CN104237808A - Electronic police system light supplementary lamp fault detecting method based on image abnormities - Google Patents

Abstract

The invention discloses an electronic police system light supplementary lamp fault detecting method based on image abnormities. Based on vehicle images captured by high-definition cameras of an electronic police system and vehicle license plate identification, the vehicle flow rate and the vehicle license plate identification rate in a unit time frame are calculated, the vehicle flow rate and the vehicle license plate identification rate in the unit time frame are compared with the normal vehicle flow rate and the normal vehicle license plate identification rate of the same time frame, whether the vehicle detection and the vehicle license plate identification on lanes are abnormal or not is judged, whether light supplementary lamps are abnormal or not is judged according to set threshold values of the vehicle flow rate and the vehicle license plate identification rate in abnormity, and light supplementary lamp fault detection is conducted. The electronic police system light supplementary lamp fault detecting method based on the image abnormities is simple in implementation, an algorithm is easy to conduct, the detection accuracy and operability are high, no hardware is added, the operation efficiency of the system is not influenced, a large amount of manpower, material resources and financial resources are saved, the maintenance cost is reduced, and the using prospect is quite wide.

Description

A kind of electronic police system light compensating lamp fault detection method based on image abnormity

One, technical field

The invention belongs to the intelligent transportation category of computer utility, particularly relate to computer image treatment field, specifically a kind of electronic police system light compensating lamp fault detection method based on image abnormity.

Two, background technology

In recent years, along with socioeconomic fast development and urban population increasing, vehicle guaranteeding organic quantity increases rapidly, cause traffic congestion clearly, traffic hazard is obviously risen, in urban transportation, make a dash across the red light, drive in the wrong direction, exceed the speed limit, the phenomenon such as lane change violating the regulations happens occasionally, cause a large amount of casualties and property loss every year, bring serious problems to country, society and family, in order to the driving behavior of specification driver, reduce the generation of traffic hazard, electronic police system is all set up at crossing in a lot of city now.

Electronic police system is exactly install 5,000,000 (or 2,000,000) pixel high-definition digital video camera at traffic light intersection, and connect traffic light signal system, all vehicles by crossing are detected, when there is situation violating the regulations in vehicle, system captures the image violating the regulations of vehicle automatically, and automatically identify the number plate of vehicles peccancy, using the vaild evidence of punishing as running red light for vehicle.

Along with electronic police system a large amount of construction at home, the maintenance work of electronic police equipment is very large, because electronic police system equipment belongs to outdoor equipment, equipment is subject to being exposed to the sun and drenching with rain for a long time, easily break down, wherein, night is the LED light supplement lamp that video camera carries out light filling, the most easily breaks down.After light compensating lamp breaks down, vehicle detection at night and Car license recognition can be greatly affected, and therefore, how can carry out automatic decision to fault after light compensating lamp breaks down, and alarm reparation is in time abnormal important.

Electronic police system light compensating lamp adopts LED stroboscopic light filling lamp usually, and the place of the most easily breaking down in LED light supplement lamp is LED lamp bead and control panel, and LED lamp bead easily burns out after aging, and the life-span of conventional LED light supplement lamp only has 2 ~ 3 years usually.LED light supplement lamp fault mainly contains two classes: 1) light compensating lamp light weakens, and normally light compensating lamp occurs that a small amount of lamp pearl is bad, and light compensating lamp light is weakened; 2) light compensating lamp does not work, and normally light compensating lamp lamp pearl all burns out or the damage of light compensating lamp control panel.

Run well for enabling electronic police system, after light compensating lamp equipment breaks down, the method of simple and fast how is adopted to judge equipment failure situation, and alarm is in time extremely important, this not only support equipment can be keeped in repair in time, a large amount of manpowers can also be reduced regular inspection is carried out to equipment.

Three, summary of the invention

The object of the invention is the problem easily broken down for electronic police system light compensating lamp equipment, a kind of vehicle image captured by statistics high-definition camera is provided, calculate vehicle detection rate and Car license recognition rate, light compensating lamp fault is judged in time, and fault is carried out to the method for alert process in time.This method can judge light compensating lamp fault in time, and analysis of failure type is carried out and alarm failure condition, and facilitate technician to safeguard faulty equipment in time, safeguards system is normally run; Without the need to the human and material resources of at substantial and financial resources, front-end equipment is regularly patrolled, reduce maintenance cost.

