CN102163360B - Tunnel smoke video detecting method and device - Google Patents

Abstract

The invention relates to the field of video monitoring, and discloses a tunnel smoke video detecting method and device. Each detecting window is attached with a long time window and a short time window which are used for storing former L values of a feature value. When video detection is carried out on the tunnel smoke, a variance of an image gray-scale computed according to an image in the detecting window is used as a feature value to be added in the long time window and short time window of the detecting window so as to obtain the summation of the mean values of the time windows of the long time window and the summation of the mean values of the time windows of the short time window. If the summation of mean values of the time windows of the long time window is not less than the multiplication of the summation of mean values of the time windows of the short time window and S, a smoke alarming signal is triggered. The change of the video image caused by the smoke can be effectively reflected by obtaining the change of the image brightness variance in the detecting window, therefore, the accuracy and flexibility of the triggering to the smoke alarming signal can be guaranteed.

Description

The video detecting method of tunnel smog and device thereof

Technical field

The present invention relates to field of video monitoring, particularly the Smoke Detection technology in the field of video monitoring.

Background technology

Freeway management department need report to the police to the smog episode in the tunnel for reasons such as traffic safety, and the intensity of a fire is very big when finding naked light usually, just can win the more disposal time if can provide to report to the police at the smog stage of development.In addition, fill the air dense smoke in the tunnel and can cause visibility to reduce, cause the generation of traffic hazard.Therefore can in time trigger smog alarm has great safe meaning.

Existing scheme is normally disposed smoke alarm by the monitoring point in the tunnel.As publication number is in the patent of CN1133461A, has utilized light scattering principle that smog is reported to the police.But this class smoke-detectors distributes by point, is difficult to hold on-the-spot macroscopical situation, need analyze in conjunction with the on-site supervision video.Yet, arrange personnel's complete monitoring video to be difficult to carry out, the one, because number of videos is numerous, need to arrange sufficient personnel could cover all videos; On the other hand, monitor staff's notice is difficult to keep for a long time high concentration, and smog episode is rare relatively in addition, and the monitor staff loses vigilance easily.

Therefore, this class monitoring can only be finished by the method for automatic analysis video.And existing disclosed video smoke detection method, as brightness/color mutation analysis (is the patent of CN1331823A referring to publication number), motion analysis (referring to Yuan Feiniu. (in April, 2008). based on the video smoke detection method of semi-invariant and direction of primary motion. Chinese image graphics journal) and carry out Smoke Detection smog detection methods such as (are the patent of CN101751558A referring to publication number) in conjunction with features such as contrast and smoothnesses with background modeling, when specifically being applied to the tunnel video analysis, because light changes frequent and complexity in the tunnel, car light, illuminating lamp even vehicle change by all causing brightness; The method of motion analysis also is easy to lose efficacy.And background modeling method can not be removed the vehicle movement interference effectively and the light frequent variations is blocked, and is difficult to obtain effectively " stable prime area ", is unfavorable for the subsequent smoke signature analysis.Thereby cause the detection effect of existing video smoke detection method not fully up to expectations, be difficult to guarantee the triggering accuracy of smog alert signal.

Summary of the invention

The object of the present invention is to provide a kind of video detecting method and device thereof of tunnel smog, to guarantee the triggering accuracy and the sensitivity of smog alert signal.

For solving the problems of the technologies described above, embodiments of the present invention provide a kind of video detecting method of tunnel smog, comprise following steps:

For each detection window in the video image is provided with window when long and short window, the window length of window is L when long in advance _Long, unit is a frame, is used to preserve L _LongThe eigenwert of image in the detection window of individual frame, the window length of short window is L _Short, unit is a frame, is used to preserve L _ShortThe eigenwert of image in the detection window of individual frame, eigenwert is the variance according to the gradation of image of image calculation in the detection window, L _LongGreater than L _Short

When carrying out the Video Detection of tunnel smog, the variance of the gradation of image of each detection window of calculating present frame;

When the variance of the gradation of image of the detection window calculated is joined this detection window long in window and the short window, upgrade the time window average of each window and each short window when long;

Calculate the time window average sum of each window when long, and calculate the time window average sum of each short window;

If when long window the time window average sum 〉=short window the time window average sum and S product, then trigger the smog alert signal, wherein, S is Smoke Detection sensitivity, the value of S preestablishes.

