CN102479416B - Method and device for eliminating false alarm in monitoring system - Google Patents
Method and device for eliminating false alarm in monitoring system Download PDFInfo
- Publication number
- CN102479416B CN102479416B CN201010565898.9A CN201010565898A CN102479416B CN 102479416 B CN102479416 B CN 102479416B CN 201010565898 A CN201010565898 A CN 201010565898A CN 102479416 B CN102479416 B CN 102479416B
- Authority
- CN
- China
- Prior art keywords
- vector
- alarm
- state description
- template
- color space
- 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.)
- Expired - Fee Related
Links
Abstract
The invention discloses a method for eliminating a false alarm in a monitoring system. The method comprises the following steps: carrying out color space conversion on image data of an alarm zone, thus reducing the color dimensionality; generating state description vector of the image data, wherein the state description vector is utilized as an quantitative index for describing image characteristics; and calculating the distance between the state description vector and a template vector, wherein the template vector is the state description vector of the image data of the alarm zone under a non-alarm state. The invention also discloses a device for eliminating a false alarm in a monitoring system, and the device comprises a color space conversion unit, a state description vector generation unit coupled with the color space conversion unit and an alarm judgment unit coupled with the state description vector generation unit. The invention further discloses a system for eliminating a false alarm in a monitoring system. By utilizing the method for eliminating the false alarm, the false alarm caused by light reflection on a side panel due to flickering of a faucet indicating lamp in an existing monitoring system can be eliminated and the alarm accuracy rate is improved.
Description
Technical field
The present invention relates to ATM field of intelligent monitoring, relate in particular to a kind of method, system and device of eliminating false alarm in supervisory system.
Background technology
The safety problem of ATM perplexs banking industry and ATM manufacturer always, in order to Strengthen Safety Management, generally adopts the method for camera head monitor, one, two or more cameras is installed between ATM is withdrawn the money, monitoring user operation process.The camera of ATM between withdrawing the money both comprised monitoring, identification user's camera, also comprised the camera of monitoring ATM itself and user's operating process.
Artificial monitoring need to expend a large amount of manpower and financial resources, nor can ensure monitoring effect.The importance of ATM intelligent monitor system in social safety is day by day remarkable.Intelligent monitor system is generally to determine whether reporting to the police according to the variation of prospect, background two two field pictures.For example, the Chinese invention patent application that application number is 201010109588.6 discloses monitor processing method and system in a kind of monitoring, in the time that human body does not appear in monitored body region, obtains monitored body original image and monitored body detecting image; In the time judging that described monitored body original image and described monitored body detecting image there are differences, send detecting alarm signal.Determine whether reporting to the police by the variation of movement images.
In the time that reality is carried out intelligent monitoring for ATM, card inserting mouth pilot lamp region has been set to shadow zone, not in image ratio scope; But, because the glimmer reflective image change of prospect, background relevant position that also can cause that causes of card inserting mouth pilot lamp is larger, thereby brought false alarm on side panel.
Iff not reporting to the police and can bring the risk of failing to report by shadow zone being simply set, this situation has directly affected the reliability of intelligent monitor system.
Summary of the invention
The problem that the present invention solves is to provide a kind of reflective method that causes false alarm in intelligent monitor system of eliminating, in eliminating because of the reflective false alarm causing, only avoid, by the shadow zone risk of failing to report of bringing of not reporting to the police is simply set, improving the availability of intelligent monitor system.
For addressing the above problem, the invention provides a kind of method of eliminating false alarm in supervisory system, comprising: the view data of alarm region is carried out to color space conversion, reduce the color dimension of described view data; To reducing the view data computing mode description vectors V of color dimension
a, as the quantizating index of Description Image feature; Calculate described state description vector V
aand distance between template vector, described template vector is the state description vector of the view data of alarm region under non-alarm condition; If described distance is less than default threshold value, determine that current state is the non-alarm condition that described template vector is corresponding, this time report to the police as false alarm.
Optionally, the view data of alarm region is carried out to color space conversion and comprise: by following conversion regime, the view data of alarm region is transformed into (r, g, b) color space from (R, G, B) color space,
r＝R/(R+G+B)
g＝G/(R+G+B)
b＝B/(R+G+B)。
Optionally, to reducing the view data computing mode description vectors V of color dimension
acomprise: select (r, or (g g), b) or (b, r) color space as normalization color space, the view data of described alarm region is carried out respectively to statistics with histogram at two components of selected normalization color space; Described histogrammic horizontal ordinate is intensity quantized level, and ordinate is the frequency that pixel occurs in this intensity; Generate the joint histogram of selected normalization color space; Generate the state description vector of above-mentioned joint histogram; The content of described state description vector is according to a class frequency value of intensity quantized level sequence.
