CN108583592A - A kind of subway service on buses or trains job information acquisition intelligent detecting method - Google Patents

A kind of subway service on buses or trains job information acquisition intelligent detecting method Download PDF

Info

Publication number
CN108583592A
CN108583592A CN201711490984.6A CN201711490984A CN108583592A CN 108583592 A CN108583592 A CN 108583592A CN 201711490984 A CN201711490984 A CN 201711490984A CN 108583592 A CN108583592 A CN 108583592A
Authority
CN
China
Prior art keywords
video
driver
audio
analysis
acquisition
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201711490984.6A
Other languages
Chinese (zh)
Inventor
刘炜
葛党朝
禹建伟
段秋广
刘志刚
张小宏
贾练虎
刘维维
刘通
杨龙
孙权
曹薇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xi'an Sub LLC
Original Assignee
Xi'an Sub LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xi'an Sub LLC filed Critical Xi'an Sub LLC
Priority to CN201711490984.6A priority Critical patent/CN108583592A/en
Publication of CN108583592A publication Critical patent/CN108583592A/en
Pending legal-status Critical Current

Links

Classifications

    • BPERFORMING OPERATIONS; TRANSPORTING
    • B61RAILWAYS
    • B61CLOCOMOTIVES; MOTOR RAILCARS
    • B61C17/00Arrangement or disposition of parts; Details or accessories not otherwise provided for; Use of control gear and control systems
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B61RAILWAYS
    • B61CLOCOMOTIVES; MOTOR RAILCARS
    • B61C17/00Arrangement or disposition of parts; Details or accessories not otherwise provided for; Use of control gear and control systems
    • B61C17/04Arrangement or disposition of driving cabins, footplates or engine rooms; Ventilation thereof
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00Television systems
    • H04N7/18Closed circuit television systems, i.e. systems in which the signal is not broadcast

Abstract

The invention discloses a kind of subway service on buses or trains job informations to acquire intelligent detecting method, vehicle carried data collecting equipment is set in the drivers' cab of subway, it is stored by the acquisition of audio, video data by vehicle carried data collecting equipment, carries out the Monitoring and forecasting system in real-time that electric car driver's value multiplies state;Ground analysis system transmits in real time into row information with vehicle carried data collecting equipment, receives the value that mobile unit uploads and multiply status early warning segment, audio, video data information, storage and analyze, and is managed collectively all electric car videos, audio file.The present invention can be achieved electric car operation information, video data, recording pen voice document and shield fusion together, Display on the same screen, Synchronization Analysis, solid reduction field operation truth, facilitate analysis personnel to count the analysis of corner, the working strength of analysis personnel is reduced, working efficiency is improved.

