business or individuals are concerned about how to collect data, and how to extract useful information from the data, then create social value and commercial value. The Internet is the earliest industry in which Big Data has played its role. Taobao, Jingdong, Amazon and other e-commerce companies provide users with more specialized and personalized services through the mastery and analysis of massive amounts of data. Big Data is also reconstructing many traditional industries. It collects and collates data from all aspects of life, analyzes and mines, obtains valuable information from it, and derives new business models. Locations of flagship stores such as McDonald's, KFC, and Apple Inc. are based on accurate data analysis.

With the advancement of safe cities, smart cities, and industrial 4.0 projects and technologies, the Internet of Things big data has gradually become a new direction for the development of big data. Internet of Things (IoT) technology, which analyzes and mines new sources of big data, such as cameras, sensors, and industrial production data, has been flourishing. With the requirements of high-definition, intelligence, networking, and digitization, the security industry has rapidly increased its data volume. It has entered the threshold of big data several years ago. The “big data” in the field of security generally has several features: First, the amount of data is huge, and the video data of a prefecture-level city for 30 days is already at the PB level; second, it is different from the traditional data structure and the data structure of the security field is compared. Complex, more than 80% are unstructured data, such as video surveillance data in the construction of intelligent safe city, snap photos of bayonet, characteristic data of intelligent analysis output, etc.; again, the data update is fast, and video surveillance is performed every second. In the end, these more personalized data are required to be accessed randomly after storage, which requires the new Internet of Things big data system to handle data more quickly and more intelligently to save and manage data.

Second, big data technology development and innovation

Smart security based on big data focuses on big data processing technology and intelligent analysis technology in terms of technology. Big data processing technology mainly solves the problems of data collection, storage and analysis and mining. The intelligent analysis technology is to realize the analysis and understanding of video, solve the problem of video structuring, and promote the conversion of video data to video information.

Big data applications mainly involve data integration, data storage, mining applications, and many other aspects. Among them, big data technology is essential for the infrastructure protection of big data applications. The main contents include:

Big Data Acquisition and Management Technology

(1) Extensible data description specification

The rapid growth of data requires an extensible set of data description specifications to enable data description, data storage, sharing, and exchange. At this stage, the form of data mainly comes from the video data collected by the camera and other types of complex structure data. Designing an ontology description framework for multidimensional data can describe multidimensional semantic content more comprehensively.

(2) Multidimensional data integration common technology

Data extraction, transformation and loading (ETL) is an effective solution to solve the problem of heterogeneous multi-dimensional data consistency and integration. It uses tools to integrate data according to a unified rule, and completes the transformation of data from multiple data sources to a unified target database.

Big data storage technology

Large-scale sensing devices and numerous business systems generate large amounts of data every day. These data are structured, semi-structured, and unstructured, which brings difficulties to unified description and storage of data.

(1) Resource Description Metadata Management Technology

Resource description metadata is the basis for transparent access to heterogeneous data. By extending the existing data object and storage resource description methods, describe the intrinsic properties of the data (keywords, data encoding formats, etc.), application requirements (performance, availability, security, durability, etc.) and resource characteristics from various aspects ( Location, access methods, service capabilities, etc.) to support intelligently graded storage virtualization and storage services.

(2) Video data management technology based on space-time domain

According to the spatial-temporal domain attribute information of video data, spatially adjacent or temporally adjacent video data is classified according to requirements and stored on the storage, and at the same time, a deduplication technology based on semantic content is used to increase the value density of the data.

(3) Big Data Storage Technology

Ten billion or even billions of billions of structured data puts tremendous pressure on storage and queries. Traditional relational databases can no longer support such applications. The distributed data storage architecture and columnar database technology designed for mass data characteristics can better meet the requirements for scalability and high reliability of large data storage systems.

Big Data Retrieval and Mining Technology

(1) Distributed intelligent full-text search technology

Big data only relying on a single node to perform intelligent full-text search is far from meeting the performance requirements. The distributed multi-node parallel processing technology can effectively reduce the response time and improve the system performance.

(2) Search technology based on image recognition

There are massive amounts of image data in smart security. The current retrieval technology is still based on feature text description retrieval. This requires a lot of manpower and material resources to carry out feature description. When the data continues to grow, this will be an impossible task. The use of image recognition and fuzzy matching technology can truly meet the business needs of users and promote the application of face matching, gait matching, and behavior matching.

(3) Visual analysis of the associated network

Using visual analysis to graphically map all kinds of different information, establish common elements and links between different data sources and different information, and establish associations between different entities to discover related clues and intelligence hidden in big data. .

Third, the application of big data in the intelligent transportation field

In recent years, with the rapid development of the economy, the holding of motor vehicles has rapidly increased, and the contradiction between the status quo and demand for traffic management has intensified. Under this circumstance, how to use advanced scientific and technological means to improve the level of traffic management and suppress the occurrence of traffic accidents is an urgent problem for the current traffic management department.