Basic ideas of the present invention are: after light compensating lamp breaks down, and capture vehicle image brightness night and reduce, and vehicle detection rate is understood with Car license recognition rate and be reduced rapidly.By analyzing vehicle detection rate and Car license recognition rate down ratio, judging which class fault light compensating lamp belongs to, and carrying out alert process.This thinking is the setting based on video camera in electronic police system and light compensating lamp: because 1 5,000,000 (or 2,000,000) pixel camera can cover 2 ~ 3 tracks usually, every root track needs 1 light compensating lamp, therefore, a camera association 2 ~ 3 light compensating lamps, the general simultaneously bad possibility of these 2 ~ 3 light compensating lamps is less, usual fault is that one of them light compensating lamp does not work or light weakens, the present invention carries out analysis according to the change that image after multiple local electronic police system light compensating lamp fault occurs and finds, after there is damage in a light compensating lamp, the image effect in this track is poor, color is darker, and the impact of other tracks is little, the chroma-luminance change of direct employing image judges that light compensating lamp is abnormal, the effect that can not obtain.Electronic police system all carries out video capture and Car license recognition to each car through crossing, therefore, the present invention adopt to the verification and measurement ratio of vehicle and Car license recognition rate judge the vehicle detection in this track and Car license recognition whether abnormal, judge whether light compensating lamp has problems with this, and judgement is lamp pearl partial destruction or all damages.

The object of the present invention is achieved like this: based on the vehicle image adopting electronic police system high-definition camera to capture and Car license recognition, vehicle flow and Car license recognition rate in the unit of account period, vehicle flow in unit time period is compared with identical period normal vehicle flow and Car license recognition rate with Car license recognition rate, judge that whether vehicle detection and the Car license recognition in this track be abnormal, the lamp pearl partial destruction of light compensating lamp or the threshold value of lamp pearl whole damaged condition time abnormal according to the vehicle flow and Car license recognition rate that more whether exceed setting, judge whether light compensating lamp has problems, carry out the detection of light compensating lamp fault.

First obtain the car data excessively that electronic police equipment is in X days each tracks under normal circumstances, and be period unit with 1 hour, the vehicle flow N in the statistics every root track of day part every day _ijk, Car license recognition rate R _ijk, the then statistics data of first X days, the average traffic flow of statistics every day same period with Car license recognition rate

Selected 6 time periods as differentiation of morning every night 20 to next day, obtain the vehicle flow value n in the track of present period _jk, Car license recognition rate r _jk, calculate Current vehicle flow n _jkwith front X day average difference with ratio S _jk, Car license recognition rate r _jkwith front X day average difference T _jk, and set S respectively _jkand T _jkthreshold value time abnormal;

To point in morning 6 at 20 in evening to next day, the S in each each track of whole period _jk, T _jkvalue judges, when finding that there is a S _jk, T _jkvalue occurs extremely, namely compares exceptional value, finds that this value is greater than the threshold values of setting, then continue to observe subsequent period, if same abnormal conditions appear in subsequent period, then more adjacent light compensating lamp covers the S of two periods of track simultaneously _jk, T _jkvalue, if it be normal that adjacent light compensating lamp covers the value in track, then tentatively judges the light compensating lamp exception in this track, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then alarm in time, by manually transferring realtime graphic, and determines whether that light compensating lamp is abnormal.

Wherein: i=1 ... X, i days before expression current time;

J=1 ... 24, represent the data of jth hour;

K=1 ... Y, represents kth root track;

X is the number of days needing the statistical average period;

Y is the track sum that single camera covers;

N _ijkrepresent the vehicle flow in i-th day j period kth root track before current time;

R _ijkrepresent the vehicle identification rate in i-th day j period kth root track before current time;

represent the average traffic flow in X days j period kth root tracks before current time;

represent the average Car license recognition rate in X days j period kth root tracks before current time;

N _jkrepresent the vehicle flow in current j period kth root track;

R _jkrepresent the Car license recognition rate in current j period kth root track;

S _jkrepresent current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio;

T _jkrepresent current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference;

Concrete implementation step is as follows:

1), choose electronic police equipment under normal circumstances before X days each tracks cross car data, and be period unit with 1 hour, the vehicle flow adding up the every root track of day part every day is N _ijk, the correct vehicle numerical digit identifying car plate is M _ijk, and calculate Car license recognition rate R _ijk,

R _ijk=M _ijk/N _ijk

2) the average traffic flow in the X days every root tracks of day part before, calculating the average traffic number of correct identification car plate average Car license recognition rate

${\stackrel{\‾}{N}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\Σ}}}{N}_{\mathrm{ijk}}$

${\stackrel{\‾}{M}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\Σ}}}{M}_{\mathrm{ijk}}$

${\stackrel{\‾}{R}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\Σ}}}{R}_{\mathrm{ijk}}$

3), selected 6 conducts in morning every night 20 to next day differentiate the time period, the vehicle flow value n of acquisition present period _jk, Car license recognition rate r _jk;

4) the difference ratio of Current vehicle flow, Car license recognition rate and front X day average, is calculated:

$S_{\mathrm{jk}}=\frac{|n_{\mathrm{jk}}-{\stackrel{\‾}{N}}_{\mathrm{jk}}|}{{\stackrel{\‾}{N}}_{\mathrm{jk}}}$

T _jk=|r _jk-R _jk|

5), judge

System at night 20 to point in morning 6 next day, each whole period is all to the S in each track _jk, T _jkvalue calculates, and judges that whether this value is abnormal, as one of them S of discovery _jk, T _jkvalue occurs extremely, namely exceptional value is compared, find that this value is greater than the threshold values of setting, then continue to observe subsequent period, if same abnormal conditions appear in subsequent period, then compare the value of two periods of adjacent lane simultaneously, if the value of adjacent lane is normal, then tentatively judge that the light compensating lamp in this track is abnormal, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks continuous two periods _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then manually checks image, determines that whether light compensating lamp is abnormal;

6), after, according to 5 judging that light compensating lamp is abnormal, alarming processing is carried out in time;

7), according to warning information, manually transfer realtime graphic, whether light compensating lamp is abnormal to utilize manual type finally to confirm, and carries out respective handling.

Light compensating lamp fault detect flow process is:

Step 1: the S calculating each period _jk, T _jkvalue;

Step 2: if present period S _jk> δ _s1and T _jk> δ _t1, then perform step 3, otherwise proceed to step 1;

Step 3: if subsequent period S _jk> δ _s1and T _jk> δ _t1, then perform step 4, otherwise proceed to step 1;

Step 4: calculate adjacent lane two period S _jk, T _jkvalue, if adjacent lane two period S _jk< δ _s1and T _jk< δ _t1, then perform step 5, otherwise proceed to step 6;

Step 5: if present period S _jk> δ _s2and T _jk> δ _t2, then current lane light compensating lamp is bad, otherwise current lane light compensating lamp part lamp pearl is bad, proceeds to step 7;

Step 6: the S judging two period all tracks on daytime _jk, T _jkwhether value normal, if daytime two period all tracks S _jk, T _jkvalue is normal, be then judged as that light compensating lamp is completely bad, otherwise be judged as fault of camera, proceed to step 7 and carry out manual confirmation;

Step 7: carry out alarming processing, by artificial contrast's image, judges that whether light compensating lamp is abnormal;

In above-mentioned steps: δ _s1represent according to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jks when judging that lamp pearl part is bad _jkthreshold values; δ _s2represent according to current time vehicle flow n _jkwith front X days same period average vehicle flow difference with ratio S _jkjudge lamp pearl whole bad time S _jkthreshold values;

δ _t1represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that lamp pearl part is bad _jkthreshold values; δ _t2represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge lamp pearl whole bad time T _jkthreshold values.