Optionally, described template vector comprises the bright template vector V of lamp
onwith the lamp template vector V that goes out
off; Distance between the template vector that the described state description vector sum of described calculating is preserved comprises: calculate described state description vector V by following conversion regime
awith the bright template vector V of described lamp
onbetween distance B 1 and described state description vector V
awith the described lamp template vector V that goes out
offbetween distance B 2:
D1＝Va·Von/(|Va|×|Von|)
D2＝Va·Voff/(|Va|×|Voff|)。
Optionally, if being less than default threshold value, described distance comprises: if there is an alarm threshold value that is less than setting in D1 and D2.
Optionally, the method for upgrading described template vector comprises: input modeling image; Modeling image is carried out to color space conversion; Generate the state description vector of alarm region; According to template number, classification preservation state description vectors; If need to continue statistics, time delay a period of time, again input modeling image; If do not need to continue statistics, the state description vector of preserving based on classification, calculates average template vector.
Optionally, described template vector comprises the bright state description template vector of lamp and the lamp state description template vector that goes out; The method of calculating average template vector is that the state description vector to preserving carries out statistics with histogram in batches.
Optionally, in described elimination supervisory system, the method for false alarm also comprises: the standard deviation of calculating average template vector; The corresponding lamp of the large average template of the standard deviation vector state description template vector that goes out, the bright state description template vector of the average template corresponding lamp of vector that standard deviation is little.
Further, the present invention also provides a kind of device of eliminating false alarm in supervisory system, comprise: color space converting unit, the state description coupling with color space converting unit vector generation unit, the alarm decision unit coupling with state description vector generation unit; Described color space converting unit is suitable for receiving the view data of alarm region, and the view data of alarm region is carried out to color space conversion, reduces the color dimension of described view data; Described state description vector generation unit is suitable for generating the state description vector of described view data, as the quantizating index of Description Image feature; Described alarm decision unit is suitable for calculating distance between described state description vector sum template vector and judges according to the relation between described distance and default threshold value whether this time report to the police is false alarm; Described template vector is the state description vector of the view data of alarm region under non-alarm condition.
Optionally, in described elimination supervisory system, the device of false alarm also comprises template storage unit, couples with described alarm decision unit, is suitable for preserving described template vector.
Optionally, in described elimination supervisory system, the device of false alarm also comprises template vector updating block, couple with described state description vector generation unit and template storage unit respectively, be suitable for receiving and classify and preserve the state description vector of modeling image, calculate average template vector and upgrade the template vector that described template storage unit is preserved; Menu unit, couples with state description vector generation unit, is suitable for arranging the state flag bit of described device; Described state description vector generation unit is suitable for obtaining this state flag bit, and according to this state flag bit, state description vector is sent to the template vector updating block or the alarm decision unit that couple with it.
Further, the present invention also provides a kind of system of eliminating false alarm in supervisory system, comprising: the device of false alarm in above-mentioned elimination supervisory system, image acquisition units, pre-alarm unit and alarm unit; Described image acquisition units is suitable for gathering at set intervals the view data of the monitored object of a frame, and sends to pre-alarm unit; Described pre-alarm unit is suitable for calculating the variable quantity of front and back two frame image datas and judging whether to exceed threshold value; Described false alarm cancellation element is suitable for receiving pre-alarm signal and the alarm region view data from pre-alarm unit, and judges whether this pre-alarm signal is false alarm; Described alarm unit is suitable for receiving reports to the police from the alerting signal of false alarm cancellation element and to monitor staff.
Optionally, calculate the variable quantity of front and back two frame image datas and judge whether to exceed threshold value and comprise: described front and back two frame image datas are carried out respectively to gray processing processing; Obtain the difference value of described front and back two frame image datas; Whether poor score value exceedes default threshold value.
Compared with prior art, the present invention has the following advantages: (1) is by adopting above-mentioned false alarm removing method, eliminate in existing supervisory system due to the card inserting mouth pilot lamp reflective false alarm causing causing on side panel that glimmers, improved warning accuracy rate; (2) by upgrading template vector, eliminate the impact bringing according to conditional variation due to exterior light, further improve warning accuracy rate (3) by utilizing normalization color space, compress the redundant information of view data, not only can be on normal alarm generation impact, can reduce on the contrary the wrong report that illumination variation causes, further improve warning accuracy rate.
Brief description of the drawings
Fig. 1, the 2nd, in prior art, the schematic diagram of the image that in ATM intelligent monitor system, camera obtains;
Fig. 3 is the reflective method flow diagram that causes false alarm in the elimination intelligent monitor system providing in one embodiment of the invention;
Fig. 4 provides template vector update method process flow diagram in one embodiment of the invention;
Fig. 5 provides a kind of automatic classification to preserve the method flow diagram of template vector in one embodiment of the present of invention;
Fig. 6, the 7th, the joint histogram that in one embodiment of the invention, lamp goes out, the bright template vector of lamp is corresponding;
Fig. 8 is the false alarm method for removing of the renewable template vector that provides in one embodiment of the invention;
Fig. 9,10,11 is in different embodiments of the invention, eliminates the structural representation of the device of false alarm in supervisory system;
Figure 12 is the structural representation of the system of false alarm in the elimination supervisory system providing in one embodiment of the invention.