Description

A kind of subway service on buses or trains job information acquisition intelligent detecting method
Technical field
The invention belongs to on-vehicle information acquisition technique fields, are related to a kind of based on the acquisition intelligence inspection of subway service on buses or trains job information Survey analysis method.
Background technology
With booming, the continuous improvement of electric car drive speed of track dealings transport undertaking, electric car operation Safety problem is more prominent;There are many factor for influencing locomotive safety running, but the driving behavior of electric car driver is in electric car The status that can not ignore is occupied in terms of safe operation.Service on buses or trains relies primarily on personnel to driver's operation or work standard inspection and adds with vehicle at present Multiply, check secretly, transferring the video recording of individual train operator rooms and carry out checking analysis driver's operation process.There is main problem in this way:1, it manages Reason technical staff personnel amount, limited time cannot to driver's operation overall process, it is round-the-clock check stare at control.2, driver's value multiplies Process is currently without the standard operation for combining the audio collecting device of subway operation information to monitor driver, to driver when breaking down Operations Analyst brings certain difficulty.3, driver's value multiply process fatigue driving, the state of mind it is bad when in no monitoring state, to row Vehicle traffic safety causes major hidden danger.In a practical situation, electric car driver needs in locomotive driving process Bu Jian Duan lookouts And calling respond, to ensure the safe operation of electric car.Home and abroad exhales response to answer and drive for electric car driver at present State-detection control technology.
Invention content
Present invention solves the problem in that provide one kind acquiring intelligent detection and analysis method based on subway service on buses or trains job information, Electric car driver value can be monitored and multiply state, acquire audio/video information and stored, three-dimensional can restore field operation actual conditions, It is convenient to check analysis at any time.
The present invention is to be achieved through the following technical solutions:
A kind of subway service on buses or trains job information acquisition intelligent detecting method, is arranged vehicle carried data collecting in the drivers' cab of subway Equipment is stored by the acquisition of audio, video data by vehicle carried data collecting equipment, carries out electric car driver's value and multiply the real-time of state Monitoring and warning;
The vehicle carried data collecting equipment includes video acquisition host and status monitoring terminal coupled respectively, department Machine room video camera, road conditions video camera and recording pen;Wherein, status monitoring terminal, drivers' cab camera acquire trainman respectively The video image of standing working position and driving task position, to carry out driver driving room, whether someone, driver's value multiply state and supervise online Survey and early warning;Road conditions video camera carries out front monitoring road conditions and signal lamp is differentiated, and recording pen realizes whole record locomotive operation mistake Voice handling situations of the journey including calling respond;
The ground analysis system being made of audio and video dump analytical equipment, data server and mobile phone APP terminal, and it is vehicle-mounted Data acquisition equipment is transmitted into row information in real time, is received the value that mobile unit uploads and is multiplied status early warning segment, audio, video data Information, storage are simultaneously analyzed, and all electric car videos, audio file are managed collectively.
The video acquisition host is also according to status monitoring terminal, the video information judgement department of drivers' cab camera acquisition Machine driver's cabin whether someone, if equal nobody, auto-associating ATI etc. correlation driving recording, speed be 0 when then judge machine Vehicle driver is in state of leaving the post and carries out early warning prompting;When speed is not 0, in driver's working position video image when someone, to department The human face region of machine carries out tracing detection, then judges that driver is in fatigue state and space early warning is carried when abnormal detection It wakes up;
It is the video information acquired by camera that driver's value, which multiplies status monitoring, and people is extracted by non-intervention mode Face multiple regions fatigue characteristic accurately detects the electric car crew state of mind, the bad time-division three-level report of the state of mind in real time It is alert to remind;Driver multiplies far from value after status monitoring terminal pickup area reaches certain threshold value, sends out the visual field and deviates and alarms online.
The driver driving state judging method:Using Adaboost algorithm, extraction Haar features train to obtain face Detector carries out tracing detection to face, to detect and navigate to human face region in complex background in real time, detects shape State detection terminal acquisition video image in whether someone;If nobody, the video image of analysis drivers' cab camera acquisition In whether someone;If also nobody in the video image of " drivers' cab camera " acquisition shows when speed is not 0 at driver In the state of leaving the post, the counter that makes to leave the post adds 1, and when leaving the post, counter is more than threshold value T1, then output leaves the post to alarm, and ends processing;Such as Fruit status detection terminal acquisition video image in nobody, drivers' cab camera acquisition video image in someone, it is believed that department Machine is in Zhan Li lookouts;If status detection terminal acquisition video image in someone, illustrate that driver is on seat, then into Enter to carry out driving condition detection.
The driving condition detection uses the method based on integrogram, cascade detectors and AdaBoost algorithms:It uses Harr-like character representation faces realize the quick calculating of character numerical value using integrogram;
Several rectangular characteristics that can most represent face are picked out as Weak Classifier, according to weighting using Adaboost algorithm Weak Classifier is configured to a strong classifier by the mode of ballot;Several strong classifiers that training obtains are composed in series a grade It is coupled the cascade filtering of structure, cascade structure can effectively improve the detection speed of grader, be carried out to the human face region of driver Tracing detection;