According to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jks when judging that lamp pearl part is bad _jkthreshold values δ _s1=0.4;

According to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jkjudge lamp pearl whole bad time S _jkthreshold values δ _s2=0.8;

According to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that lamp pearl part is bad _jkthreshold values δ _t1=0.5;

According to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge lamp pearl whole bad time T _jkthreshold values δ _t2=0.9.

The number of days X of statistical average period is needed to be 5 ~ 8; The value of the track total number Y that setting single camera covers is 1 ~ 3.

Good effect of the present invention is:

1, do not need outfield inspection and the internal field inspection of carrying out timing, save a large amount of human and material resources and financial resources, reduce maintenance cost.

2, break down to light compensating lamp and detect fast, the type of automatic discrimination fault, timely alarm, makes fault solve faster, to ensure the quality gathering image.

3, the method realizes simple, and algorithm is easy, and detection accuracy is high, workable, does not increase any hardware, can not increase too many calculated amount, not the operational efficiency of influential system, and prospect of the application is very wide.

Five, embodiment

See accompanying drawing.

The present invention adopt vehicle detection rate that high-definition camera is captured and Car license recognition rate judge the vehicle detection in this track and Car license recognition whether abnormal, judge whether light compensating lamp has problems, carry out light compensating lamp fault detect.In the present embodiment, electronic police system video camera is 5,000,000 pixel high-definition digital video cameras, and each pixel camera covers 2 ~ 3 tracks, and every root track needs 1 light compensating lamp, a camera association 2 ~ 3 light compensating lamps.

Set 5 ~ 8 days for needing the number of days X of statistical average period; The track total number Y that setting single camera covers is 1 ~ 3.

Setting is according to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jks when judging that lamp pearl part is bad _jkthreshold values δ _s1=0.4;

Setting is according to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jkjudge lamp pearl whole bad time S _jkthreshold values δ _s2=0.8;

Setting is according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that lamp pearl part is bad _jkthreshold values δ _t1=0.5;

Setting is according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge lamp pearl whole bad time T _jkthreshold values δ _t2=0.9.

First obtain the car data excessively that electronic police equipment is in X days each tracks under normal circumstances, and be period unit with 1 hour, the vehicle flow N in the statistics every root track of day part every day _ijk, Car license recognition rate R _ijk, the then statistics data of first X days, the average vehicle flow of statistics every day same period with Car license recognition rate

Selected 6 time periods as differentiation of morning every night 20 to next day, obtain the vehicle flow value n of present period _jk, Car license recognition rate r _jk, calculate Current vehicle flow n _jkwith front X day average difference with ratio S _jk, Car license recognition rate r _jkwith front X day average difference T _jk, and S during light compensating lamp lamp pearl partial destruction when setting is abnormal respectively _jkthreshold value and light compensating lamp lamp pearl T when all damaging _jkthreshold value.

To point in morning 6 at 20 in evening to next day, the S in each each track of whole period _jk, T _jkvalue judges, when finding that there is a S _jk, T _jkvalue occur abnormal after, namely exceptional value is compared, S when light compensating lamp lamp pearl partial destruction or lamp pearl all damage when finding that this value is greater than the exception of setting _jk, T _jkthreshold values, then continue to observe subsequent period, if same abnormal conditions appear in subsequent period, then more adjacent light compensating lamp covers the S of two periods of track simultaneously _jk, T _jkvalue, if it be normal that adjacent light compensating lamp covers the value in track, then tentatively judges the light compensating lamp exception in this track, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then alarm in time, by manually transferring realtime graphic, and determines whether that light compensating lamp is abnormal.

Wherein: i=1 ... X, i days before expression current time; J=1 ... 24, represent the data of jth hour;

K=1 ... Y, represents kth root track; X is the number of days needing the statistical average period;

Y is the track sum that single camera covers;

N _ijkrepresent the vehicle flow in i-th day j period kth root track before current time;

R _ijkrepresent the vehicle identification rate in i-th day j period kth root track before current time;

represent the average traffic flow in X days j period kth root tracks before current time;

represent the average Car license recognition rate in X days j period kth root tracks before current time;

N _jkrepresent the vehicle flow in current j period kth root track;

R _jkrepresent the Car license recognition rate in current j period kth root track;