Embodiment
For above-mentioned purpose of the present invention, feature and advantage can more be become apparent, below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in detail.
In the present invention, image is transformed into normalization color space from rgb color space and refers to:
(1) by following conversion regime, the view data of alarm region is transformed into (r, g, b) color space from (R, G, B) color space:
r＝R/(R+G+B)
g＝G/(R+G+B)
b＝B/(R+G+B)。
Wherein, utilize any two normalization components, can generate the 3rd component, for example:
b＝1-r-g
(2) r, two components of g form normalization color space (r, g).
Can select any two components in rgb color space to carry out above-mentioned normalization calculating, can realize and reduce color dimension through normalization, reduce the object of redundant information.
In the present invention, " intensity quantized level " is for describing the component of normalization color space.For example, the span of r and g is evenly divided into 256 quantized levels, obtains (r0-r255), (g0-g255), each quantized level corresponds to an intensity quantized level.
In the present invention, " statistics with histogram " is for the component of the normalization color space of image.Histogrammic horizontal ordinate is intensity quantized level, and ordinate is the frequency that pixel occurs in this intensity.
In the present invention, " joint histogram " is for the normalization color space of image.For example, in normalization color space (r, g), the histogrammic horizontal ordinate of the histogram of r component and g component is together in series, obtains r, the joint histogram of g component.That is: 256 intensity quantized levels of the corresponding r of horizontal ordinate 0-255,256 intensity quantized levels of the corresponding g of horizontal ordinate 256-511, ordinate is the frequency that pixel occurs at corresponding intensity quantized level.R, g component combination together characterized the color statistical information of image.
In the present invention, " state description vector ", for the normalization color space of image, is the vector drawing according to joint histogram.Its content is according to a class frequency value of intensity quantized level sequence.State description vector (a0, a1, a2 ..., a511), have 512 elements, front 256 elements and be 1, rear 256 elements and be 1, that is:
For example, piece image comprises 1000 pixel datas, has 10 pixels to occur in r 0 intensity, and 20 pixels occur in r 1 intensity, ..., 15 pixels are in r 255 intensity appearance, and state description vector r component is (0.01,0.02, ..., 0.015), and be 1; Have 30 pixels to occur in g 0 intensity, 5 pixels occur in g1 intensity ..., 50 pixels occur in g 255 intensity, state description vector g component be (0.03,0.005 ..., 0.05), and be 1; The state description vector of this width image be (0.01,0.02 ..., 0.015,0.03,0.005 ..., 0.05).State description vector is the quantizating index for Description Image feature.
In the present invention, " the bright state description template vector of lamp " and " lamp go out state description template vector " is illustrated under the card inserting mouth pilot lamp " lamp is bright " and " lamp goes out " two states in ATM, the image of alarm region corresponding state description vector respectively.They are " reference vector "." template vector " is the abbreviation of " state description template vector ".
In order to eliminate the reflective false alarm causing in intelligent monitor system, the present invention proposes a kind of solution based on color joint histogram: first to image being carried out to color space conversion to reduce illumination effect, then obtain state description template vector by calculating color joint histogram, and on the basis of bulk statistics, use joint histogram standard deviation to determine template vector and the corresponding relation that lamp is bright, lamp goes out this two states; In the time of practical application, on-the-spot test picture is calculated to color joint histogram vector too, and carry out distance with corresponding state description template vector and relatively report to the police finally judging whether.
Set forth detail in the following description so that fully understand the present invention.But the present invention can be different from alternate manner described here and implements with multiple, and those skilled in the art can do similar popularization without prejudice to intension of the present invention in the situation that.Therefore the present invention is not subject to the restriction of following public embodiment.
Be an image 101 that camera obtains in ATM intelligent monitor system as shown in Figure 1, 2.The content that image 101 shows comprises ATM and wall around thereof.ATM comprises display screen 102, plug-in card district 104, card inserting mouth pilot lamp 103, and user input area.The card inserting mouth pilot lamp 103 that Fig. 1 shows is for extinguishing state, and the card inserting mouth pilot lamp showing in Fig. 2 is blink states.In Fig. 2, also shown the retroreflective regions 105 on ATM side panel when card inserting mouth pilot lamp 103 glimmers, problem to be solved by this invention is the caused false alarm of retroreflective regions 105 just, and retroreflective regions 105 is above-mentioned alarm region.
