Blowdown intelligent control method based on real-time videoTechnical field
The present invention relates to vision signal control technology fields, more particularly, to a kind of blowdown intelligence based on real-time videoIt can monitoring method.
Background technique
Currently, it has been recognized that sewage discharge for various negative effects caused by environment, is just taking measures pairThe discharge of sewage is limited.So far, the measure taken substantially has the following two kinds：One is must be managed by environmental protectionPersonnel carry out personal monitoring to enterprise's effluent discharge outlet, the information of enterprise's sewage effluent are obtained, to the enterprise of discharge in violation of regulations sewageIndustry is punished.Such measure is not only time-consuming and laborious, not can guarantee continuous monitoring in daily 24 hours.And due to human factor compared withIt is more, it is difficult to ensure that preventing blowdown and accurate to the punishment of blowdown enterprise, rationally.Another measure is then for personal monitoring instituteThere are the shortcomings that and design, it be sewage monitoring instrument is mounted on sewage disposal device, and by sewage monitoring instrument withWater pump and the valve being mounted on blowoff line connect, and water pump and valve can only be realized according to the output signal of sewage monitoring instrumentIt opens or closes, will be emitted in time by the middle water for handling, meeting pollution emission standard, raw sewerage is prevented to discharge,Its high degree of automation can save a large amount of manpowers.But due to sewage disposal device all be installation enterprise place in, it is aOther enterprise illegally installs around the sewage pipe of sewage disposal device, valve, finally by raw sewerage from enterpriseEffluent discharge outlet discharge；Moreover, because set valve does not have power-off restoration function, when valve is in the open state and prominentSo when power-off, valve will still be kept it turned on.In this way, being in valve if artificially sewage disposal device is allowed to power offThen raw sewerage is discharged by the valve opened from enterprise's effluent discharge outlet for open state.
In this regard, in the prior art, application No. is the Chinese invention patent applications of CN03238004.6 to disclose a kind of sewageAutomatic monitoring device is discharged, sensor is equipped with, is connected to signal processor, the output of signal processor and power-off bullet with sensorSpring homing position type executing agency connects, and spring to break homing position type executing agency connects with sewage discharge valve.Enterprise can be installed inIndustry effluent discharge outlet, the water timing monitoring that enterprise is discharged.When the water of enterprise's discharge meets pollution emission standard, valve is beatenOpen, otherwise valve close, alarm, can prevent enterprise illegally install bypass sewage disposal device drainage pipeline and will be unprocessedSewage discharge.However, this method still will use sensor, and the detection position of sensor is fixed, as long as sewageDischarger gets around the position, then still can not effectively monitor the truth of sewage discharge.
Summary of the invention
Unfavorable in order to overcome the requirement of sensor fixed installation to monitor blowdown, the present invention provides one kind based on real-timeThe blowdown intelligent control method of video, including：
(1) blowdown intelligent monitoring model is established by way of machine learning；
(2) blowdown monitoring is carried out by the video information that unmanned plane obtains in real time.
Further, the step (1) includes：
(10) geographical information library is established；
(20) video information identification model is established according to the geographical information library by machine learning mode.
Further, the step (2) includes：
(30) UAV Video information is obtained；
(40) it according to UAV Video information and the model, determines sewage location and issues warning information.
Further, the step (10) includes：At least two width obtained around Sewage outlet are continuously clapped in timeThe image taken the photograph, the image can uniquely identify corresponding Sewage outlet, and position a certain in image and image is correspondingLatitude and longitude information be saved in database jointly, as mark sewage discharge ground geographical information library.
Further, the machine learning is that engineering is carried out in a manner of unsupervised learning according to vegetation growth state imageIt practises.
Further, the machine learning is to carry out engineering to vegetation growth state image using stochastic gradient descent methodIt practises.
Further, the step (20) includes：
(2021) key frame information determines：Assuming that earth's surface vegetation map corresponds to vegetation health status Cj as Ei；Vegetation health shapeThe corresponding possible cause Sm of state Cj constitutes set { Sm, Pm }, then using vegetation health status Cj as key frame, wherein Pm is possible formerThe probability of vegetation health status Cj caused by becoming because of Sm, i, j and m are the natural number since 1；
(2022) probability of the appearance for possible cause of vegetation health status Cj is defined：
p(Sm|Cj)=χgh(pj),
Wherein
M=1,2,3,4,5,6；AndFor with forMean value, ξmFor the m rank diagonal matrix of variance,
(2023) according to Probability p (Sm|Cj) determine when vegetation health status Cj takes current meaning and the matching degree in season：
It calculatesWherein p ' indicates to carry out difference to p；
It calculatesWhether less than the first preset threshold：When smallYu Shi determines that the serial number for the possible cause that j is indicated in Cj meets Ei corresponding season, otherwise enables j=j+1, jump to step(2022), it if j reaches its maximum value by traversal, enables j=1 and continues step (2024), u and v are natureNumber；
(2024) when correction corresponding possible cause of the Sm as Cj and the matching degree in season：
It calculatesWhether less than secondPreset threshold：When being less than, determines that Sm meets season as the corresponding possible cause of Cj, otherwise enable m=m+1, jump to stepSuddenly (2022) enable m=1 if m reaches its maximum value by traversal.