S _jkrepresent current time flow n _jkwith front X days same period average traffic flows difference with ratio;

T _jkrepresent current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference;

Concrete implementation step is as follows:

1), choose electronic police equipment under normal circumstances before X days each tracks cross car data, and be period unit with 1 hour, the vehicle flow adding up the every root track of day part every day is N _ijk, the correct vehicle numerical digit identifying car plate is M _ijk, and calculate Car license recognition rate R _ijk,

R _ijk=M _ijk/N _ijk

2) the average traffic flow in the X days every root tracks of day part before, calculating the average traffic number of correct identification car plate average Car license recognition rate

${\stackrel{\&OverBar;}{N}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\&Sigma;}}}{N}_{\mathrm{ijk}}$

${\stackrel{\&OverBar;}{M}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\&Sigma;}}}{M}_{\mathrm{ijk}}$

${\stackrel{\&OverBar;}{R}}_{\mathrm{jk}}=\underset{i}{\overset{X}{\mathrm{\&Sigma;}}}{R}_{\mathrm{ijk}}$

3), selected 6 conducts in morning every night 20 to next day differentiate the time period, the vehicle flow value n of acquisition present period _jk, Car license recognition rate r _jk;

4) the difference ratio of Current vehicle flow, Car license recognition rate and front X day average, is calculated:

$S_{\mathrm{jk}}=\frac{|n_{\mathrm{jk}}-{\stackrel{\&OverBar;}{N}}_{\mathrm{jk}}|}{{\stackrel{\&OverBar;}{N}}_{\mathrm{jk}}}$

T _jk=|r _jk-R _jk|

5), judge

System at night 20 to point in morning 6 next day, each whole period is all to the S in each track _jk, T _jkvalue calculates, and judges that whether this value is abnormal, as one of them S of discovery _jk, T _jkvalue occur abnormal after, namely exceptional value is compared, S when light compensating lamp lamp pearl partial destruction or lamp pearl all damage when finding that this value is greater than the exception of setting _jk, T _jkthreshold values, then continue to observe subsequent period, if there are same abnormal conditions in subsequent period, then compare the value of two periods of adjacent lane simultaneously, if the value of adjacent lane is normal, then tentatively judge that the light compensating lamp in this track is abnormal, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks continuous two periods _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then manually checks image, determines that whether light compensating lamp is abnormal;

6), after, according to 5 judging that light compensating lamp is abnormal, alarming processing is carried out in time;

7), according to warning information, manually transfer realtime graphic, whether light compensating lamp is abnormal to utilize manual type finally to confirm, and carries out respective handling.

Drawings illustrate and use the present invention to carry out fault detect to electronic police system light compensating lamp and judge flow process:

Step 1: the S calculating each period _jk, T _jkvalue;

Step 2: if present period S _jk> δ _s1and T _jk> δ _t1, then perform step 3, otherwise proceed to step 1;

Step 3: if subsequent period S _jk> δ _s1and T _jk> δ _t1, then perform step 4, otherwise proceed to step 1;

Step 4: calculate adjacent lane two period S _jk, T _jkvalue, if adjacent lane two period S _jk< δ _s1and T _jk< δ _t1, then perform step 5, otherwise proceed to step 6;

Step 5: if present period S _jk> δ _s2and T _jk> δ _t2, then current lane light compensating lamp is bad, otherwise current lane light compensating lamp part lamp pearl is bad, proceeds to step 7;

Step 6: the S judging two period all tracks on daytime _jk, T _jkwhether value normal, if daytime two period all tracks S _jk, T _jkvalue is normal, be then judged as that light compensating lamp is completely bad, otherwise be judged as fault of camera, proceed to step 7 and carry out manual confirmation;

Step 7: carry out alarming processing, by artificial contrast's image, judges that whether light compensating lamp is abnormal.

In above-mentioned steps: δ _s1represent according to current time flow n _jkwith front X days same period average vehicle flow difference with ratio S _jks when judging that lamp pearl part is bad _jkthreshold values; δ _s2represent according to current time flow n _jkwith front X days same period average vehicle flow difference with ratio S _jkjudge lamp pearl whole bad time S _jkthreshold values;

δ _t1represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that lamp pearl part is bad _jkthreshold values; δ _t2represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge lamp pearl whole bad time T _jkthreshold values.