In the process of ATM intelligent monitor system work, judge whether to report to the police by the variation of two two field pictures before and after comparing.First supervisory system generally obtains colored monitoring image, and colored monitoring image is carried out to gray processing processing, uses various gray processing formula to carry out; Then two two field pictures are carried out to grey scale difference processing, if just think when the absolute value of difference exceedes threshold value and produced enough variations, report to the police thereby cause, and transmitting, monitoring image identify the alarm region in monitoring image.Described threshold value is empirical value.
For fear of false alarm, in the present invention, the image of alarm region is further analyzed.
As shown in Figure 3, in the elimination intelligent monitor system providing in one embodiment of the invention, the reflective method that causes false alarm comprises the steps:
S101, carries out color space conversion to alarm region;
S102, the state description vector of generation alarm region;
S103, the distance between computing mode description vectors and template vector;
S104, if distance is less than threshold value, current state is state corresponding to reference vector, is false alarm;
S105, if distance is not less than threshold value, the state that the non-reference vector of current state is corresponding, is true warning.
In step S101, in order to reduce illumination effect, by following conversion regime, the image of alarm region is transformed into (r, g, b) color space from (R, G, B) color space:
r＝R/(R+G+B)
g＝G/(R+G+B)
b＝B/(R+G+B)
Three components in (r, g, b) color space and be 1, wherein having one-component is redundancy component.In the present embodiment, select (r, g) color space as normalization color space, reduced illumination redundant information.In other embodiments, also can select (g, b) or (b, r) as normalization color space.
For monitoring image, (R, G, B) color space has comprised a large amount of illumination redundant informations, has compressed this part redundant information after normalization, and the chromatic characteristic of more concentrated reflection own is favourable to follow-up further processing.Not only can, on normal alarm generation impact, can not reduce on the contrary the wrong report that illumination variation causes.
Color space conversion regime has multiple, in other embodiments of the invention, can also be by the image of alarm region from (R, G, B) color space is transformed into the color spaces such as YUV, HSV, LAB, CMYK, reach the object of brightness, chrominance separation, but after conversion, arbitrary component can not be derived by other 2 component, their dimensionality reduction is realized by abandoning luminance component.In above-described embodiment, selecting the image of alarm region to be transformed into (r, g, b) color space from (R, G, B) color space, is mainly to have considered that its conversion is simple, dimensionality reduction advantage easily.
In step S102, the process that generates the state description vector of alarm region is:
S201, carries out respectively statistics with histogram to alarm region image at r, the g component of (r, g) color space;
S202, the joint histogram of generation r, g component;
S203, generates the state description vector V of above-mentioned joint histogram
a.
In step S103, the distance between computing mode description vectors and template vector comprises:
Calculate described state description vector V
awith the bright template vector V of described lamp
onbetween distance B 1;
Calculate described state description vector V
awith the described lamp template vector V that goes out
offbetween distance B 2.
The bright template vector V of described lamp
onwith the lamp template vector V that goes out
offimplication be: under the bright state of card inserting mouth pilot lamp and under the state that goes out of card inserting mouth pilot lamp, calculate respectively the state description vector of alarm region view data; Its computing method are also to utilize the step in S201～S203 to complete, and are saved as reference vector is for comparing.
The method of calculating the distance between two state description vectors is:
D1＝Va·Von/(|Va|×|Von|)
D2＝Va·Voff/(|Va|×|Voff|)
Utilize the included angle cosine distance between vector, judge two similarity distances between vector.
If there is an alarm threshold value T0 (empirical value) who is less than setting in D1 and D2, the current state of card inserting mouth pilot lamp or for the bright state of lamp or be the lamp state that goes out, the state of corresponding alarm region or for the bright state of lamp or be the lamp state that goes out, this warning is false alarm, can get rid of.
If D1 and D2 are greater than alarm threshold value T0, represent that the state of current alarm region is neither the bright state of lamp, the state that neither lamp goes out, this warning is true warning.
By above-mentioned method, get rid of the false alarm that retroreflective regions causes, improve the warning accuracy of intelligent monitor system, save time and relevant resource.
Further, because outside illumination condition may change, so the state description template vector of setting up comprises the bright template vector of lamp and the lamp template vector that goes out, just need at set intervals to upgrade, better to avoid the appearance of false alarm.
As shown in Figure 4, in one embodiment of the present of invention, provide a kind of template renewal method, comprising:
S300, arranges zone of alarm;
S301, input modeling image (setting up template vector image), is the view data of zone of alarm;
S302, carries out color space conversion to alarm region;
S303, the state description vector of generation alarm region;
S304, classification preservation state description vectors;
S305, if need to continue statistics, time delay a period of time, gathers and inputs modeling image again;
S306, if do not need to continue statistics, calculates average template vector.