Embodiments of the present invention also provide a kind of video detecting device of tunnel smog, comprise:

The time window module is set, be used to each detection window in the video image that window when long and short window are set, the window length of window is L when long _Long, unit is a frame, is used to preserve L _LongThe eigenwert of image in the detection window of individual frame, the window length of short window is L _Short, unit is a frame, is used to preserve L _ShortThe eigenwert of image in the detection window of individual frame, eigenwert is the variance according to the gradation of image of image calculation in the detection window, L _LongGreater than L _Short

The variance computing module is used for when carrying out the Video Detection of tunnel smog, the variance of the gradation of image of each detection window of calculating present frame;

The time window average update module, window and short window when the variance that is used for the gradation of image of detection window that the variance computing module is calculated joins this detection window long are upgraded the time window average of each window and each short window when long;

The time window average and computing module, be used to calculate the time window average sum of each window when long, and the time window average sum of each short window;

Detection module is used to judge whether to meet the following conditions: when long window the time window average sum 〉=short window the time window average sum and S product, and when judgement satisfies condition, trigger the smog alert signal, wherein, S is Smoke Detection sensitivity, and the value of S preestablishes.

Embodiment of the present invention compared with prior art, the key distinction and effect thereof are:

Subsidiary window when long of each detection window and short window, in order to preserve preceding L the value (numerical value of L is identical with window length) of eigenwert, this eigenwert is the variance according to the gradation of image of image calculation in the detection window.When carrying out the Video Detection of tunnel smog, variance with the gradation of image of each detection window of the present frame that calculates, when joining this detection window long in window and the short window, upgrade the time window average of each window and each short window when long, obtain the time window average sum of each window when long, with the time window average sum of each short window.If when long window the time window average sum 〉=short window the time window average sum and S product, then trigger the smog alert signal, wherein, S is Smoke Detection sensitivity, the value of S preestablishes.Because the time window average sum of window when long, time window average sum with short window, the brightness of image variance that is actually in the detection window of reflection changes, therefore change by the brightness of image variance of obtaining in the detection window, the variation that smog brings video image can be reflected effectively, thereby the triggering accuracy and the sensitivity of smog alert signal can be guaranteed.And, owing to can preestablish the value of Smoke Detection sensitivity S, therefore can effectively distinguish the concentration of smog.

Further, detection window is evenly arranged in the top of video image.By at the video image top detection window being set, the interference that can avoid vehicle movement to bring.Because smog can spread to tunnel top when rising, and detection window is arranged on the validity that the top can not have influence on algorithm, has also reduced operand simultaneously.

Further, calculate each when long window the time window average sum before, can be earlier with the time window average window during less than Lower Threshold or greater than Upper threshold long, be labeled as invalid window when long, with the time window average window more than or equal to Lower Threshold and when being less than or equal to Upper threshold long, be labeled as effective window when long.Calculate each when long window the time during window average sum, calculate the time window average sum of window when effectively long.Whether the window average judges the validity of this detection window, the wrong report that can effectively avoid the unusual fluctuation of indivedual detection window feature to bring during by analyzing and testing window long at threshold range.

Further, before triggering the smog alert signal, if whether the interval time that can also judge earlier current time and the moment of last triggering smog alert signal greater than preset duration, then enters the step of triggering smog alert signal greater than presetting duration again.By analyzing the interval that smoke characteristics occurs, can avoid repeatedly reporting to the police with a smog episode.

Further, before the step that triggers the smog alert signal, whether the high light area that can also first judge the track is less than the 1/N of track area, and N is predefined value, during less than the 1/N of track area, advance to trigger the smog alert signal at the high light area of judging the track again.By comparing high-brightness region ratio in the track, the wrong report that can avoid car light direct projection camera to cause.

Description of drawings

Fig. 1 is according to the imaging pattern synoptic diagram that is installed in the top, tunnel in the first embodiment of the invention;

Fig. 2 is according to the imaging pattern synoptic diagram that is installed in tunnel-side in the first embodiment of the invention;

Fig. 3 is the video detecting method process flow diagram according to the first embodiment of the invention tunnel smog;

Fig. 4 is according to the position view in video image of detection window in the first embodiment of the invention;

Fig. 5 is the video detecting method process flow diagram according to the second embodiment of the invention tunnel smog;

Fig. 6 is the video detecting device structural representation according to the big tunnel smog of third embodiment of the invention.

Embodiment

In the following description, in order to make the reader understand the application better many ins and outs have been proposed.But, persons of ordinary skill in the art may appreciate that even without these ins and outs with based on the many variations and the modification of following each embodiment, also can realize each claim of the application technical scheme required for protection.