Four, accompanying drawing explanation

Accompanying drawing is fault detect process flow diagram of the present invention.

Claims (5)

1. the electronic police system light compensating lamp fault detection method based on image abnormity, it is characterized in that: based on the vehicle image adopting electronic police system high-definition camera to capture and Car license recognition, vehicle flow and Car license recognition rate in the unit of account period, vehicle flow in unit time period is compared with identical period normal vehicle flow and Car license recognition rate with Car license recognition rate, judge that whether vehicle flow and the Car license recognition in this track be abnormal, time abnormal according to the vehicle flow and Car license recognition rate that more whether exceed setting, whether the threshold decision light compensating lamp of light compensating lamp lamp pearl partial destruction or whole damaged conditions has problems, carry out the detection of light compensating lamp fault:

First obtain the car data excessively that electronic police equipment is in X days each tracks under normal circumstances, and be period unit with 1 hour, the vehicle flow N in the statistics every root track of day part every day _ijk, Car license recognition rate R _ijk, the then statistics data of first X days, the average traffic flow of statistics every day same period with Car license recognition rate

Selected 6 time periods as differentiation of morning every night 20 to next day, obtain the vehicle flow value n in present period track _jk, Car license recognition rate r _jk, calculate Current vehicle flow n _jkwith front X day average difference with ratio S _jk, Car license recognition rate r _jkwith front X day average difference T _jk; And S during light compensating lamp lamp pearl partial destruction when setting is abnormal respectively _jkthreshold value and light compensating lamp lamp pearl T when all damaging _jkthreshold value;

To point in morning 6 at 20 in evening to next day, the S in each each track of whole period _jk, T _jkvalue judges, when finding that there is a S _jk, T _jkvalue occur abnormal after, namely exceptional value is compared, light compensating lamp lamp pearl partial destruction or the S that all damages when finding that this value is greater than the exception of setting _jk, T _jkthreshold values, then continue to observe subsequent period, if same abnormal conditions appear in subsequent period, then more adjacent light compensating lamp covers the S of two periods of track simultaneously _jk, T _jkvalue, if it be normal that adjacent light compensating lamp covers the value in track, then tentatively judges the light compensating lamp exception in this track, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then alarm in time, by manually transferring realtime graphic, and determines whether that light compensating lamp is abnormal;

Wherein: i=1 ... X, i days before expression current time;

J=1 ... 24, represent the data of jth hour;

K=1 ... Y, represents kth root track;

X is the number of days needing the statistical average period;

Y is the track sum that single camera covers;

N _ijkrepresent the vehicle flowrate in i-th day j period kth root track before current time;

R _ijkrepresent the vehicle identification rate in i-th day j period kth root track before current time;

represent the average traffic flow in X days j period kth root tracks before current time;

represent the average Car license recognition rate in X days j period kth root tracks before current time;

N _jkrepresent the vehicle flow in current j period kth root track;

R _jkrepresent the Car license recognition rate in current j period kth root track;

S _jkrepresent current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio;

T _jkrepresent current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference.

2. as claimed in claim 1 based on the electronic police system light compensating lamp fault detection method of image abnormity, it is characterized in that: concrete implementation step is as follows:

1), choose electronic police equipment under normal circumstances before X days each tracks cross car data, and be period unit with 1 hour, the vehicle flow adding up the every root track of day part every day is N _ijk, the correct vehicle numerical digit identifying car plate is M _ijk, and calculate Car license recognition rate R _ijk,

R _ijk=M _ijk/N _ijk

2) the vehicle average discharge in the X days every root tracks of day part before, calculating the average traffic number of correct identification car plate average Car license recognition rate

3), selected 6 conducts in morning every night 20 to next day differentiate the time period, the vehicle flow value n of acquisition present period _ik, Car license recognition rate r _jk;

4) the difference ratio of vehicle flow, Car license recognition rate and front X day average, is calculated:

T _jk=|r _jk-R _jk|

5), judge

System at night 20 to point in morning 6 next day, each whole period is all to the S in each track _jk, T _jkvalue calculates, and judges that whether this value is abnormal, as one of them S of discovery _jk, T _jkvalue occur abnormal after, namely exceptional value is compared, light compensating lamp lamp pearl partial destruction or S when all damaging when finding that this value is greater than the exception of setting _jk, T _jkthreshold values, then continue to observe subsequent period, if there are same abnormal conditions in subsequent period, then compare the value of two periods of adjacent lane simultaneously, if the value of adjacent lane is normal, then tentatively judge that the light compensating lamp in this track is abnormal, and according to the size of threshold values, judgement is that light compensating lamp is completely bad or part lamp pearl is bad; If all there is S in all tracks continuous two periods _jk, T _jkvalue is abnormal, then judge the S in all tracks of daytime period _jk, T _jkwhether value is normal, if daytime is normal, then manually checks image, determines that whether light compensating lamp is abnormal;

6), after, according to 5 judging that light compensating lamp is abnormal, alarming processing is carried out in time;

7), according to warning information, manually transfer realtime graphic, whether light compensating lamp is abnormal to utilize manual type finally to confirm, and carries out respective handling.

3., as claimed in claim 1 based on the electronic police system light compensating lamp fault detection method of image abnormity, it is characterized in that: light compensating lamp fault detect flow process is:

Step 1: the S calculating each period _jk, T _jkvalue;

Step 2: if present period S _jk> δ _s1and T _jk> δ _ti, then perform step 3, otherwise proceed to step 1;

Step 3: if subsequent period S _jk> δ _s1and T _jk> δ _t1, then perform step 4, otherwise proceed to step 1;

Step 4: calculate adjacent lane two period S _jk, T _jkvalue, if adjacent lane two period S _jk< δ _s1and T _jk< δ _t1, then perform step 5, otherwise proceed to step 6;

Step 5: if present period S _jk> δ _s2and T _jk> δ _t2, then current lane light compensating lamp is bad, otherwise current lane light compensating lamp part lamp pearl is bad, proceeds to step 7;

Step 6: the S judging two period all tracks on daytime _jk, T _jkwhether value normal, if daytime two period all tracks S _jk, T _jkvalue is normal, be then judged as that light compensating lamp is completely bad, otherwise be judged as fault of camera, proceed to step 7 and carry out manual confirmation;

Step 7: carry out alarming processing, by artificial contrast's image, judges that whether light compensating lamp is abnormal;

In above-mentioned steps: δ _s1represent according to current time vehicle flow n _jkwith front X days same period average traffic flows difference with ratio S _jks when judging that light compensating lamp lamp pearl part is bad _jkthreshold values; δ _s2represent according to current time vehicle flow n _jkwith front X days same period average traffic difference with ratio S _jkjudge light compensating lamp lamp pearl whole bad time S _jkthreshold values; δ _t1represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that light compensating lamp lamp pearl part is bad _jkthreshold values; δ _t2represent according to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge light compensating lamp lamp pearl whole bad time T _jkthreshold values.

4., as claimed in claim 3 based on the electronic police system light compensating lamp fault detection method of image abnormity, it is characterized in that:

According to current time vehicle flow n _jkwith front X days same period average traffic vehicle flowrates difference with ratio S _jks when judging that light compensating lamp lamp pearl part is bad _jkthreshold values δ _s1=0.4;

According to current time vehicle flow n _jkwith front X days same period average vehicle flow difference with ratio S _jkjudge light compensating lamp lamp pearl whole bad time S _jkthreshold values δ _s2=0.8;

According to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkt when judging that light compensating lamp lamp pearl part is bad _jkthreshold values δ _t1=0.5;

According to current time Car license recognition rate r _jkcar license recognition rate average with front X days synchronization difference T _jkjudge light compensating lamp lamp pearl whole bad time T _jkthreshold values δ _t2=0.9.

5. as claimed in claim 1 based on the electronic police system light compensating lamp fault detection method of image abnormity, it is characterized in that: setting needs the number of days X of statistical average period to be 5 ~ 8; The value of the track total number Y that setting single camera covers is 1 ~ 3.