Summary of the invention
It is an object of the invention to overcome above-mentioned the deficiencies in the prior art, it is provided that a kind of new intelligent early-warning device that the construction machinery under transmission line of electricity is monitored safely and the method for early warning analysis thereof.The system and method for the present invention is based on video image intelligent analysis technology, there is the ability independently analyzed video data, only having external force destructive insident to occur when, just return useful data and information, to reduce operation management expense, all the time monitoring screen is dig-inned without monitoring personnel, as long as paying close attention to early warning monitoring device to beam back the validity event data of Surveillance center, thus alleviate the working strength of supervisor.
In order to reach foregoing invention purpose, the technical scheme that patent of the present invention provides is as follows:
First the present invention there is provided a kind of supervising device having and carrying out safe early warning for event of constructing under power transmission line route.This device is powered by DC source, may be installed on electric power pylon, and accessible cvbs analog video signal or SDI digital video signal also carry out graphical analysis and process to video signal.
Construction machinery safety prewarning monitoring system under a kind of power transmission line route, it is characterised in that this prewarning monitoring system includes main control module, Computer Vision module, communication module and power management module, wherein:
Computer Vision module is for being acquired the vedio data around transmission line of electricity, video and graphic is carried out data analysis, identify external force destructive insident, produce warning message, and the warning message of dependent event and image/video are passed through network interface transfers to communication module;
Communication module, transmits to Surveillance center by the way of radio communication for warning message and related data Computer Vision module inputted;
Power management module, connects main control module, Computer Vision module and communication module respectively, is used for performing power management work;
Main control module, connects Computer Vision module, communication module and power management module respectively, manages its mode of operation and monitoring state.
Under power transmission line route of the present invention in construction machinery safety prewarning monitoring system, the input interface of described Computer Vision module is bnc interface, utilizes this bnc interface to input to cvbs analog video signal or SDI high-definition digital video signal.
Under power transmission line route of the present invention in construction machinery safety prewarning monitoring system, described Computer Vision module utilizes the video signal of image A/D chip Gather and input;For cvbs analog video signal, image A/D chip is the TW9910 of TECKWELL company, is output as BT656 signal;For SD1 high-definition digital video signal, image A/D chip is the GV7610 of gennum company, is output as BT1120 signal.
Under power transmission line route of the present invention in construction machinery safety prewarning monitoring system, described power management module includes reserve battery and electric power management circuit, and described reserve battery continues as main control module when external input power breaks down and communication module provides working power.
Under power transmission line route of the present invention in construction machinery safety prewarning monitoring system, use serial ports 485 to communicate between described main control module with Computer Vision module, the mode of operation of Computer Vision module is set and obtains the work state information of Computer Vision module;Communication interface between main control module and power management module is RS485, and main control module obtains the duty of power management module, controls the operational module of power management module.
The present invention is next to that provide a kind of early warning analysis method for event of constructing under power transmission line route.The method is on the basis of Computer Vision, whether the feature that under joint line, big machinery construction event occurs, have the external force destructive insident to transmission line of electricity to occur, and result reported Surveillance center in the monitoring range of comprehensive descision visual field, there is monitoring range big, actively the feature of warning.
A kind of construction machinery safe early warning monitoring method under power transmission line route, it is characterised in that the method comprises the steps:
The first step, scene calibration, in the scene the depth of field is set by calibration algorithm, standard is set to the scene depth of field, accurately to calculate physical location and the size of target in scene;Described scene calibration algorithm realizes step:
Line segment is demarcated in 1a. input;To image, three line segments are inputted as three demarcation line segments from input module;
1b. is according to demarcating line segment, it is thus achieved that camera parameter;Listing respectively and demarcate the corresponding calibration equation of line segment with every, then solution is by the calibration equation group of gained, it is thus achieved that camera parameter include the height H of camera, the pitching angle theta of camera and horizontal vanishing line ι;
1c. is according to camera parameter, it is thus achieved that and export the calibrating parameters of target;Height according to camera, the angle of pitch, the camera parameter of horizontal vanishing line, calculate target in image includes that target corresponds to the actual width calibrating parameters of the coordinate of real space, the true altitude of target, target, and exports this calibrating parameters from output module;
Second step, arranges early warning rule, and including arranging safe distance and holdup time under warning region, line, described warning region is to the delimitation forbidding construction area under line, if there being moving target to enter this region, it will be detected;Under line, safe distance is to the setting of the range of activity of moving target in warning region, if the scope of moving target motion has exceeded the permissible value of safe distance under line, it will be detected;
3rd step, moving object detection, moving object detection is that the motion target area in video scene splits from background, with the moving target in accurately detection transmission line of electricity monitoring scene image, including people, car, filter the interference factor including flating, the tree of swing, brightness flop, shade, rain and snow simultaneously;
4th step, motion target tracking, it is achieved in that and first does target prediction to after moving object detection success, for estimating that the next frame of target moves, carry out object matching again, for the stable objects of tracking and matching and filter false target, finally carry out target update, the renewal of stable objects in present frame;
5th step, moving object classification, based on physical space, moving target is identified, including feature extraction and type decision, feature extraction is in order to extract space characteristics and the temporal characteristics of target, and type decision is then in order to judge the type of target;
6th step, event judges, moving target is after detection with Classification and Identification, in conjunction with the rule of second step setting, it may be judged whether have big machinery to swarm into warning region, and judge whether its moving region exceedes set safe distance;
7th step, report to the police and export, loudspeaker and warning lamp by site setup carry out scene and warn, and relevant information and video data remotely send to Surveillance center simultaneously, and this Surveillance center is provided with rear end management software so that supervisor can remotely check the real-time video at scene.