Further, the step (30) includes：
(301) framing sampling is carried out to the video of camera acquisition；
(302) sample image is normalized；
(303) feature extraction is carried out to the image after normalization using convolutional neural networks.
Further, the step (40) includes：
The video information obtained based on the unmanned plane and the model, when determining that the corresponding possible cause of certain image is dirtyWhen water, that is, eliminate the surface vegetation indicated according to image after the predetermined reason such as weather, pest and disease damage color andThe comparison of the characteristics of image such as growth conditions determines that possible cause is caused by sewage discharge, then according to the figure of figure unmanned plane acquisitionAs corresponding second latitude and longitude information of information, make again unmanned plane shooting and second latitude and longitude information it is immediate, describedThe image of the corresponding sewage draining exit of already existing latitude and longitude information in geographical information library determines that Location for Sewage and sewage draining exit mayThe location being related to issues warning information to the relevant monitoring of the sewage draining exit or administrative staff.
Beneficial effects of the present invention include：The figure such as color, growth conditions of surface vegetation can be obtained by machine learningAs feature with whether be that sewage causes being associated between this possible cause, based on 6 rank depth probability analysis methods carry out mayThe matching of reason and season, so that way compared with prior art reduces the operand more than 37% or so.
Detailed description of the invention
Fig. 1 shows the flow chart of the method for the present invention.
Specific embodiment
As shown in Figure 1, preferred embodiment in accordance with the present invention, the blowdown intelligence based on real-time video that the present invention provides a kind ofEnergy monitoring method, including：
(1) blowdown intelligent monitoring model is established by way of machine learning；
(2) blowdown monitoring is carried out by the video information that unmanned plane obtains in real time.
Preferably, the step (1) includes：
(10) geographical information library is established；
(20) video information identification model is established according to the geographical information library by machine learning mode；
Preferably, the step (2) includes：
(30) UAV Video information is obtained；
(40) it according to UAV Video information and the model, determines sewage location and issues warning information.
Preferably, the step (10) includes：At least two width obtained around Sewage outlet are continuously shot in timeImage, which can uniquely identify corresponding Sewage outlet, and position a certain in image and image is correspondingLatitude and longitude information is saved in database jointly, the geographical information library as mark sewage discharge ground.
Preferably, the machine learning is that engineering is carried out in a manner of unsupervised learning according to vegetation growth state imageIt practises.
Preferably, the machine learning is to carry out engineering to vegetation growth state image using stochastic gradient descent methodIt practises.
Preferably, the step (20) includes：
(2021) key frame information determines：Assuming that earth's surface vegetation map corresponds to vegetation health status Cj as Ei；Vegetation health shapeThe corresponding possible cause Sm of state Cj constitutes set { Sm, Pm }, then using vegetation health status Cj as key frame, wherein Pm is possible formerThe probability of vegetation health status Cj caused by becoming because of Sm, i, j and m are the natural number since 1；
(2022) probability of the appearance for possible cause of vegetation health status Cj is defined：
p(Sm|Cj)=χgh(pj),
Wherein
M=1,2,3,4,5,6；AndFor with forMean value, ξmFor the m rank diagonal matrix of variance,
(2023) according to Probability p (Sm|Cj) determine when vegetation health status Cj takes current meaning and the matching degree in season：
It calculatesWherein p ' indicates to carry out difference to p；
It calculatesWhether less than the first preset threshold：When smallYu Shi determines that the serial number for the possible cause that j is indicated in Cj meets Ei corresponding season, otherwise enables j=j+1, jump to step(2022), it if j reaches its maximum value by traversal, enables j=1 and continues step (2024), u and v are natureNumber；
(2024) when correction corresponding possible cause of the Sm as Cj and the matching degree in season：
It calculatesWhether less than secondPreset threshold：When being less than, determines that Sm meets season as the corresponding possible cause of Cj, otherwise enable m=m+1, jump to stepSuddenly (2022) enable m=1 if m reaches its maximum value by traversal.
Preferably, the step (30) includes：
(301) framing sampling is carried out to the video of camera acquisition；
(302) sample image is normalized；
(303) feature extraction is carried out to the image after normalization using convolutional neural networks.
Preferably, the step (40) includes：
The video information obtained based on the unmanned plane and the model, when determining that the corresponding possible cause of certain image is dirtyWhen water, that is, eliminate the surface vegetation indicated according to image after the predetermined reason such as weather, pest and disease damage color andThe comparison of the characteristics of image such as growth conditions determines that possible cause is caused by sewage discharge, then according to the figure of figure unmanned plane acquisitionAs corresponding second latitude and longitude information of information, make again unmanned plane shooting and second latitude and longitude information it is immediate, describedThe image of the corresponding sewage draining exit of already existing latitude and longitude information in geographical information library determines that Location for Sewage and sewage draining exit mayThe location being related to issues warning information to the relevant monitoring of the sewage draining exit or administrative staff.
The above-described embodiments merely illustrate the principles and effects of the present invention, and is not intended to limit the present invention.It is any ripeThe personage for knowing this technology all without departing from the spirit and scope of the present invention, carries out modifications and changes to above-described embodiment.CauseThis, institute is complete without departing from the spirit and technical ideas disclosed in the present invention by those of ordinary skill in the art such asAt all equivalent modifications or change, should be covered by the claims of the present invention.