In step S300, the object that zone of alarm is set is that instruction may produce the scope of false alarm, within the scope of this, normal state is incessantly a kind of, and before and after supervisory system adopts, the method for two two field picture grey scale differences or the alarming result that other comparative approach obtain likely belong to false alarm.In the present embodiment, alarm region is above-mentioned retroreflective regions.
In step S301, modeling image is the image of alarm region.In other embodiments of the invention, if adopt whole monitoring image, step S300 can cancel, and increases an operation steps of choosing alarm region in step S301.
Step S302, S303 are corresponding respectively with above-mentioned step S101, S102.
In step S304, the state description vector classification obtaining is kept in template storage unit.In the present embodiment, template vector has two classes, and a class is the bright template of lamp, and another kind of is the lamp template of going out.Classification thresholds T1 is for judging that newly-generated state description vector belongs to the template vector of which kind of type.In the present embodiment, provide a kind of method of template type of automatic decision state description vector, with reference to figure 5, comprising:
S401, if the state description vector generating for the first time saves as type i;
S402, if not, input vector and type i template vector distance calculated;
S403, if distance is less than classification thresholds T1, saves as type i template vector sample;
S404, if distance is greater than classification thresholds T1, saves as Type II template vector sample.
The method of the distance between above-mentioned compute vector still adopts the included angle cosine range formula between vector:
D＝a·b/(|a|×|b|)
Obtaining after some samples, in step S306, the sample of type i and Type II template vector is carried out respectively to statistics with histogram in batches and obtain type i and Type II average template vector.
Concrete, being located at the picture that certain time period has obtained a collection of type i, these pictures amount to and include 10000 pixels, by statistics, wherein have 100 pixels to occur in r 0 intensity, 200 pixels occur in r1 intensity ..., 150 pixels occur in r 255 intensity, average template vector r component be (0.01,0.02 ..., 0.015), and be 1; Have 300 pixels to occur in g 0 intensity, 50 pixels occur in g1 intensity ..., 500 pixels occur in g 255 intensity, average template vector g component be (0.03,0.005 ..., 0.05), and be 1; The average template vector of the picture of whole type i be (0.01,0.02 ..., 0.015,0.03,0.005 ..., 0.05).
The average template vector of two types that said method obtains will be served as two types of template vectors, compare with the state description vector of image to be detected.
In the present embodiment, for the go out corresponding relation of template of clear and definite these two types of template vectors and the bright template of lamp, lamp, can be by operating personnel's manual configuration.
Further, in one embodiment of the invention, by the standard deviation of compute type I and Type II template vector, the bright template of automatic corresponding lamp, the lamp template of going out.Its principle is: when lamp goes out, color background is often more single, and lamp bright after, in region, both had powerful connections, also had light, so color distribution can be relatively broad, the template of type i and Type II is calculated respectively to standard deviation, the corresponding lamp that standard deviation the is large state that goes out, the bright state of corresponding lamp that standard deviation is little.As shown in Figure 6, Figure 7, in Fig. 6, the standard deviation of (r-g) joint histogram is 0.0194, the indication lamp template of going out; In Fig. 7, the standard deviation of (r-g) joint histogram is 0.0077, the bright template of indication lamp.
Further, above-mentioned false alarm method for removing and template vector update method can combine, and become the false alarm method for removing that upgrades template vector.As shown in Figure 8, in one embodiment of the present of invention, provide a kind of false alarm method for removing, comprising:
S501, input modeling image; Or S502, input test image;
S503, carries out color space conversion to alarm region;
S504, the state description vector of generation alarm region;
S505, if not test pattern, classification preservation state description vectors;
S506, if be test pattern at present, no longer needs to continue to gather statistical sample, calculates average template vector;
S507, the average template obtaining with S506 vector substitutes the template vector of preserving;
S508, if test pattern calls the template vector of preservation, the distance of calculation template vector sum state description vector also judges whether it is less than alarm threshold value;
S509, according to the comparative result of S508, judges whether to report to the police; If distance is less than threshold value, current state is state corresponding to reference vector, is false alarm.
Further, although the method that the above embodiments are described for the alarm region that has two kinds of normal conditions, method provided by the invention also can have multiple normal condition for monitored object, whether need further to analyze alarm region be in true system of reporting to the police.When application, only various normal conditions all need to be set up respectively to corresponding template vector, obtain after alarm region, calculate respectively and judge whether the state description vector of alarm region can be summed up as template vector (as utilized included angle cosine formula between vector), if can be summed up as template vector, this is reported to the police as false alarm, can get rid of.