Statistics frequency of wink is carried out using PERCLOS algorithms and is blinked the duration, and driver's state for extended periods does not change, eyes It is closed into certain threshold value, illustrates that driver is in fatigue state, if human face region in continuous variation, shows that driver is carrying out just Chang is looked at, and continues to acquire next frame image.
The USB flash disk for acquiring storage dump is also connected on the video acquisition host, USB flash disk accesses vehicle before electric car outbound Loading system host, extracts USB flash disk after electric car storage, and drivers' cab regards i.e. in the storage same day entire driving conditions of electricity car in USB flash disk Frequency evidence and value multiply status information.
The analysis of the ground analysis system includes:
Ground analysis system carries out the associative search of electric car crew's operation ATI record file and audio-video document, with Electric car operation information, video data, recording pen voice document are carried out centered on travelling data with screen fusion and Synchronization Analysis, are stood Body restores field operation actual conditions, completes crucial corner handling situations analysis;
Ground analysis system also carries out log and synchronizes positioning, and the whole log to show this dump record is detailed Thin information makes according to timi requirement video and recording in video display process, and log also can be as video be automatically toward incision It changes;
The terminal of ground analysis system checks the real time execution situation and warning message of online vehicle, when program request check The instantaneous value of line driver multiplies state, monitored over time electricity car driver's state, when electric car driver Chu Xian look at interruption phenomenon triggering When audio alert, data record is automatically formed, and records the audio/video information of electricity car driver when alarming;
Electric car in ground analysis system also statistical analysis for a period of time runs violation situation, by the risk period, by pass Key place, press typical problem, press workshop to violation information carry out statistic of classification summarize.
The dump that the audio and video dump analytical equipment carries out is:
Audio and video automatic dump device will be believed by being communicatively coupled with video acquisition host in the USB flash disk connected thereon Breath carries out dump, and supports the dump simultaneously of 6 road USB flash disks;What real-time reception mobile unit uploaded multiplies status early warning segment, audio, regards Frequency data information memory, and by the clock synchronization system of audio, video data and ATI, which includes that voice and video is defeated Enter module, audio-video signal acquisition module, audio-video signal processing module, audio and video identification module and time synchronization module, it is real Intellectual analysis now is carried out to the audio, video data for being dumped to ground;When judging driver driving exception in crucial corner, do not call Early warning is carried out when response.
The method that the audio identification and time synchronizes is:
End-point detection and feature extraction are carried out to the voice signal of input and voice prompt signal first, end-point detection uses The detection method of short-time energy and short-time average zero-crossing rate is detected using energy and zero-crossing rate as feature, in noise In the case of not low, according to the energy of voice signal be more than noise energy it is assumed that by comparing input signal energy with The size of speech energy threshold value, is distinguish voice segments and non-speech segment;MFCC or LPCC can be used in the feature extraction It realizes;
Then machine learning training, the feature and training pattern of input speech signal are carried out to the feature of voice prompt signal It is matched and is identified, identify the voice prompt signal in input speech signal, and determine that the voice prompt signal is being recorded At the time of in record;
Further according to voice recognition result, the moment where the voice prompt signal in the sound recordings detected is transported with ATI It is synchronized at the time of in row record file where the voice prompt signal.
Compared with prior art, the present invention has technique effect beneficial below:
Subway service on buses or trains job information provided by the invention acquires intelligent detecting method, and the electric car driver based on audio and video drives It sails with calling respond condition detection method, effective management and control with solution to driver's standardized work;It can realize electric visitor Vehicle operation information, video data, recording pen voice document are same to shield fusion, Display on the same screen, Synchronization Analysis;Solid reduction field operation Truth facilitates analysis personnel to count the analysis of corner, reduces the working strength of analysis personnel, improves working efficiency, The effective management and control to driver's standardized work may be implemented, the improper band of locomotive electricity car driving behavior is prevented and reduced to effect The security risk come;Operation can largely be reduced and lack of standardization cause safety accident.
Subway service on buses or trains job information provided by the invention acquires intelligent detecting method, may be implemented tired in driving a vehicle to driver Labor, detection early warning of leaving the post, do not need driver and operate, do not increase driver's burden.Locomotive operation is recorded by recording pen whole process Calling respond in the process and other voice handling situations, meanwhile, recording pen also records the language for having running crane control device to send out Sound standby signal.By USB flash disk dump audio, video data, pass through audio and video dump apparatus automatic high speed dump USB flash disk, recording on ground Data, realize three-in-one Ground analysis.
Subway service on buses or trains job information provided by the invention acquires intelligent detecting method, and driver's value may be implemented and multiply assistant:Department System is not alarmed when machine working condition is good, when occur driver leave the post, sight deviation, the defective modes such as fatigue when, send out voice Driver Zhu Yi lookouts are reminded in alarm;After driver corrects bad working condition, system is automatically stopped audio alert.