For the needs of the traffic management department and the characteristics of China's roads, it is possible to implement real-time monitoring and data collection of all-weather real-time traffic lanes and non-motorized lanes on the monitored road sections through the integration of image processing and pattern recognition techniques. The front-end bayonet processing system analyzes the captured images to obtain data such as number plate number, number plate color, car body color, vehicle logo, vehicle sub-brand, etc., and obtains the vehicle information together with the passing time, location and driving direction of the vehicle. Such information is transmitted to the database of the bayonet system control center through the computer network for data storage, inquiries, comparisons, etc. When an accident, escape, violation, or suspicious vehicle is found, the system will automatically send an alarm signal to the interception system and related personnel. It provides important information and evidence for the timely detection of violations such as traffic violation investigation, traffic accident escape, and robbery of motor vehicles. At the same time, with the construction of the city's Smart system, the new Smart IPC monitoring front-end will also become a bayonet system, which will make the city bayonet system more rigorous, and it will be able to obtain more passing data and can more accurately depict Vehicle dynamics information.

The front-end bayonet system can also record target information passing through the bayonet in a timely and accurate manner, grasping the traffic status of vehicles entering and leaving the district at any time, and providing important reference data for traffic induction. In order to solve the challenge of mass data analysis of passing cars, intelligent transportation should change its thinking and actively try to use big data technology to solve the problem of intelligent traffic data analysis and mining.

In cities in a developed economic area, traffic accidents have occurred frequently in recent times, and the vicious traffic incidents caused by lost drivers and drug drivers have had a bad influence on the people. The municipal party committee leaders decided to carry out a traffic improvement operation, focusing on the remediation of key traffic violations such as “yellow-standard vehicles, overdue unspent vehicles, illegal vehicles, illegal drivers, and blocked license plates”. In addition to increasing the intensity of patrolling the road, the traffic police department is also required to monitor and dig traffic violations through traffic big data.

The major source of traffic big data comes from the passing records of the bayonet on the traffic road. However, the bayonet coverage is limited. Can we obtain traffic data from social resources? Through coordination, the traffic police department has obtained surveillance videos from key traffic areas such as parking entrances, gas stations, and public parking spaces. Through the massive video cloud analysis platform, the vehicle-related data in the surveillance video will be taken out, including the license plate, vehicle models, and driver characteristics, and deposited into the big data platform.

****Analysis compares the two schemes. For local vehicles, if the models matching the vehicle's inventory in the same license plate record are inconsistent, they may be *** and can be extracted for further analysis. For foreign vehicles, a space-time analysis scheme is used. If the same license plate appears in two places far away from each other in the same time period, it means that it is *** and it is extracted for further analysis.

* Poisoned and lost drivers are not allowed to drive motor vehicles on the road, otherwise it will cause great traffic hazards. The traffic police department passed the existing key personnel pool of poisoned and lost drivers. When the bayonet is deployed with a high-definition bayonet, the driver's head photo can be taken out. Automatically compare faces in photos of drivers and drivers who have missed drivers. If there is a high degree of similarity, it may indicate that they are drivers and drivers who have lost their lives and are extracted and analyzed.

* By querying the vehicle with missing license plate in the passing record of the bayonet, you can query the vehicle with blocked license plate. However, it is not enough to find that the license plate is blocked by the vehicle. It is also necessary to find the real license plate of the car. Therefore, it is necessary to use the search map function of traffic big data to search through all the passing records through the characteristics of the obscured license plate vehicle to find the vehicle with the highest similarity and to extract it for further analysis.

* Whether the vehicle in the recorded car record is a yellow-labeled vehicle or an overdue vehicle that has not expired. If it belongs, it is extracted for further processing.

As a result of careful investigation of suspected vehicles that have been analyzed through the traffic big data platform through several means mentioned above, and confirming that they are indeed illegal, the vehicle trajectory analysis function of the traffic big data platform is used to acquire the suspicion of vehicle activity, set cards in advance, intercept and handle them.

The rectification action analyzed and extracted more than 3 billion records of passing cars, analyzed social videos for 300,000 hours, and extracted data from more than 70 subsystems. During the two-month rectification period, more than a dozen drivers and drivers lost their lives through the traffic big data platform. The number of *** vehicles exceeded 80, and more than a dozen cars were not scrapped. The relevant leaders are very satisfied with the performance of the traffic big data platform in this rectification campaign, and stated that “this rectification operation has effectively and efficiently fulfilled the predetermined target with the help of the traffic big data platform”.

Fourth, summary

There are many possibilities for smart traffic based on big data, and the intelligence of traffic is the fundamental trend. Using big data technology and intelligent analysis technology to integrate other data of urban management will really promote the construction of smart transportation and lay a good foundation for traffic management. . At present, big data technologies are mainly applied to the traffic police and traffic management departments. With the further networking of traffic data, integrating data from various sources such as parking lots, railways, rail transit, and public transportation will provide more abundant cities. Transportation applications will make the roads open, and parking spaces will no longer be difficult to find, improving the overall operational efficiency of the city.