Further, the present invention also provides a kind of device of eliminating false alarm in supervisory system.As shown in Figure 9, in one embodiment of the invention, eliminate the structural representation of the device of false alarm in supervisory system.Comprising color space converting unit 901, the state description vector generation unit 902 coupling with color space converting unit 901, the alarm decision unit 903 coupling with state description vector generation unit 902.Color space converting unit 901 is responsible for the image of the alarm region receiving to be transformed into (r, g, b) color space from (R, G, B) color space.Conversion formula is described in step S101; State description vector generation unit 902 is responsible for generating the state description vector of alarm region, generates the method for state description vector as described in step S102; Alarm decision unit 903 is responsible for the distance between computing mode description vectors and template vector and is judged the relation of this distance and the threshold value of setting, as described in step S103, S104 and S105.
Of the present invention another executed in example, and as shown in figure 10, the device of eliminating false alarm in supervisory system also comprises template storage unit 904, couples with alarm decision unit 903, is responsible for preserving template vector.In other embodiments, the function of preservation template vector also can be completed by alarm decision unit 903.
Further, for new template more, in one embodiment of the invention, provide the another kind of device of eliminating false alarm in supervisory system.As shown in figure 11, this device comprises: image input block 1001, for receiving modeling image or test pattern; Color space converting unit 1002; State description vector generation unit 1003; Template vector updating block 1004, the preservation state description vectors that is suitable for classifying, finally calculates average template vector and upgrades the template vector in template storage unit 1005; The menu unit 1007 and the alarm decision unit 1006 that couple with state description vector generation unit 1003.
In the present embodiment, the data flow of eliminating the device of false alarm in supervisory system depends on the setting of menu unit 1007.Menu unit 1007 receives user's selection, and state flag bit Flag is set; State flag bit Flag can be set to 0 or 1, and corresponding templates is upgraded function and false alarm recognition function respectively.
In the time that template renewal function is chosen, device receives data stream 1011 for modeling image, and state description vector generation unit 1003 obtains the state flag bit Flag in menu unit 1007, and value is 0.State description vector generation unit 1003 sends to state description vector the template vector updating block 1004 coupling with it.Complete template renewal by template vector updating block 1004 and operate, content of operation is as described in step S304, S305 and S306.
In the time that false alarm recognition function is chosen, device receives data stream 1012 for test pattern, i.e. alarm region image, and state description vector generation unit 1003 obtains the state flag bit Flag in menu unit 1007, and value is 1.State description vector generation unit 1003 sends to state description vector the alarm decision unit 1006 coupling with it.Alarm decision unit 1006 obtains the bright template of lamp and the lamp template of going out from template storage unit 1005, and the distance between computing mode description vectors and two templates, determines whether false alarm.
In other embodiments of the invention, the also state flag bit of setting device otherwise, or the function of finishing device is switched in other way.For example, image input block receives user's function and selects, and increases state flag bit to all view data, and the state flag bit that view data is entrained is depended in function switching.
Further, as shown in figure 12, in one embodiment of the invention, also provide a kind of system of eliminating false alarm in supervisory system.This system comprises: image acquisition units 1101, pre-alarm unit 1102, false alarm cancellation element 1103 and alarm unit 1104.
Image acquisition units 1101 can be camera or other capture apparatus, for gathering at set intervals the view data of the monitored object of a frame.
Pre-alarm unit 1102 is connected with image acquisition units 1101, receives and judges whether front and back two frame image datas change larger.In the present embodiment, the color image data that camera obtains has carried out gray processing processing in pre-alarm unit 1102, and processing mode is:
Gray(i，j)＝0.299*R(i，j)+0.587*G(i，j)+0.114*B(i，j)
Wherein i, j is horizontal coordinate, and RGB is color component, and Gray is gray-scale value.
Relatively, whether the difference of the gray-scale value of two each pixels of frame exceedes the first predetermined value.If the pixel quantity exceeding reaches the second predetermined value, generate pre-alarm signal.Meanwhile, alarm region (comprising the pixel that gray-value variation exceedes the first predetermined value) is extracted.Wherein, the first predetermined value and the second predetermined value are all empirical values.
In other embodiments of the invention, also can carry out difference to front and back two frame image datas in other way, compare obtaining difference value and default threshold value, then judge whether to generate pre-alarm signal.
False alarm cancellation element 1103 receives from the pre-alarm signal of pre-alarm unit 1102 and alarm region view data.Adopt the method for above-mentioned elimination false alarm, get rid of false alarm wherein, and real pre-alarm signal is sent to alarm unit 1104.
Alarm unit 1104 can be the light-emitting devices such as sound-producing device or warning lamp such as alarm bell, for reporting to the police to monitoring of tools personnel.
By adopting above-mentioned false alarm removing method, Apparatus and system, eliminate in existing supervisory system due to the card inserting mouth pilot lamp reflective false alarm causing causing on side panel that glimmers, improve warning accuracy rate.