Subway service on buses or trains job information provided by the invention acquires intelligent detecting method, and safety management assistant may be implemented:System Original ex-post analysis is changed to runtime on-line monitor management by system, greatly reduces the workload of manual analysis, improves safety management effect Rate and timeliness.By vehicle-mounted collecting device and comprehensive ground information analysis platform, it can not only realize that driver's value multiplies shape on vehicle The real-time monitoring alarming of state at the same by all rolling stock status informations (driver's value multiply state, type of alarm, alarm video, vehicle Fast information, wheelpath etc.) upload to ground surface platform, locomotive operating condition can be monitored in real time, between driver's value is taken the opportunity, driver's value Multiply state.Managerial cadre etc. can by mobile terminal cell phone application, web browser receive specify fleet, locomotive alert notice, And locomotive position can be understood in real time by electronic map, situations such as driver's value multiplies state.To reinforce to department in safe level The management of machine eliminates accident in possible trouble, reduces enterprises' loss to greatest extent, ensures traffic safety.
Subway service on buses or trains job information provided by the invention acquires intelligent detecting method, and three-in-one function may be implemented:It realizes Automatically by USB flash disk video data and recording pen audio data high speed be dumped in local server.Auto-associating locomotive ATI numbers According to keeping audio, video data automatically synchronous with ATI clocks.Realize that electric car operation information, video data, recording pen voice document are same Screen fusion, Synchronization Analysis.The voice signal of input is detected in crucial corner, whether analysis driver's calling respond is marked It is accurate.
Description of the drawings
Fig. 1 is the system structure diagram of the present invention;
Fig. 2 is the camera schematic view of the mounting position of the present invention;
Fig. 3 is the onboard system detection method flow chart of the present invention.
Fig. 4 is audio identification of the present invention and the flow chart that the time synchronizes.
Specific implementation mode
With reference to specific embodiment, the present invention is described in further detail, it is described be explanation of the invention and It is not to limit.
A kind of subway service on buses or trains job information provided by the invention acquires intelligent detecting method, is arranged in the drivers' cab of subway Vehicle carried data collecting equipment is stored by the acquisition of audio, video data by vehicle carried data collecting equipment, carries out electric car driver's value Multiply the Monitoring and forecasting system in real-time of state;
The vehicle carried data collecting equipment includes video acquisition host and status monitoring terminal coupled respectively, department Machine room video camera, road conditions video camera and recording pen;Wherein, status monitoring terminal, drivers' cab camera acquire trainman respectively The video image of standing working position and driving task position, to carry out driver driving room, whether someone, driver's value multiply state and supervise online Survey and early warning;Road conditions video camera carries out front monitoring road conditions and signal lamp is differentiated, and recording pen realizes whole record locomotive operation mistake Voice handling situations of the journey including calling respond;
The ground analysis system being made of audio and video dump analytical equipment, data server and mobile phone APP terminal, and it is vehicle-mounted Data acquisition equipment is transmitted into row information in real time, is received the value that mobile unit uploads and is multiplied status early warning segment, audio, video data Information, storage are simultaneously analyzed, and all electric car videos, audio file are managed collectively.
Referring to Fig. 1, system of the invention is made of vehicle carried data collecting equipment and ground analysis system two parts.Vehicular data It is taken the photograph by video acquisition host, status monitoring terminal (being mounted in front of driver control console, towards driver), drivers' cab according to collecting device Camera, road conditions video camera, four part of recording pen composition.Vehicle carried data collecting realizes that electric car driver's value multiplies the real-time monitoring of state Early warning and the acquisition of audio, video data storage.Mobile wireless network interface card can be loaded in vehicle carried data collecting device Host, with realizing vehicle Real-time communication, by floor synthetic analysis station and mobile phone APP terminal to electric car driver value multiply state monitor on-line, early warning With analysis.
Vehicle carried data collecting carries out following operation:
(1) driver's value multiplies state on_line monitoring and early warning
Using intelligent image processing method, excessively non-intervention mode extracts face multiple regions fatigue characteristic.Not by driver Body difference driving habit, wears the influences such as sunglasses, while not interfering driver's normal driving.Accurately detect that electric car multiplies in real time Business person's state of mind, the bad time-division three-level warning reminding of the state of mind.
After driver reaches certain threshold value far from " value multiplies status monitoring terminal " pickup area, sends out the visual field and deviate in report from a liner It is alert.
(2) " road conditions camera " realizes that front monitoring road conditions and signal lamp are differentiated.
(3) calling respond and other voice handling situations in locomotive operational process are recorded by recording pen whole process, together When, recording pen also records the voice prompt signal for having running crane control device to send out.
(4) USB flash disk real-time storage function:Before electric car outbound, USB flash disk is accessed onboard system host by crew member, after storage USB flash disk is extracted, the video data of drivers' cab and value in the same day entire driving conditions of electricity car are stored in USB flash disk and multiplies status information.
The specific vehicle carried data collecting equipment composition that is given below is introduced:
Video acquisition host is powered using DC110V, and governor circuit is by DM3730, data memory module, network data exchange The external components such as module, power management chip, memory, FLASH, mobile network form.Using (SuSE) Linux OS, system will be schemed As separating after DSP is converted, by DSP coprocessor concurrent operations, complicated image data is quickly handled.Realize locomotive Security information, locomotive state information, the acquisition of locomotive monitoring information, processing and record, the storage of video information, and will alarm Information uploads floor control center.
Status monitoring terminal is using the external components such as low noise camera, the infrared active light sources of 940nm, optical filter composition.Using Infrared ray active illumination and infrared ray optical filtering mode effectively filter out the interference of extraneous various light, have good inhibition to background It is interfered with sunglasses are eliminated, reduces the interference of external environment, while any interference will not be generated to crew.
Drivers' cab camera and road conditions camera use outsourcing finished product machine, high-resolution, low-light (level) to support wide dynamic work( Can, wide dynamic range is more than 75dB, supports 3D digital noise reduction functions, supports SMART IR technologies, supports electronic flutter-proof function, branch Hold digital enlarging function
Recording pen uses the metal shell of firm, durable beauty, electroplating surface processing, the interference far from environmental noise;It adopts With highly sensitive omnidirectional water proof and dust proof microphone, omnidirectional's pickup, clear sound are natural;Using professional audio treatment technology, have Effect prevents voice signal to be distorted and decay;Water proof and dust proof grade IP55;Built-in lightning strike protection, protection to error-polarity connection of power supply and electrostatic Protection.
As depicted in figs. 1 and 2, the present invention passes through " status detection terminal " and " drivers' cab camera " difference harvester first The video image of vehicle driver's standing working position and driving task position.Judge whether someone by camera, if equal nobody, automatically It is associated with the related driving recording such as ATI, then judges that trainman is in state of leaving the post and carries out early warning prompting when speed is not 0; When speed is not 0, in driver's working position video image when someone, tracing detection is carried out to the human face region of driver, when detecting Then judge that driver is in fatigue state and space early warning is reminded when now abnormal." road conditions camera " realizes front monitoring road conditions and letter Signal lamp is differentiated.Recording pen realizes calling respond and other voice handling situations in whole record locomotive operational process, meanwhile, Recording pen also records the voice prompt signal for having running crane control device to send out.
Driver driving condition adjudgement algorithm testing process is as shown in Figure 3.Specific practice:Utilize Adaboost algorithm, extraction Haar features train to obtain human-face detector, tracing detection are carried out to face, to be detected simultaneously in complex background in real time Navigate to human face region, in the video image of detection " status detection terminal " acquisition whether someone;If nobody, analysis " department Machine room camera " acquisition video image in whether someone;If do not had in the video image of " drivers' cab camera " acquisition yet People, when speed is not 0 then illustrating that driver is in leaves the post state, and the counter that makes to leave the post adds 1, and when leaving the post, counter is more than threshold value T1, then output are left the post to alarm, be ended processing;Nobody in the video image of " if status detection terminal " acquisition, " drivers' cab is taken the photograph Someone in the video image of picture head " acquisition, it is believed that driver is in Zhan Li lookouts;The video figure of " if status detection terminal " acquisition Someone as in, then illustrate that driver is on seat, then enters and carry out driving condition detection.
Driving condition detection uses the method based on integrogram, cascade detectors and AdaBoost algorithms, uses Harr- Like character representation faces use " integrogram " to realize the quick calculating of character numerical value;One is picked out using Adaboost algorithm Weak Classifier is configured to one strong point by a little rectangular characteristics (Weak Classifier) that can most represent face in the way of Nearest Neighbor with Weighted Voting Class device;Several strong classifiers that training obtains are composed in series to the cascade filtering of a cascade structure, cascade structure can be effective Ground improves the detection speed of grader, and tracing detection is carried out to the human face region of driver, and carrying out statistics using PERCLOS algorithms blinks Eye frequency and blink the duration, PERCLOS refer within the unit interval eyes closed degree be more than a certain threshold value (70%, 80%) time accounts for the ratio of total time;Frequency of wink is used to indicate that the speed of variation (is opened eyes or be closed) to eye state;Blink Duration refers to eyes from opening to being closed again to the time opened used in this process.
Driver's state for extended periods does not change, eyes closed to certain threshold value, illustrates that driver is in fatigue state, if face area Domain illustrates that driver carries out Zheng Chang lookouts in continuous variation, continues to acquire next frame image.
Ground analysis system carries out following operation:
Ground analysis system is made of audio and video dump analytical equipment, data server, mobile phone APP terminal composition.It is responsible for The value that real-time reception mobile unit uploads multiplies status early warning segment, audio, video data information and storage, is managed collectively all electricity Car video, audio file, build Trinitarian audio/video information united analysis management platform.System realizes that electric car multiplies The associative search function of business person's operation ATI record file and audio-video document, realizes electric car centered on the travelling datas such as ATI Operation information, video data, recording pen voice document are same to shield fusion, Synchronization Analysis.By it is a variety of record data comprehensive analysis, Solid reduction field operation actual conditions complete crucial corner handling situations analysis, improve analysis efficiency.
Realize that log synchronizes positioning, for showing the whole log details of this dump record.Operation After record, it can make in video display process automatically according to timi requirement video and recording, log also can be with video certainly It is dynamic to switch down.
Administrative staff can check the real time execution situation and warning message of online vehicle by PC machine or mobile terminal, When program request check that the instantaneous value of online driver multiplies state.Monitored over time electricity car driver's state, when electric car driver Chu Xian are looked at When being interrupted phenomenon triggering audio alert, system automatically forms data record, and the sound for recording electricity car driver when alarming regards Frequency information, vehicle-mounted part can be locally stored to be analyzed by USB flash disk dump.Electric car operation in statistical analysis for a period of time Violation situation, can by the risk period, by crucial place, by typical problem, press workshop, statistic of classification remittance is carried out to violation information Always, data support is provided effectively to formulate risk management and control measure.
Audio and video automatic dump device realizes that the high speed dump of USB flash disk acquisition information, dump speed can reach 200M per second More than, can support the dump simultaneously of 6 road USB flash disks, parallel dump speed close to gigabit, whole process is not necessarily to manual intervention.Real-time reception vehicle It carries multiply status early warning segment, audio, video data information that equipment uploads to store, is managed collectively all electric car videos, audio File, and by the clock synchronization system of audio, video data and ATI, the clock synchronization system include voice and video input module, Audio-video signal acquisition module, audio-video signal processing module, audio and video identification module and time synchronization module are realized to dump Audio, video data to ground carries out intellectual analysis, when judging driver driving exception in crucial corner, when non-calling respond into Row early warning.Build Trinitarian audio/video information united analysis management platform.
Audio identification and time synchronization implementation mode are introduced (referring to Fig. 4):Voice signal to input and voice prompt letter Number carry out end-point detection and feature extraction.End-point detection can be used short-time energy, short-time average zero-crossing rate, cepstrum feature, entropy, answer The modes such as polygamy (KC complexity and C0 complexity) are realized.MFCC (mel cepstrum coefficients) or LPCC (lines can be used in feature extraction Property prediction cepstrum coefficient) realize.
Machine learning training is carried out to the feature of voice prompt signal, feature and the training pattern of input speech signal carry out Matching and identification identify the voice prompt signal in input speech signal, and determine the voice prompt signal in sound recordings At the time of middle.
According to voice recognition result, the moment where the voice prompt signal in the sound recordings detected is run with ATI It is synchronized at the time of in record file where the voice prompt signal.
A synchronous embodiment is provided below:
When carrying out end-point detection and feature extraction, short-time energy and short-time average zero-crossing rate are selected in this implementation, in short-term can Amount and the detection method of short-time average zero-crossing rate are also referred to as double-threshold comparing method, it is on the basis of short-time energy detection method On, in addition short-time average zero-crossing rate, is detected using energy and zero-crossing rate as feature, it is not very low feelings in signal-to-noise ratio Under condition, the hypothesis of noise energy is more than according to the energy of voice signal, by comparing energy and the speech energy threshold of input signal The size of value can be distinguish voice segments and non-speech segment.That selected in this implementation is MFCC, and voice is extracted using MFCC Information characteristics belong to the technology of comparative maturity, can refer to the prior art to carry out.
Rising by each voice segments in detection " green light passes through " voice signal and input speech signal in the present embodiment Initial point and terminal extract the phonetic feature of voice segments in " green light passes through " voice signal and input signal after end-point detection.
Machine learning training is carried out to " green light passes through " phonic signal character of extraction in this implementation, input speech signal Feature is matched and is identified with training pattern, identifies own the voice signal of " green light passes through " in input speech signal, and At the time of determining each " green light passes through " voice signal place.
It is transported with ATI at the time of the present embodiment is arbitrarily selected in input speech signal where " green light a passes through " voice signal It is compared at the time of owning " green light passes through " voice signal place in row record file, it, will at the time of finding out time difference minimum At the time of the moment is as where " green light passes through " voice signal selected in sound recordings, while passing through sound recordings and ATI Log sheet corresponds to the time difference at two moment of voice signal, corrects the time in recording pen, and recording pen and ATI is made to supervise Control the time synchronization of device.
It realizes that electricity car operation information, video data, recording pen voice document are same using the present invention and shields fusion, same to screen display Show, Synchronization Analysis.Solid reduction field operation truth, facilitates analysis personnel to count the analysis of corner, reduces analysis The working strength of personnel improves working efficiency.Realize the effective management and control to driver's standardized work.
Example given above is to realize the present invention preferably example, and the present invention is not limited to the above embodiments.This field Technical staff any nonessential addition, the replacement made according to the technical characteristic of technical solution of the present invention, belong to this The protection domain of invention.