Under power transmission line route of the present invention in construction machinery safe early warning monitoring method, in described 3rd step, the specific practice of moving object detection includes:
3a. carries out Image semantic classification, for eliminating its impact on background model;
3b. carries out zone marker, for background model carries out foreground segmentation, and marks connected region one by one, safeguards the state of detection, and for judging the state that detection program is presently in, if occurring, detection is abnormal, then restart detection;
3d. does region to be strengthened, for rejecting the false areas including shade, highlighted and leaf swing;
3e. carries out regional split and merges, and is used for solving target over-segmentation and the mutual occlusion issue of target.
Under power transmission line route of the present invention in construction machinery safe early warning monitoring method, in described 5th step, space characteristics includes the major axis of region contour fitted ellipse, the short axle of region contour fitted ellipse, the short axle of matching and the angle of horizontal direction, the class circle property of region contour, the degree of compacting in region, the area in region, upper part and the lower part area ratio in region;Temporal characteristics includes the history classification information of the velocity characteristic of target, the periodic movement feature of target and target.
Under power transmission line route of the present invention in construction machinery safe early warning monitoring method, the type decision of described 5th step includes every frame classification and overall classification, described every frame classification is: according to major axis, the short axle of region contour fitted ellipse, the short axle of matching and horizontal direction angle, the class circle property of region contour, the degree of compacting in region, the area in region, upper part and lower part area ratio and the velocity characteristic of target, the periodic movement feature of target in region of region contour fitted ellipse, every frame classification is carried out, to obtain the history classification information of target by probability classification;Described overall classification is classified according to the history classification information of target, to judge the type of target.
Under power transmission line route of the present invention in construction machinery safe early warning monitoring method, as further prioritization scheme, the thought of described probability classification is: respectively according to people, crowd, car, car group and toy type, gives the weights that the major axis of region contour fitted ellipse, the short axle of region contour fitted ellipse, the short axle of matching are different from the feature of the class of horizontal direction angle, region contour circle property, the degree of compacting in region, the area in region, the upper and lower part area ratio in region and the velocity characteristic of target, the periodic movement feature of target;Calculate the characteristic probability value of people, crowd, car, Che Qun, toy type, and the present frame that type the is target classification information that select probability value is maximum the most respectively.
Based on technique scheme, under the power transmission line route of the present invention, construction machinery safety prewarning monitoring system and monitoring method achieve following beneficial effect in practical application:
First the intelligent early-warning device that the present invention relates to has possessed the video monitoring function of ordinary video supervising device, on this basis, use Computer Vision that video signal is carried out field assay, big machinery construction, personnel are swarmed into etc. cause tripping operation, break, fall the accidents such as tower formation, develop, the overall process that deteriorates is monitored in real time, and have give warning in advance, the function of initiative alarming, it is to avoid the generation of these events brings heavy losses to national economy and industrial or agricultural, people's lives electricity consumption.
Detailed description of the invention
Under we combine accompanying drawing and specific embodiment transmission line of electricity a kind of to the present invention below, the system and method for construction machinery safe early warning monitoring is described in detail; to further appreciate that its structure composition and working method, but the scope of protection of the invention is not limited thereto.
Certain job site is positioned near power transmission line, big machinery often travel under electric power pylon through, the most also personnel walk up and down from line.For the practical situation of this on-the-spot application, the present invention and system and method can the situation of effective monitoring site construction, if it find that there is big machinery superelevation to drive through or operation, also active reporting of being warned in scene.