Although the present invention with specific embodiment openly as above; but it is not for limiting the present invention; any those skilled in the art without departing from the spirit and scope of the present invention; can utilize method and the technology contents of above-mentioned announcement to make possible variation and amendment to technical solution of the present invention; therefore; every content that does not depart from technical solution of the present invention; any simple modification, equivalent variations and the modification above embodiment done according to technical spirit of the present invention, all belong to the protection domain of technical solution of the present invention.
Claims (8)
1. a method of eliminating false alarm in supervisory system, is characterized in that, comprising:
The view data of alarm region is carried out to color space conversion, reduce the color dimension of described view data; Describedly the view data of alarm region is carried out to color space conversion comprise: by following conversion regime, the view data of alarm region is transformed into (r, g, b) color space from (R, G, B) color space,
r＝R/(R+G+B)
g＝G/(R+G+B)
b＝B/(R+G+B)；
To reducing the view data computing mode description vectors V of color dimension
a, as the quantizating index of Description Image feature; Described to reducing the view data computing mode description vectors V of color dimension
acomprise: select (r, or (g g), b) or (b, r) color space as normalization color space, the view data of described alarm region is carried out respectively to statistics with histogram at two components of selected normalization color space; Described histogrammic horizontal ordinate is intensity quantized level, and ordinate is the frequency that pixel occurs in this intensity; Generate the joint histogram of selected normalization color space; Generate the state description vector of above-mentioned joint histogram; The content of described state description vector is according to a class frequency value of intensity quantized level sequence;
Calculate described state description vector V
aand distance between template vector, described template vector is the state description vector of the view data of alarm region under non-alarm condition;
If described distance is less than default threshold value, determine that current state is the non-alarm condition that described template vector is corresponding, this time report to the police as false alarm.
2. the method for false alarm in elimination supervisory system as claimed in claim 1, is characterized in that:
Described template vector comprises the bright template vector V of lamp
onwith the lamp template vector V that goes out
off;
Distance between the template vector that the described state description vector sum of described calculating is preserved comprises:
Calculate described state description vector V by following conversion regime
awith the bright template vector V of described lamp
onbetween distance B 1 and described state description vector V
awith the described lamp template vector V that goes out
offbetween distance B 2:
D1＝V
a·V
on/(|V
a|×|V
on|)
D2＝V
a·V
off/(|V
a|×|V
off|)。
3. the method for false alarm in elimination supervisory system as claimed in claim 2, is characterized in that, comprises if described distance is less than default threshold value:
If there is an alarm threshold value that is less than setting in D1 and D2.
4. the method for false alarm in elimination supervisory system as claimed in claim 1, is characterized in that, also comprises and upgrades described template vector; The described template vector of described renewal comprises:
Input modeling image;
Modeling image is carried out to color space conversion;
Generate the state description vector of alarm region;
According to template number, classification preservation state description vectors;
If need to continue statistics, time delay a period of time, again input modeling image;
If do not need to continue statistics, the state description vector of preserving based on classification, calculates average template vector.
5. the method for false alarm in elimination supervisory system as claimed in claim 4, is characterized in that:
Described template vector comprises the bright state description template vector of lamp and the lamp state description template vector that goes out;
The method of calculating average template vector is that the state description vector to preserving carries out statistics with histogram in batches.
6. the method for false alarm in elimination supervisory system as claimed in claim 5, is characterized in that, also comprises:
Calculate the standard deviation of average template vector; The corresponding lamp of the large average template of the standard deviation vector state description template vector that goes out, the bright state description template vector of the average template corresponding lamp of vector that standard deviation is little.
7. a device of eliminating false alarm in supervisory system, is characterized in that, comprising: color space converting unit, state description vector generation unit, alarm decision unit;
Described color space converting unit is suitable for the view data of alarm region to carry out color space conversion, reduces the color dimension of described view data; Describedly the view data of alarm region is carried out to color space conversion comprise: by following conversion regime, the view data of alarm region is transformed into (r, g, b) color space from (R, G, B) color space,
r＝R/(R+G+B)
g＝G/(R+G+B)
b＝B/(R+G+B)；
Described state description vector generation unit is suitable for the view data computing mode description vectors V to reducing color dimension
a, as the quantizating index of Description Image feature; Described to reducing the state description vector V of color dimension
acomprise: select (r, or (g g), b) or (b, r) color space as normalization color space, the view data of described alarm region is carried out respectively to statistics with histogram at two components of selected normalization color space; Described histogrammic horizontal ordinate is intensity quantized level, and ordinate is the frequency that pixel occurs in this intensity; Generate the joint histogram of selected normalization color space; Generate the state description vector of above-mentioned joint histogram; The content of described state description vector is according to a class frequency value of intensity quantized level sequence;
Described alarm decision unit is suitable for calculating described state description vector V
aand distance between template vector, if described distance is less than default threshold value, determine that current state is the non-alarm condition that described template vector is corresponding, this time report to the police as false alarm; Described template vector is the state description vector of the view data of alarm region under non-alarm condition.