Claims (8)

1. a kind of subway service on buses or trains job information acquires intelligent detecting method, which is characterized in that vehicle is arranged in the drivers' cab of subway Data acquisition equipment is carried, is stored by the acquisition of audio, video data by vehicle carried data collecting equipment, is carried out electric car driver's value and multiply The Monitoring and forecasting system in real-time of state;
The vehicle carried data collecting equipment includes video acquisition host and status monitoring terminal coupled respectively, drivers' cab Video camera, road conditions video camera and recording pen;Wherein, status monitoring terminal, drivers' cab camera acquire trainman's standing respectively The video image of working position and driving task position, with carry out driver driving room whether someone, driver's value multiply state on_line monitoring and Early warning;Road conditions video camera carries out front monitoring road conditions and signal lamp is differentiated, and recording pen realizes whole record locomotive operational process packet Include the voice handling situations including calling respond;
The ground analysis system being made of audio and video dump analytical equipment, data server and mobile phone APP terminal, with vehicle-mounted data Collecting device in real time into row information transmit, receive mobile unit upload value multiply status early warning segment, audio, video counts it is believed that It ceases, store and analyze, be managed collectively all electric car videos, audio file.
2. subway service on buses or trains job information as described in claim 1 acquires intelligent detecting method, which is characterized in that the video Acquisition host also according to status monitoring terminal, drivers' cab camera acquisition video information judge driver driving room whether someone, If equal nobody, the correlation driving recording such as auto-associating ATI then judges that trainman is in shape of leaving the post when speed is not 0 State simultaneously carries out early warning prompting;When speed is not 0, in driver's working position video image when someone, the human face region of driver is carried out Tracing detection then judges that driver is in fatigue state and space early warning is reminded when abnormal detection;
It is the video information acquired by camera that driver's value, which multiplies status monitoring, and it is more to extract face by non-intervention mode A region fatigue characteristic accurately detects that the electric car crew state of mind, the bad time-division three-level alarm of the state of mind carry in real time It wakes up;Driver multiplies far from value after status monitoring terminal pickup area reaches certain threshold value, sends out the visual field and deviates and alarms online.
3. subway service on buses or trains job information as claimed in claim 1 or 2 acquires intelligent detecting method, which is characterized in that described Driver driving state judging method:Using Adaboost algorithm, extraction Haar features train to obtain human-face detector, to face into Line trace detects, to detect and navigate to human face region, the acquisition of detecting state detection terminal in complex background in real time Video image in whether someone;If nobody, analysis drivers' cab camera acquisition video image in whether someone;If Also nobody in the video image of " drivers' cab camera " acquisition shows that driver is in state of leaving the post when speed is not 0, make from Hilllock counter adds 1, and when leaving the post, counter is more than threshold value T1, then output leaves the post to alarm, and ends processing;If status detection terminal is adopted Nobody in the video image of collection, someone in the video image of drivers' cab camera acquisition, it is believed that driver is in the loud and clear prestige of standing;Such as Someone in the video image of fruit status detection terminal acquisition, then illustrate that driver is on seat, then enters and carry out driving condition inspection It surveys.
4. subway service on buses or trains job information as claimed in claim 3 acquires intelligent detecting method, which is characterized in that the driving State-detection uses the method based on integrogram, cascade detectors and AdaBoost algorithms:It is leted others have a look at using Harr-like mark sheets Face realizes the quick calculating of character numerical value using integrogram;
Several rectangular characteristics that can most represent face are picked out as Weak Classifier, according to Nearest Neighbor with Weighted Voting using Adaboost algorithm Mode Weak Classifier is configured to a strong classifier;Several strong classifiers that training obtains are composed in series a level link The cascade filtering of structure, cascade structure can effectively improve the detection speed of grader, to the human face region of driver into line trace Detection;
Statistics frequency of wink is carried out using PERCLOS algorithms and is blinked the duration, and driver's state for extended periods does not change, eyes closed To certain threshold value, illustrate that driver is in fatigue state, if human face region in continuous variation, shows that driver is normal loud and clear in progress It hopes, continues to acquire next frame image.
5. subway service on buses or trains job information as described in claim 1 acquires intelligent detecting method, which is characterized in that the video The USB flash disk for acquiring storage dump is also connected on acquisition host, USB flash disk accesses onboard system host, electric car before electric car outbound USB flash disk is extracted after storage, and the video data of drivers' cab and value in the same day entire driving conditions of electricity car are stored in USB flash disk and multiplies state letter Breath.
6. subway service on buses or trains job information as described in claim 1 acquires intelligent detecting method, which is characterized in that the ground The analysis of analysis system includes:
Ground analysis system carries out the associative search of electric car crew's operation ATI record file and audio-video document, with driving Data grid technology carries out electric car operation information, video data, recording pen voice document with screen fusion and Synchronization Analysis, and solid is also Former field operation actual conditions complete crucial corner handling situations analysis;
Ground analysis system also carries out log and synchronizes positioning, to show that the whole log of this dump record is believed in detail Breath, according to timi requirement video and recording, makes in video display process, log also can down switch automatically with video;
The terminal of ground analysis system checks the real time execution situation and warning message of online vehicle, when program request check online department The instantaneous value of machine multiplies state, and monitored over time electricity car driver's state triggers voice when loud and clear prestige interruption phenomenon occurs in electric car driver When alarm, data record is automatically formed, and records the audio/video information of electricity car driver when alarming;
Electric car in ground analysis system also statistical analysis for a period of time runs violation situation, by the risk period, by crucially Point summarizes violation information progress statistic of classification by typical problem, by workshop.
7. subway service on buses or trains job information as described in claim 1 or 6 acquires intelligent detecting method, which is characterized in that described Audio and video dump analytical equipment carry out dump be:
Audio and video automatic dump device by being communicatively coupled with video acquisition host, by information in the USB flash disk connected thereon into Row dump, and support the dump simultaneously of 6 road USB flash disks;What real-time reception mobile unit uploaded multiplies status early warning segment, audio, video counts According to information storage, and by the clock synchronization system of audio, video data and ATI, which includes voice and video input mould Block, audio-video signal acquisition module, audio-video signal processing module, audio and video identification module and time synchronization module, realization pair The audio, video data for being dumped to ground carries out intellectual analysis;When crucial corner judge driver driving exception, non-calling respond Shi Jinhang early warning.
8. subway service on buses or trains job information as claimed in claim 7 acquires intelligent detecting method, which is characterized in that the audio Identify that the method synchronized with the time is:
End-point detection and feature extraction are carried out to the voice signal of input and voice prompt signal first, end-point detection is using in short-term The detection method of energy and short-time average zero-crossing rate is detected using energy and zero-crossing rate as feature, signal-to-noise ratio not In the case of low, according to the energy of voice signal be more than noise energy it is assumed that by comparing input signal energy and voice The size of energy threshold, is distinguish voice segments and non-speech segment;The feature extraction can be used MFCC or LPCC and realize;
Then machine learning training is carried out to the feature of voice prompt signal, feature and the training pattern of input speech signal carry out Matching and identification identify the voice prompt signal in input speech signal, and determine the voice prompt signal in sound recordings At the time of middle;
Further according to voice recognition result, the moment where the voice prompt signal in the sound recordings detected is remembered with ATI operations It is synchronized at the time of in record file where the voice prompt signal.
CN201711490984.6A 2017-12-30 2017-12-30 A kind of subway service on buses or trains job information acquisition intelligent detecting method Pending CN108583592A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711490984.6A CN108583592A (en) 2017-12-30 2017-12-30 A kind of subway service on buses or trains job information acquisition intelligent detecting method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711490984.6A CN108583592A (en) 2017-12-30 2017-12-30 A kind of subway service on buses or trains job information acquisition intelligent detecting method

Publications (1)

Publication Number Publication Date
CN108583592A true CN108583592A (en) 2018-09-28

Family

ID=63633158

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711490984.6A Pending CN108583592A (en) 2017-12-30 2017-12-30 A kind of subway service on buses or trains job information acquisition intelligent detecting method

Country Status (1)

Country Link
CN (1) CN108583592A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109545027A (en) * 2018-12-24 2019-03-29 郑州畅想高科股份有限公司 A kind of practical traning platform, crew's simulation training method and device
CN110213548A (en) * 2019-07-01 2019-09-06 南京派光智慧感知信息技术有限公司 A kind of track train driving behavior comprehensive monitoring alarming method for power
CN110618635A (en) * 2019-10-08 2019-12-27 中兴飞流信息科技有限公司 Train cab operation specification monitoring system based on AI technology

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101639894A (en) * 2009-08-31 2010-02-03 华南理工大学 Method for detecting train driver behavior and fatigue state on line and detection system thereof
CN103035238A (en) * 2012-11-27 2013-04-10 中国科学院自动化研究所 Encoding method and decoding method of voice frequency data
CN103150560A (en) * 2013-03-15 2013-06-12 福州龙吟信息技术有限公司 Method for realizing intelligent safe driving of automobile
CN103310590A (en) * 2012-03-06 2013-09-18 上海骏聿数码科技有限公司 System and method for driver fatigue analysis and early-warning
CN104732251A (en) * 2015-04-23 2015-06-24 郑州畅想高科股份有限公司 Video-based method of detecting driving state of locomotive driver
CN106408877A (en) * 2016-11-17 2017-02-15 西南交通大学 Rail traffic driver fatigue state monitoring method
CN107506698A (en) * 2017-08-01 2017-12-22 无锡南理工科技发展有限公司 The method of public transportation vehicle anti-fatigue-driving management based on Internet of Things

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101639894A (en) * 2009-08-31 2010-02-03 华南理工大学 Method for detecting train driver behavior and fatigue state on line and detection system thereof
CN103310590A (en) * 2012-03-06 2013-09-18 上海骏聿数码科技有限公司 System and method for driver fatigue analysis and early-warning
CN103035238A (en) * 2012-11-27 2013-04-10 中国科学院自动化研究所 Encoding method and decoding method of voice frequency data
CN103150560A (en) * 2013-03-15 2013-06-12 福州龙吟信息技术有限公司 Method for realizing intelligent safe driving of automobile
CN104732251A (en) * 2015-04-23 2015-06-24 郑州畅想高科股份有限公司 Video-based method of detecting driving state of locomotive driver
CN106408877A (en) * 2016-11-17 2017-02-15 西南交通大学 Rail traffic driver fatigue state monitoring method
CN107506698A (en) * 2017-08-01 2017-12-22 无锡南理工科技发展有限公司 The method of public transportation vehicle anti-fatigue-driving management based on Internet of Things

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109545027A (en) * 2018-12-24 2019-03-29 郑州畅想高科股份有限公司 A kind of practical traning platform, crew's simulation training method and device
CN110213548A (en) * 2019-07-01 2019-09-06 南京派光智慧感知信息技术有限公司 A kind of track train driving behavior comprehensive monitoring alarming method for power
CN110213548B (en) * 2019-07-01 2021-09-07 南京派光智慧感知信息技术有限公司 Rail train driver behavior comprehensive monitoring and warning method
CN110618635A (en) * 2019-10-08 2019-12-27 中兴飞流信息科技有限公司 Train cab operation specification monitoring system based on AI technology

Similar Documents

Publication Publication Date Title
CN109300471B (en) Intelligent video monitoring method, device and system for field area integrating sound collection and identification
CN101655907B (en) Trainman driving state monitoring intelligent alarm system
CN108583592A (en) A kind of subway service on buses or trains job information acquisition intelligent detecting method
CN103714631B (en) ATM cash dispenser intelligent monitor system based on recognition of face
CN206097257U (en) Security protection intelligent monitoring system
EP3279700B1 (en) Security inspection centralized management system
WO2019153193A1 (en) Taxi operation monitoring method, device, storage medium, and system
CN101196505A (en) Passageway security checking terminal system and security checking method using the same
CN111064925B (en) Subway passenger ticket evasion behavior detection method and system
CN109286788A (en) A kind of safety defense monitoring system based on cloud computing
CN205990117U (en) Intelligent security guard apparatus for controlling elevator
CN103700220A (en) Fatigue driving monitoring device
CN110525456A (en) A kind of train safe driving monitoring system and method
CN110781798A (en) Vehicle-mounted suspect locking system based on raspberry group and face recognition
CN112686090A (en) Intelligent monitoring system for abnormal behaviors in bus
CN112530144A (en) Method and system for warning violation behaviors of thermal power plant based on neural network
CN111654665A (en) Intelligent tail money cabinet monitoring system
CN105227918B (en) A kind of intelligent control method
KR102043922B1 (en) A cctv divided management system and method it
RU2684484C1 (en) Method and cognitive system for video analysis, monitoring, control of driver and vehicle state in real time
CN110889919A (en) Production workshop staff-fixing supervision method and system
CN212208379U (en) Automatic attendance temperature measuring system
CN105095891A (en) Human face capturing method, device and system
CN112633057A (en) Intelligent monitoring method for abnormal behaviors in bus
CN112165611A (en) One-way intelligent channel image and light curtain article remaining identification system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20180928

RJ01 Rejection of invention patent application after publication