As it is shown in figure 1, construction machinery safety prewarning monitoring system includes main control module 11, Computer Vision module 12, communication module 13 and power management module 14 under power transmission line route of the present invention.Wherein:
Computer Vision module 12 is for being acquired the vedio data around transmission line of electricity, video and graphic is carried out data analysis, identify external force destructive insident, produce warning message, and the warning message of dependent event and image/video are passed through network interface transfers to communication module 13;
Communication module 13, transmits to Surveillance center by the way of radio communication for warning message and related data Computer Vision module 12 inputted;
Power management module 14, connects main control module 11, Computer Vision module 12 and communication module 13 respectively, is used for performing power management work;
Main control module 11, connects Computer Vision module 12, communication module 13 and power management module 14, respectively with mode of operation and the duty of monitoring management modules.
In 4 comprising modules within above-mentioned early warning monitoring device, Computer Vision module 12, communication module 13, the mode of operation of power management module 14 and monitoring state are managed by main control module 11;Vedio data around transmission line of electricity is acquired by Computer Vision module 12, and video and graphic is carried out data analysis, identifies external force destructive insident, produces warning message, and is transmitted to communication module 13 with image/video by the warning message of dependent event.Communication module 13 provides the wireless communication modes such as 3G, WIFI, the warning message of intelligent early-warning device and related data can be reached by the way of various wireless communication Surveillance center.Power management module 14 is then that other three modules perform power management work.
Specifically, main control module 11 is responsible for being monitored the duty of whole intelligent early-warning device.Main control module 11 uses the arm7 of low-power consumption as core processor, and the power consumption of whole module is the lowest, and at about 2W during duty, stand-by power consumption even can arrive 500mW.The power supply of Computer Vision module 12 with communication module 13 can be managed by main control module 11 as required, can open and close the power supply of the two module as required, to reach the purpose of power saving;The two module can also be carried out not power-off and restart control, to realize not power-off service.
The core processor of Computer Vision module 12 uses the DM8127 of American TI Company.The input interface of Computer Vision module 12 is bnc interface, accessible cvbs analog video signal or SDI high-definition digital video signal.Computer Vision module 12 first has to the video signal of input is carried out data acquisition, and the circuit realiration of this part is realized by image A/D chip.Simulating incoming video signal for cvbs, image AD is the TW9910 of TECKWELL company, is output as BT656 signal.For SDI high-definition digital video, image AD is the GV7610 of gennum company, is output as BT1120 signal.Above-mentioned BT656 or BT1120 digital video signal is analyzed by Computer Vision module 12, after being processed by DSP such as some row filtering, region enhancing, characteristics extraction, pattern recognition, picture coding, compression of images, analysis result and vedio data are delivered to communication module 13 by network interface.
Communication module 13 is to support the wireless router of 3G, WIFI, communication module 13 is responsible for the wireless communication setting up main control module 11 with Computer Vision module 12 to background monitoring center, data and the information of prior-warning device are uploaded to background monitoring center, and receive the instruction from background monitoring center.
Power management module 14 is mainly made up of with electric power management circuit reserve battery.Electric power management circuit carries out branch switch management to the power supply of whole device.Reserve battery can continue as main control module 11 when external input power breaks down and communication module 13 provide working power, makes prior-warning device can work in the way of low-power consumption 48 hours, and carrying out remote maintenance for attendant provides field data for reference.
Communication interface between above-mentioned main control module 11 and power management module 14 is RS485, can obtain the status information of power management module in time, and the mode of operation of power module is controlled management.Using serial ports 485 to communicate between main control module 11 with Computer Vision module 12, the pattern that can be operated Computer Vision module 12 arranges and obtains its work state information.The output interface of Computer Vision module 12 is 10/100M network interface, is connected to communication module 13.Also there is the connection of a serial ports 485 between main control module 11 and communication module 13, the work state information of communication module 13 can be obtained, and mode of operation is configured.
The present invention relates to construction machinery under power transmission line route and carry out the method for safe early warning be: by video scene is analyzed, realize the background modeling to video scene, real moving region is extracted by technology such as inter-frame difference, morphologic filter, shadow detection and territorial classifications, based on mode identification technology, region is carried out matched jamming obtain moving target and carry out target classification, target location detection obtained compares with the warning region of user setup, it may be judged whether have personnel's (vehicle) to invade;Target location detection obtained compares with the warning region of user setup, by 3D calibration algorithm Coordinate Conversion, judge size and the shape facility of moving target, the statistics target holdup time in warning region, exceed the time threshold of setting, carry out alarm, target location detection obtained compares with the safe distance of user setup, if the safe distance that the position, region of target is demarcated through user, then it is judged as that safe distance detection occurs abnormal, send alarm signal, if the area coordinate information at target place is on the safe distance that user demarcates, the most now will not produce abnormal alarm, it is thus possible to by the object of some high airflights, such as aircraft, flying birds etc. get rid of, improve the accuracy of safe distance algorithm detection.