8. the device of false alarm in elimination supervisory system as claimed in claim 7, is characterized in that, also comprises:
Template vector updating block, is suitable for upgrading template vector; Described renewal template vector comprises: input modeling image; Modeling image is carried out to color space conversion; Generate the state description vector of alarm region; According to template number, classification preservation state description vectors; If need to continue statistics, time delay a period of time, again input modeling image; If do not need to continue statistics, the state description vector of preserving based on classification, calculates average template vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010565898.9A CN102479416B (en) | 2010-11-29 | 2010-11-29 | Method and device for eliminating false alarm in monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010565898.9A CN102479416B (en) | 2010-11-29 | 2010-11-29 | Method and device for eliminating false alarm in monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102479416A CN102479416A (en) | 2012-05-30 |
| CN102479416B true CN102479416B (en) | 2014-08-13 |
Family
ID=46092042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010565898.9A Expired - Fee Related CN102479416B (en) | 2010-11-29 | 2010-11-29 | Method and device for eliminating false alarm in monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102479416B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104700532B (en) * | 2013-12-11 | 2018-03-09 | 杭州海康威视数字技术股份有限公司 | A kind of video alarm method and apparatus |
| EP2897112B1 (en) * | 2014-01-17 | 2019-03-06 | Wincor Nixdorf International GmbH | Method and apparatus for the prevention of false alarms in monitoring systems |
| CN104320388A (en) * | 2014-10-13 | 2015-01-28 | 杭州海兴电力科技股份有限公司 | S-FSK carrier based active alarming method |
| CN105336074A (en) | 2015-10-28 | 2016-02-17 | 小米科技有限责任公司 | Alarm method and device |
| CN108447208B (en) * | 2018-05-17 | 2019-10-29 | 普联技术有限公司 | Light warning method, equipment, web camera and storage medium |
| CN113628404A (en) * | 2020-05-08 | 2021-11-09 | 华为技术有限公司 | Method and device for reducing invalid alarm |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101808234A (en) * | 2010-02-09 | 2010-08-18 | 成都市华为赛门铁克科技有限公司 | Monitor processing method and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010176533A (en) * | 2009-01-30 | 2010-08-12 | Secom Co Ltd | Transaction monitoring device |
-
2010
- 2010-11-29 CN CN201010565898.9A patent/CN102479416B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101808234A (en) * | 2010-02-09 | 2010-08-18 | 成都市华为赛门铁克科技有限公司 | Monitor processing method and system |
Non-Patent Citations (1)
| Title |
|---|
| JP特开2010-176533A 2010.08.12 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102479416A (en) | 2012-05-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102479416B (en) | Method and device for eliminating false alarm in monitoring system | |
| CN103714325B (en) | Left object and lost object real-time detection method based on embedded system | |
| CN100579174C (en) | Motion detection method and device | |
| CN106897720A (en) | A kind of firework detecting method and device based on video analysis | |
| CN104966304A (en) | Kalman filtering and nonparametric background model-based multi-target detection tracking method | |
| KR20180037164A (en) | Monitoring system for automatically selecting cctv, monitoring managing server for automatically selecting cctv and managing method thereof | |
| CN102892007B (en) | Promote the synchronous method and system with obtaining tracking of color balance between multiple video camera | |
| CN102176758A (en) | Video quality diagnosis system and realization method thereof | |
| US20060221181A1 (en) | Video ghost detection by outline | |
| CN110516529A (en) | It is a kind of that detection method and system are fed based on deep learning image procossing | |
| CN111222478A (en) | Construction site safety protection detection method and system | |
| CN105187800A (en) | Detector with high definition module structure | |
| CN105335701A (en) | Pedestrian detection method based on HOG and D-S evidence theory multi-information fusion | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| JP7429756B2 (en) | Image processing method, device, electronic device, storage medium and computer program | |
| CN113706523A (en) | Method for monitoring belt deviation and abnormal operation state based on artificial intelligence technology | |
| CN110348343A (en) | A kind of act of violence monitoring method, device, storage medium and terminal device | |
| CN110636281A (en) | Real-time monitoring camera shielding detection method based on background model | |
| CN111460917B (en) | Airport abnormal behavior detection system and method based on multi-mode information fusion | |
| EP4191521A1 (en) | Machine learning device and image processing device | |
| CN113723229A (en) | Signal lamp detection method and device and computer readable storage medium | |
| CN109389132A (en) | A kind of mist Concentration Testing method for early warning and system based on image | |
| CN205378015U (en) | Detecting device with two cameras | |
| US20230048649A1 (en) | Method of processing image, electronic device, and medium | |
| CN111145219B (en) | Efficient video moving target detection method based on Codebook principle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140813 Termination date: 20191129 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |