Research on Video Monitoring Application of Intelligent Station

Liu Yi ,Zhao Hui

(1.Nanchang Railway Bureau ExecutiveLeadership Academy,Information TeachingandResearchSection,Nanchang,Jiangxi,330002;2.Information&Network Center,Central South University,Changsha,Hunan,410083)

Abstract:The paper first expounds the basic situation and application background of the monitoring arrangement of the intelligent station and adopts the relevant security monitoring means to give the main monitoring layout system applicable to the station,including some important technologies and overall ideas to study the station application solution and propose existing equipment problems which have theoretical value and application prospects.

Keywords:intelligent video surveillance;station passenger transportation security;software and hardware system framework

Introduction

The railway occupies an important position in China’s transportation system.It has always been a leader in the transportation system and its passenger and cargo transportation far exceeds other modes of transportation.With the leaping development of China’s railway construction in recent years,ensuring that railways do not have safety problems has become the most important part of ensuring the normal operation of the railway system.The railway departments at all levels have taken numerous measures to prevent the occurrence of railway accidents.The role of video surveillance systems in transportation,security and other aspects is becoming more and more obvious.These monitoring systems play akey rolein handlingtrain station accidentsand restoring the normal order of rail transport after security problems have occurred.At the same time,however,how to manage the massive amount of video data that is continuously acquired,how to automatically extract the information and knowledge that people want from them,is an urgent problem.The existing monitoring layout structure does not meet the requirements of station monitoring and management,so it is necessary to focus on the development of new monitoring technology,establish an effective operation and maintenance system and ensure that video surveillance systems can effectively prevent and supervise in railway transportation production.

I.Background of the Project

In the 21st century,many railway stations have already established independent video surveillance systems.Their video surveillance systems are mainly used in station waiting halls,station plazas,station exits,entrances,for human traffic statistics,real-time monitoring of the surrounding environment of the station,etc.It has greatly impacted the traditional concept of safety of the world.At present,the ground transportation point monitoring of transport passengers has been fully covered. The comprehensive monitoring system mainly includes real-time monitoring of the areas where passengers gather in a waiting area,ticketing area,train or bus station in order to grasp the location of passengers and their behaviors.Thereby,the station’s ability to respond to unexpected situations can be improved and the safety of transportation process can be ensured.

The increase in efficiency is accompanied by an exponential increase in video file size.At the same time,in order to analyze and process these data,it takes a lot of manpower and energy to search and analyze.These have led to an ever-increasing workforce for related work.However,this kind of video surveillance that requires human participation,because of the human physiological weakness,makes the accuracy not high and and analysis errors and false reports caused by data analysis problems often occur.This also makes the monitoring system no longer relatively cost-effectiveand practical.

At the same time,due to the relatively complicated events in the station,the workload of monitoring and management is too much and it is impossible to train more professionals for special management in a short time.This makes the station more complicated and difficult to handle the warning of abnormal events.But this cannot negate the importance of video surveillance.In the case of mass processing events and incident tracing,a large amount of data is difficult to retrieve and it also makesthework of staff moredifficult.

Based on the above problems,the key monitoring areas are divided by modern intelligent application systems and event warning and scene analysis are performed on the events in the area.In addition,a more comprehensive decision tree for data mining should be established in the post-indexing process.It fundamentally changestheworkflow of the staff,frees them from the original complicated mechanical work and gives them more opportunities for event resolution and handling responsibility.[1]The traditional manual detection is transferred to the machine through the application of machine learning and recognition algorithms.It not only ensures uninterrupted and efficient operation for 24 hours a day,but also eliminatesthe problemof unstable video surveillance caused by subjective factors.At the same time,video-based processing is also beneficial to give the video more exclusive attributes and it is faster and more convenient to reproduce the problem scene when the video issearched and retrieved.

II.Application Requirements

The main function of the intelligent station is to achieve the following aspects in management operations:

(i)The intelligent station has a large space area and each panel function has an independent space.The entire space is divided into multiple areas according to functions and multiple service sections such as waiting,ticketing,office,and business consultingareseparately performed.

(ii)It is necessary to realize information sharing and data docking between stations and different public administration services and become an important part of intelligent city.

(iii)It isvital to efficiently manage the operation of the station and provide services by means of automation technology to automatically change business operations of the station during the actual operation.It not only improves the efficiency but also reduces the error rate generated by manual services,reduces the number of manual jobs and reduces the output cost of personnel.

(iv)Integrated management platform and service platform with higher user experience should be set such as baggage storage and mailing platform,intelligent guidance mode for entering and leaving the station,combined with video service voice reminder and other servicefunctions.

In the monitoring system used at the current station,it is mainly used to set up monitoring equipment within a certain range to meet the monitoring dead angle.After the deployment is complete,the video data is sent to the integrated management display system. There is a specially-assigned person to control the monitor screen on duty.Real-time observation of the operational status of each node and the flow of people in each service hall.When an emergency occurs,it can be pre-processed in time and it can also support the recurrence of illegal behavior and accident conditions.

As much as possible to dig out the effective fragments in the video data,from the perspective of the majority of passengers’own security to improve operational management efficiency.The use of new intelligent detection methods based on traditional monitoring methods is extremely effective.The artificial intelligence technology based on machine learning and neural network is used to image the video data in the monitoring system and timely detect,for example,the character image sound information and the customized service detection mechanism and timely process the emergencies.It should comprehensively analyze the entire scene,detect real-time occurrence,alert the emergency situation at the first time and provide the most reasonable solution.At the same time,the cause and effect of the incident are recorded to facilitate the division of responsibility afterwards. This combination of intelligent detection technology and manual processing is the core function of the intelligent monitoringsystem[2].

The detection functions supported by intelligent video analysis mainly include cross-line,intrusion,reverse driving,entry/exit,remnant/object moving,occlusion, direction, speed, fireworks, motion detection,automatic tracking,traffic statistics,target density,and soon.

It is important to analyze the demand standard of intelligent station for video processing and the development trend and function of current intelligent detection technology and combine the two to design the core functions of video surveillance system upgrading in the following areas.

Waiting area:it is vital to realize the real-time monitoring of the waiting area,analyze whether there is passenger gathering problem in the waiting area and realize intelligent guidance.At the same time,it is also necessary to implement a matching detection technology for passenger items.When an abnormality occurs,the systemwill give an alarm.

Ticketingarea:it needsto monitor and speculate the number of people gathered in the ticketing window and analyze the average processing speed and waiting time of each window to provide more choices for ticket buyers.

Office area:the face recognition technology is used to analyze the staff in the office area,strictly control the occurrence of personnel changes and issue alarms and reminders to the sudden abnormal situations,conducting behavior analysis and testing torecord violations.

Station area:in order to ensure the safety of driving,the items dropped in the train track will be alarmed in time.The personnel composition inside and outside the warning line is tested in real time to guarantee the personal safety of passengers and the legacy items are analyzed to ensure that there will be noabnormal incidents.

Equipment room:in order to minimize the loss of equipment,it is important to regularly perform fireworks detection on important equipment such as UPS room,centralized air-conditioning room and strong electric room,monitor whether there is illegal entry and exit in the equipment room and immediately report thealarm.

III.Intelligent Station Video Surveillance System Functions

1.Front-end Information Acquisition Camera

The installation of video capture devices requiresacomprehensiveconsideration of thevarious requirements of the design features. It comprehensively analyzes of needs of public security policing,vehicle logistics and freight supply and power supply departments,designed to meet the requirementsof comprehensivescheduling.

In the data collection of railway monitoring system,the main application scenarios include indoor and outdoor.The indoor environment includes the passenger hall and server management room.The outdoor environment includes bridges,ramps,tunnels,for railway trains.Outdoor scenes often have poor environmental conditions and are often affected by abnormal environmental conditions. When considering the construction of outdoor monitoring systems,it isoften necessary to take measures against weather,snow,lightning,thundering and earthquakes to prevent external interference.This increases the difficulty factor in designing and implementing an outdoor monitoring system.How to ensure the stability of the system and how to reduce the maintenance difficulty of the equipment must be considered.At the same time,in some scenarios,the monitoring equipment should have functions such as remote management and cloud group upgrade,which has higher requirements for the cost of hardware and software.In general,the following features must be available for front-end cameras:

(i)It must guarantee normal use under night conditions.The camera should have a thermal infrared function and the dual filters of different day and night can be used alternately,so that the quality of the video can be guaranteed in low brightness and night environment.

(ii)Long-distance high-definition shooting can be performed on long-distance trains in the wilderness to use the long-distance zoom function.It is the main method to ensure image quality on the extended railway.

(iii)It can precisely locate important small targets.Thevideocontrol device isrequired to have a jitter filtering function after long-distance zooming and a multi-step rotation adjustment method is used for microstep adjustment.

(iv)It owns the function that can dynamically adjust the video width for reference to changes in the field environment.When there is a need to perform backlight compensation or increase the video range change,it can be realized by the technology of automatically changing electronic shutter.

(v)The alarm linkage technology needs to have enough reserved settings to make calls.The accuracy of the reserved set is the measure that determines the PTZcamera.

(vi)It is necessary to use the noise reduction technology at the right time to compress the file size during video storage and minimize the occupied space while ensuring a certain coderate.

2.System Storage

The railway has the characteristics of large distance base when performing data transmission,so the requirements for network resources will be relatively high.In order to encode the video,it is necessary to select a camera device with a higher hardware coding capability.The specific feature is to ensure that the performance parameters of the encoder meet the conditions,including the control of video transmission delay and the design of video and audio support formats and input and output standards.

H.264 and MPEG-4 technologies are used for video file transmission and storage.The video data is spatially and frequency-domain-converted during encoding and the content of each frame of the video is compared in the form of quantization and prediction to remove redundant information.At the same time,the selected camera equipment and the entire internet management system should support the management of relay equipment carried on the outside of the equipment,realizing the linkage alarm function or the full-duplex voice intercom function.Protocol support for remote upgrades and remote transfer of device data is also required for ease of deployment and post-maintenance.When performing monitoring operations,it is often necessary to maintain a long-term admission connection action,which optimizes the video storage and retrieval functions and performance.Filtered searches for multiple search parameters are supported such as location time and events.Online playback and remote local download operationsare also required.

A complete and flexible video storage mode needs to be established and the content of the video for the self-judgment event trigger save,timed video save and manual video save needs to be comprehensively considered, operating current mainstream video data storage architectures and redundant processing.The recording format cycle of different cycles is divided to support the design planning and online deployment mode developed by theproject requirements.

The specific functions are as follows:

(i)Multiple storage methods can coexist and it supportsmanual and automatic trigger mode.

(ii)It supports manual processing of video information storage.

(iii)Multiple archival storage modescan coexist.

(iv)It supports multiple conditional retrieval,playbacks condition settings and meets the demand for time space.

(v)It supports local download and remote return visit.

(vi)It supports multi-user parallel operating systems.

3.Platform Software in Railway Integrated Video Surveillance System

The core data service platform and the client data processing station are integral parts of the video surveillance system.The core service platform can read the core data of the server from the core database and the client processing station is used to configure user’s information and manage device information.For example,it is used to assign user rights and manage device operation logs and so on.The daily monitoring and maintenance of the parameters such as the performance and status of the video equipment should be supported in the monitoring system.At the same time,it is also necessary to provide users with functions of registration,authority authentication,and additional data clearing and rights management for the administrator.The log management function can provide the function to satisfy the administrator’s query for the user’s login and logout operations.The specific function[3]isas follows:

(i)Monitoring function:it satisfies the unified management call center of the monitoring equipment integrated in all areas and realizes the read control requirementsof the control equipment data.

(ii)Voice function:it realizes full-duplex language interaction or single-item audio stream broadcasting function that meets the protocol at the management network and the monitoring device that supportsthe voice service.

(iii)Data integration:A data integration service based on Internet of Things device information exchange protocol is implemented on the operating monitoring device to realize an official database that can storemonitoringinformation of theentireregion.

(iv) Network management: it implements network management for monitoring devices and implements data synchronization between the device and themanagement center.

(v)Alarm management:it records and displays alarms of devices,systems and communication parts and recordstheprocessingstatusof alarms.

(vi)Device Control:it realizes the data read and write function of the sensor or controller under the monitor allowed by thepermission.

(vii)Data Storage:it sets up a unified data storage center for devices connected to the network center to maintain data backup requirements for videoor other parameters.

IV.Key Technological Issues

In the monitoring system of railway stations,with the spread of intelligent detection technology,there are new requirements for more accurate retrieval technology.How to detect the occurrence of violations,how to deal with massive video surveillance data for quick browsing and analysis,is the key point of current detection technology[4].In the anomaly detection and video summary,the key issuesare asfollows:

(i)In a crowded scenario,how to extract an effective feature training event model and perform anomalous event detection is a problem.In a crowded scene, abnormal event detection becomes a challenging task due to complex background and severe occlusion.Plenty of existing work is based on the characteristics of motion trajectory,optical flow or space-time gradient and the event model is trained by statistical methods to perform anomaly detection.However,in acrowded scene,trackingisdifficult and information such as optical flow also contains a lot of noise.In order to improve detection performance,training event models is a critical issue if extracting features that can better reflect group behavior in crowded scenarios.

(ii)How to generate a video summary that is easier for people to understand and has a higher compression ratio is another problem.Video Concentration is a motion-based video summary developed in recent years that generates a high-compression video summary that preserves the dynamicsof the target in the original video.However,existing video condensing algorithms only eliminate time and space redundancy without eliminating redundancy on the content.In the video enrichment process,too many motion observation targets will reduce the video concentration efficiency and affect the concentrated video visual effects[5].Therefore,how to eliminate redundancy from time,space and content in video enrichment is a problem worth studying.With the development of intelligent monitoring,how to monitor multiple cameras in large scenes is particularly important.To display the dynamic data that appears completely,the video files with larger specifications must be condensed,which is easy to retrieve and browse and has become more and more valuable and needed.Matching moving target trajectories is a key point and it worths further study.

V.Related Improvement Suggestions

(i)System robustness must be improved.It is necessary tooptimizethe algorithmic model of motion detection and background description in video processing for complex situation changes in different environments such as human sloshing effects or lighting conditions.External disturbances caused by shadows and weather or surface reflections will be minimized during calculations.

(ii) Video data should be processed comprehensively.It is necessary to establish a video data interaction platform,collect multiple camera data in the same scene and use a comprehensive algorithm for joint computing.The 3D integrated scene recovery function can solve the problem of the external environment occlusion and also display the scene moreintuitively.

(iii)The accuracy of monitoring must be guaranteed.Different alarm rules are set according to different timeperiodsof different scenes.Thespecific needs are specifically designed to avoid generating falsealarmmessageswhen they arenot needed.

(v)The alarm accuracy must be improved.Relying solely on image processing technologies to generate alerts for violations is not sufficient to meet the accuracy requirements of alert.Therefore,the combination of manual alarm and automatic alarm is necessary.It can not only use the shorter time in actual use,reduce manpower consumption,but also improve the accuracy of the alarm and meet the actual requirements.

VI.Conclusion

The technology of intelligent video surveillance is gradually maturing.In the field of practical application and deployment,there are still some functions that are not perfect for the application of monitoring technology,including video stream information fusion technology and large-scale video integration systemdesign.

When passengers enter the service platform,it should effectively provide video warning information to meet the needs of passengers so as to improve the service quality of the service platform more efficiently.The combination of humanized sound and image also need to be added to the video surveillance system.

At present,the video processing on the market is mostly based on a single processing system.This low-integration processing platform is designed at a high cost while not providing cost-effective system processingefficiency.Therefore,in thefuturemarket,relying on the development of network computing processing technology,the application technology that promotes large-scale integrated processing video will becomeanew mainstream.