As shown in Figure 2, it is achieved the comprising the concrete steps that of construction machinery safe early warning monitoring analysis method under power transmission line route of the present invention:
The first step, scene calibration.
In video monitoring, the position that same target is the most different has different sizes, monitors the most at a distance, and this just brings problem for intelligent video analysis.To solve this problem it is necessary to arrange the depth of field by calibration algorithm in the scene, a standard is set deeply to scene.Demarcated by the depth of field, physical location and the size of target in scene can be calculated exactly, this provides for improved accuracy and the reliability of intelligent video analysis.
The scene calibration that the present invention relates to uses a kind of scaling method based on line segment to realize, and scene calibration algorithm concrete implementation step be given below:
Step one: line segment is demarcated in input;To image, three line segments are inputted as three demarcation line segments from input module;
Step 2: according to demarcating line segment, it is thus achieved that camera parameter;List respectively and demarcate the calibration equation that line segment is corresponding with every, the calibration equation group that then solution is made up of the calibration equation of gained, include the height H of camera, the angle of pitch of camera and the camera parameter of horizontal vanishing line to obtain;
Step 3: according to camera parameter, it is thus achieved that and export the calibrating parameters of target;Height according to camera, the angle of pitch, the camera parameter of horizontal vanishing line, calculate target in image includes that target corresponds to the actual width calibrating parameters of the coordinate of real space, the true altitude of target, target, and exports this calibrating parameters from output module.
At least there is a demarcation line segment in described three demarcation line segments and be in the far-end of scene corresponding to image, at least there is a demarcation line segment and be in the near-end of scene corresponding to image;And described three demarcation line segments all can not be with image lower edges parallel.
Second step, arranges early warning rule, including arranging safe distance, holdup time etc. under warning region and line.
Warning region is to the delimitation forbidding construction area under line, if there being moving target to enter this region, it will be detected.
Under line, safe distance is to the setting of the range of activity of moving target in warning region, if the scope of moving target motion has exceeded the permissible value of safe distance under line, it will be detected.
3rd step, moving object detection.
Moving object detection is that the motion target area in video scene splits from background.
Moving object detection can accurately detect the moving target in transmission line of electricity monitoring scene image, including people, car, can filter the impact of the interference factors such as flating, the tree of swing, brightness flop, shade, rain, snow simultaneously.
Specific practice is first to carry out Image semantic classification, for eliminating its impact on background model.Then carrying out zone marker, for background model being carried out foreground segmentation, and marking connected region one by one, now the state of detection is safeguarded, for judging the state that detection program is presently in, abnormal if there is detection, then restart detection;Then do region to strengthen, be used for rejecting the false areas such as shade, highlighted and leaf swing;Finally carry out regional split and merge, be used for solving target over-segmentation and the mutual occlusion issue of target.
4th step, motion target tracking.
Motion target tracking is achieved in that
After moving object detection success, target prediction to be done, for estimating the next frame motion of target;Then carry out object matching, for the stable objects of tracking and matching, and filter some false targets;Finally carry out target update, the template renewal of stable objects in present frame.
5th step, moving object classification.
Moving object classification is mainly moving target to be carried out examination not, including feature extraction and type decision based on physical space.Wherein feature extraction is in order to extract space characteristics and the temporal characteristics of target, and type decision is then in order to judge the type of target.
Described space characteristics includes: the major axis of region contour fitted ellipse;The short axle of region contour fitted ellipse;The short axle of matching and horizontal direction angle;The class circle property of region contour;The degree of compacting in region;The area in region;Upper and the lower part area ratio in region.
Described temporal characteristics includes: the velocity characteristic of target, the periodic movement feature of target, the history classification information of target.
Type decision includes every frame classification and overall classification.
Described every frame classification is: according to major axis, the short axle of region contour fitted ellipse, the short axle of matching and horizontal direction angle, class circle property, the degree of compacting in region, the area in region, the upper and lower part area ratio in region and the velocity characteristic of target, the periodic movement feature of target of region contour of region contour fitted ellipse, every frame classification is carried out, to obtain the history classification information of target by probability classification.
The design philosophy of probability classification described above is: respectively according to types such as people, crowd, car, Che Qun, toys, and the imparting major axis of region contour fitted ellipse, the short axle of region contour fitted ellipse, the short axle of matching justify the weights that the features such as the periodic movement feature of property, the degree of compacting in region, the area in region, the upper and lower part area ratio in region and the velocity characteristic of target, target are different from the class of horizontal direction angle, region contour;Calculate the characteristic probability value of the types such as people, crowd, car, Che Qun, toy, and the present frame that type the is target classification information that select probability value is maximum the most respectively.
Described overall classification is classified according to the history classification information of target, to judge the type of target.
7th step, event judges.
Moving target is after detection with Classification and Identification, in conjunction with the rule of second step setting, it may be judged whether have big machinery to swarm into warning region, and judge whether its moving region exceedes set safe distance.
8th step, output of reporting to the police.
The form of output of reporting to the police is to carry out scene by on-the-spot loudspeaker and warning lamp to warn promptings, and relevant information and video data are sent remotely to the rear end pipe platform software of Surveillance center simultaneously, and supervisor can remotely check on-the-spot real-time video.
As a concrete application case, as it is shown on figure 3, the specific implementation process of the safety prewarning monitoring system of the present invention and early warning analysis method is as follows:
The first step, headend equipment including, front end early warning monitoring device 1, high-speed ball-forming machine 2, sun-generated electric power 4, audible-visual annunciator 3 are installed connect, specifically, front end early warning monitoring device 1 connects high-speed ball-forming machine 2, audible-visual annunciator 3 and rear end Management software platform 5 respectively with transmitting data information, sun-generated electric power 4 connects front end early warning monitoring device 1, high-speed ball-forming machine 2 and audible-visual annunciator 3 respectively to provide power supply, and front end early warning monitoring device 1 is connected to Surveillance center and transfers data to rear end Management software platform 5.On the cross-arm of the steel tower that the equipment such as front end early warning monitoring device 1, high-speed ball-forming machine 2, audible-visual annunciator 3 and sun-generated electric power 4 are typically mounted at ultra-high-tension power transmission line.
Second step, carries out rule setting on rear end Management software platform 5 to front end early warning monitoring device 1.Rule arranges and includes that the demarcation to scene and warning region divide and safe distance, holdup time etc. are configured.The size of people, dilly is mainly demarcated by scene calibration, also has and demarcates the size of big machinery exactly, as distinguishing big machinery and dilly, people, the foundation of toy later.After headend equipment connects, front end early warning monitoring device 1 can carry out facility registration, it can be seen that early warning monitoring device is in line states, and it can be seen that the monitoring scene at scene on rear end Management software platform 5 by Management software platform 5 after starting the most to the back-end.
Rear end Management software platform 5 can be delimited by the way of drawing closed line in monitoring scene warning region, and be set in the interior time threshold being detained of warning region.Below transmission line of electricity, suitable position sets safe distance by the way of setting-out section, is safety zone at setting-out line section lower zone, and top is then deathtrap, and the most all of moving target does not allows to arrive drawn line segment superjacent air space.After setting warning region, holdup time parameter and safe distance line, preserve and arrange, and remote reboot front end early warning monitoring device 1, parameter is set will be saved in front end early warning monitoring device 1 at this moment.
3rd step, front end early warning monitoring device 1 starts normal work after restarting, if it find that there is big machinery to swarm into, and the border of range of movement is up to or over default safe distance line, then produce warning message, and video and data are got forwarded to rear end Management software platform 5.The on-the-spot loudspeaker started in acoustooptic alarm 3 is propagandaed directed to communicate, and warning lamp starts flicker, reminds the big machinery principle transmission line of electricity of hot work in progress.
Usually under line personnel walk about and sailing into of dilly is rolled away from, when entering the warning region delimited, owing to having done sizing calibration for moving target, so big machinery will not be judged as, and owing to its range of movement still has distance from safe distance line, so audible and visible alarm will not be produced at the scene, but it can also be seen that monitoring system has identified it has swarmed into warning region and its movement locus is tracked record on rear end Management software platform 5.If the most above-mentioned head end video supervising device 1 has the ability independently analyzed video data, only having external force destructive insident to occur when, just return useful data and information, to reduce flow to reduce operation management expense.If video monitoring system has an intelligent video analysis function, then dig-inning monitoring screen all the time without monitoring personnel, as long as paying close attention to early warning monitoring device to beam back the validity event data of Surveillance center, thus alleviating the working strength of supervisor.