For making the purpose, technical solutions and advantages of the present invention clearer, embodiments of the present invention are described in further detail below in conjunction with accompanying drawing.

First embodiment of the invention relates to a kind of video detecting method of tunnel smog.In the present embodiment, can the monitor video image that obtain be passed to back-end analysis equipment (being video analysis equipment) by digital network or video signal cable by the video monitoring video camera, the frame of video that monitors is carried out the Video Detection of tunnel smog by video analysis equipment, this moment, transmission equipment required to have less time-delay and antijamming capability, in time intactly was transferred to back-end analysis equipment to guarantee the front end vision signal.Back-end analysis equipment is one the reliable and stable computing equipment than the epistasis energy, includes but not limited to server, embedded computer etc.; Also computing equipment can be integrated in the video monitoring video camera, the frame of video that monitors be carried out the Video Detection of tunnel smog by video camera.The video monitoring video camera can be installed in tunnel buttress inwall or sidewall higher position, pays the utmost attention to the traffic monitoring dedicated video camera of functions such as high light inhibition.The camera imaging effect that is installed in tunnel top as shown in Figure 1, the camera imaging effect that is installed in tunnel-side is as shown in Figure 2.

The idiographic flow of the video detecting method of tunnel smog as shown in Figure 3, in step 301, for each detection window in the video image is provided with window when long and short window, the window length of window is L when long in advance _Long, unit is a frame, is used to preserve L _LongThe eigenwert of image in the detection window of individual frame, the window length of short window is L _Short, unit is a frame, is used to preserve L _ShortThe eigenwert of image in the detection window of individual frame, the eigenwert in the present embodiment is the variance according to the gradation of image of image calculation in the detection window, L _LongGreater than L _Short

Specifically, detection window can be evenly arranged in the tip position of video image, as detection window at evenly distributed 8 squares of the tip position of video image, arrangement mode as shown in Figure 4,1～8 points out each Smoke Detection window position in video pictures respectively.Need guarantee when camera is installed has high-contrast object clearly in the detection window more than 1, as light, markings, warning sign etc.Subsidiary window when long of each detection window when eigenwert statistics (when long window) and a short window (short-time characteristic primary system timing window) are in order to L value before the preservation eigenwert.L is a window length.The window length computing formula is L=f * t: be the frame per second of this video, unit is the frame per second, the time span of window when t is, unit is second, the unit of window length is a frame.Short window time span t _ShortWindow time span t during≤length _Long, window time span 30s≤t when long usually _Long≤ 100s, short window time span 3s≤t _Short≤ 10s, t _Short≤ t _LongAccording to the time time span of window and above-mentioned formula can calculate the window length L of window when long respectively _LongWindow length L with short window _Short

When need carry out the Video Detection of tunnel smog, enter step 302, import current video frame images.

Then, in step 303, select the next not detection window of traversal.

Then, in step 304, calculate the variance s of gradation of image in the detection window ²Particularly, can pass through computing formula

Calculate s ²Y wherein _iBe the gray-scale value of Smoke Detection window interior pixel point, n is the quantity of pixel.

Then, in step 305, when the variance of the gradation of image of the detection window calculated is joined this detection window long in window and the short window, the average of window and short window when upgrading this detection window long.Because that window is preserved when long is L _LongThe s of individual frame ², that short window is preserved is L _ShortThe s of individual frame ², that is to say that each value when long in window/short window in fact all is the s of a certain frame in this detection window ²Therefore, with the s of current calculating ²When joining this detection window long in window and the short window, the time window average (mean value of all values in the instant window) of window and short window during long after can obtaining upgrading.

Then, in step 306, judge window when long the time window average whether in effective range, such as, judge window when long the time window average whether more than or equal to Lower Threshold and be less than or equal to Upper threshold, if judge window when long the time window average more than or equal to Lower Threshold and be less than or equal to Upper threshold, then assert window when long the time window average in effective range, enter step 307, window was labeled as effective window when long in the time of should growing; Otherwise (when promptly long window the time window average less than Lower Threshold or greater than Upper threshold), assert window when long the time window average not in effective range, enter step 308, window is labeled as invalid window when long during with this length.Effective range can be chosen to be 1000 to 3000 usually, and promptly the Xiamen is limited to 1000, is limited to 3000 to the doorstep.In addition, be appreciated that Lower Threshold and Upper threshold also can be set to other numerical value, do not exemplify one by one at this.

In step 309, judge whether detection window has traveled through to finish, be then to enter step 310 if be judged to be; If be judged to be not, then get back to step 303, continue to select the next not detection window of traversal.

This shows, to step 309, can (be s by step 303 with the eigenwert of each detection window ²) when joining this detection window long in window and the short window, the time window average of window and each short window when each after obtaining upgrading is long.

Then, in step 310, calculate the time window average sum s of window when effectively long _LongAnd the time window average sum s of all short window _Short

Then, in step 311, judge whether to satisfy: s _Long〉=S _Short* S, wherein, S is the Smoke Detection sensitivity of definition, and the value of S preestablishes, and span can be decided to be [2,4].That is to say, judge the time window average sum S of short window _ShortThe time window average sum s of window when long relatively _LongThe certain proportion (inverse of decline ratio is the Smoke Detection sensitivity S) that whether descended is if judge the s that satisfies condition _Long〉=s _Short* S then enters step 312, triggers the smog alert signal; If do not satisfy condition s _Long〉=S _Short* S, then the processing of current video frame finishes, and gets back to step 302.

Be not difficult to find, in the present embodiment, because the time window average sum of window when long, time window average sum with short window, the brightness of image variance that is actually in the detection window of reflection changes, therefore change by the brightness of image variance of obtaining in the detection window, can reflect the variation that smog brings video image effectively, thereby can guarantee the triggering accuracy and the sensitivity of smog alert signal.And, owing to can preestablish the value of Smoke Detection sensitivity S, therefore can effectively distinguish the concentration of smog.

And, the interference that can avoid vehicle movement to bring by the top that detection window is arranged on video image.Because smog can spread to tunnel top when rising, and detection window is arranged on the validity that the top can not have influence on algorithm, has also reduced operand simultaneously.In addition, be appreciated that detection window also can be arranged on other positions in the video image.

In addition, need to prove that in the present embodiment, to step 308, the validity of window detects during to length by step 306, is calculating s _LongThe time, calculating be effectively long the time window time window average sum.Whether the window average judges the validity of this detection window when utilizing analyzing and testing window long at threshold range, with the wrong report of effectively avoiding the unusual fluctuation of indivedual detection window feature to bring.But in actual applications, also can omit step 306, calculate s to step 308 _LongThe time, directly calculate all time window average sums of window when long.

Second embodiment of the invention relates to a kind of video detecting method of tunnel smog.Second embodiment improves on the basis of first embodiment, and main improvements are: at the judgement s that satisfies condition _Long〉=s _ShortBehind * the S, do not trigger the smog alert signal immediately, but whether the interval time of further judging current time and the last moment of triggering the smog alert signal is greater than presetting duration, and whether the high light area in track is less than the 1/N of track area, current time with on once trigger the interval time in the moment of smog alert signal greater than presetting duration, and the high light area in track triggers the smog alert signal during less than the 1/N of track area again.

Idiographic flow as shown in Figure 5, step 501 is identical to step 311 to step 511 and step 301, is not giving unnecessary details at this.

In step 512, whether the interval time of judging current time and the last moment of triggering the smog alert signal is greater than presetting duration, if greater than presetting duration, then enter step 513, if not greater than presetting duration, then the processing of current video frame finishes, and gets back to step 502.

In step 513, whether the high light area of further judging the track is less than the 1/N of track area, and N is predefined value, as N=10.Specifically, can be pre-created the sign template of track scope, sign template in track is a two values matrix, and pixel is corresponding one by one in the value in this template and the video image, and the pixel of track imaging is labeled as 1, otherwise is labeled as 0.By the sign template of the track scope created, determine the track scope.Then, in this step or before this step, calculate in the scope of track brightness greater than pixel quantity H and the track area A of T.For 256 grades of gray level images, T ∈ [245,250].In this step, suppose N=10, judge whether to satisfy 10 * H＜A,, the 1/N of the high light area in track less than the track area is described then, then enter step 514, trigger the smog alert signal if satisfy 10 * H＜A; If the high light area in track is not less than the 1/N of track area, then the processing of current video frame finishes, and gets back to step 502.

In the present embodiment, by analyzing the interval that smoke characteristics occurs, can avoid repeatedly reporting to the police with a smog episode.And by comparing high-brightness region ratio in the track, the wrong report that can avoid car light direct projection camera to cause.

In addition, be appreciated that step 512 and step 513 do not have clear and definite precedence relationship, also can first execution in step 513 execution in step 512 again, promptly high-brightness region ratio in the track is relatively earlier judged the interval of analysing the smoke characteristics appearance again, perhaps, also can executed in parallel.

Need to prove that each method embodiment of the present invention all can be realized in modes such as software, hardware, firmwares.No matter the present invention be with software, hardware, or the firmware mode realize, instruction code can be stored in the storer of computer-accessible of any kind (for example permanent or revisable, volatibility or non-volatile, solid-state or non-solid-state, fixing or removable medium or the like).Equally, storer can for example be programmable logic array (Programmable Array Logic, be called for short " PAL "), random access memory (Random Access Memory, be called for short " RAM "), programmable read only memory (Programmable Read Only Memory, be called for short " PROM "), ROM (read-only memory) (Read-Only Memory, be called for short " ROM "), Electrically Erasable Read Only Memory (Electrically Erasable Programmable ROM, be called for short " EEPROM "), disk, CD, digital versatile disc (Digital Versatile Disc is called for short " DVD ") or the like.

Third embodiment of the invention relates to a kind of video detecting device of tunnel smog.As shown in Figure 6, the video detecting device of this tunnel smog comprises:

The time window module is set, be used to each detection window in the video image that window when long and short window are set, the window length of window is L when long _Long, unit is a frame, is used to preserve L _LongThe eigenwert of image in the detection window of individual frame, the window length of short window is L _Short, unit is a frame, is used to preserve L _ShortThe eigenwert of image in the detection window of individual frame, institute's eigenwert is the variance according to the gradation of image of image calculation in the detection window, L _LongGreater than L _Short

The variance computing module is used for when carrying out the Video Detection of tunnel smog, the variance of the gradation of image of each detection window of calculating present frame.The variance computing module can calculate the variance s of the gradation of image of detection window according to following formula ²:

Wherein, y _iBe the gray-scale value of detection window interior pixel point, n is the quantity of pixel.

The time window average update module, window and short window when the variance that is used for the gradation of image of detection window that the variance computing module is calculated joins this detection window long are upgraded the time window average of each window and each short window when long.

The validity mark module, be used for the time window average window during less than Lower Threshold or greater than Upper threshold long, be labeled as invalid window when long, with the time window average window more than or equal to Lower Threshold and when being less than or equal to Upper threshold long, be labeled as effective window when long.Wherein, Lower Threshold can be made as 1000, and Upper threshold can be made as 3000.

The time window average and computing module, be used to calculate the time window average sum of each window when long, and the time window average sum of each short window.In the present embodiment, this time window average and computing module calculate window when long the time during window average sum, calculate the time window average sum of window when effectively long.

Detection module is used to judge whether to meet the following conditions: the time window average sum s of window when long _LongThe time window average sum s of 〉=short window _ShortWith the product of S, and when judgement satisfies condition, trigger the smog alert signal, wherein, S is Smoke Detection sensitivity, and the value of S preestablishes.

In the present embodiment, detection window is evenly arranged in the top of video image, with the interference of avoiding vehicle movement to bring.

Be not difficult to find that first embodiment is and the corresponding method embodiment of present embodiment, present embodiment can with the enforcement of working in coordination of first embodiment.The correlation technique details of mentioning in first embodiment is still effective in the present embodiment, in order to reduce repetition, repeats no more here.Correspondingly, the correlation technique details of mentioning in the present embodiment also can be applicable in first embodiment.

Four embodiment of the invention relates to a kind of video detecting device of tunnel smog.The 4th embodiment improves on the basis of the 3rd embodiment, and main improvements are: detection module also is used at the judgement s that satisfies condition _Long〉=s _ShortDuring * S, whether the interval time of further judging current time and the moment of last triggering smog alert signal is greater than presetting duration, and whether the high light area in track is less than the 1/N of track area, N is predefined value, and when judging greater than the high light area that presets duration and track, trigger the smog alert signal again less than the 1/N of track area.

Be not difficult to find that second embodiment is and the corresponding method embodiment of present embodiment, present embodiment can with the enforcement of working in coordination of second embodiment.The correlation technique details of mentioning in second embodiment is still effective in the present embodiment, in order to reduce repetition, repeats no more here.Correspondingly, the correlation technique details of mentioning in the present embodiment also can be applicable in second embodiment.

Need to prove, each unit of mentioning in each equipment embodiment of the present invention all is a logical block, physically, a logical block can be a physical location, it also can be the part of a physical location, can also realize that the physics realization mode of these logical blocks itself is not most important with the combination of a plurality of physical locations, the combination of the function that these logical blocks realized is the key that just solves technical matters proposed by the invention.In addition, for outstanding innovation part of the present invention, above-mentioned each the equipment embodiment of the present invention will not introduced not too close unit with solving technical matters relation proposed by the invention, and this does not show that there is not other unit in the said equipment embodiment.

Though pass through with reference to some of the preferred embodiment of the invention, the present invention is illustrated and describes, but those of ordinary skill in the art should be understood that and can do various changes to it in the form and details, and without departing from the spirit and scope of the present invention.

Claims (14)

1. the video detecting method of a tunnel smog is characterized in that, comprises following steps:

In advance for each detection window in the video image is provided with window when long and short window, described when long the window length of window be L _Long, unit is a frame, is used to preserve preceding L _LongThe eigenwert of image in the detection window of individual frame, the window length of described short window is L _Short, unit is a frame, is used to preserve preceding L _ShortThe eigenwert of image in the detection window of individual frame, described eigenwert is the variance according to the gradation of image of image calculation in the detection window, described L _LongGreater than described L _Short

When carrying out the Video Detection of tunnel smog, the variance of the gradation of image of each detection window of calculating present frame;

When the variance of the gradation of image of the detection window calculated is joined this detection window described long in window and the short window, upgrade the time window average of each window and each short window when long, this time window average be the average of the variance of gradation of image;

Calculate the time window average sum of each window when long, and calculate the time window average sum of each short window;

If described when long window time window average sum 〉=described short window the time window average sum and S product, then trigger the smog alert signal, wherein, described S is Smoke Detection sensitivity, the value of S preestablishes.

2. the video detecting method of tunnel smog according to claim 1 is characterized in that, described detection window is evenly arranged in the top of video image.

3. the video detecting method of tunnel smog according to claim 1 is characterized in that, described calculating each when long window the time window average sum step before, also comprise following steps:

With the time window average window during less than Lower Threshold or greater than Upper threshold long, be labeled as invalid window when long, with the time window average window more than or equal to described Lower Threshold and when being less than or equal to described Upper threshold long, be labeled as effective window when long;

Described calculating each when long window the time window average sum step in, calculate the time window average sum of window when effectively long.

4. the video detecting method of tunnel smog according to claim 3 is characterized in that, described Xiamen is limited to 1000, describedly is limited to 3000 to the doorstep.

5. the video detecting method of tunnel smog according to claim 1, it is characterized in that, judge described when long window time window average sum 〉=described short window the time window average sum and S product after, before the triggering smog alert signal, also comprise following steps:

If whether the interval time of judging current time and the last moment of triggering the smog alert signal greater than preset duration, then enters the step of described triggering smog alert signal greater than presetting duration again.

6. the video detecting method of tunnel smog according to claim 5 is characterized in that, judge described interval time greater than described preset duration after, before the step that triggers the smog alert signal, also comprise following steps:

Whether the high light area of judging the track is less than the 1/N of track area, and described N is predefined value;

If judge the 1/N of the high light area in track, the step that then enters described triggering smog alert signal again less than the track area.

7. according to the video detecting method of each described tunnel smog in the claim 1 to 6, it is characterized in that, calculate the variance s of the gradation of image of detection window according to following formula ²:

$s^2=\frac{1}{n-1}{\mathrm{\Σ}}_{i=1}^n{(y_i-\stackrel{\‾}{y})}^2$

Wherein, y _iBe the gray-scale value of detection window interior pixel point, n is the quantity of pixel.

8. the video detecting device of a tunnel smog is characterized in that, comprises:

The time window module is set, be used to each detection window in the video image that window when long and short window are set, described when long the window length of window be L _Long, unit is a frame, is used to preserve preceding L _LongThe eigenwert of image in the detection window of individual frame, the window length of described short window is L _Short, unit is a frame, is used to preserve preceding L _ShortThe eigenwert of image in the detection window of individual frame, described eigenwert is the variance according to the gradation of image of image calculation in the detection window, described L _LongGreater than described L _Short

The variance computing module is used for when carrying out the Video Detection of tunnel smog, the variance of the gradation of image of each detection window of calculating present frame;

The time window average update module, window and short window when the variance that is used for the gradation of image of detection window that described variance computing module is calculated joins this detection window described long, upgrade the time window average of each window and each short window when long, this time window average be the average of the variance of gradation of image;

The time window average and computing module, be used to calculate the time window average sum of each window when long, and the time window average sum of each short window;

Detection module is used to judge whether to meet the following conditions: described when long window time window average sum 〉=described short window the time window average sum and S product, and when described condition is satisfied in judgement, trigger the smog alert signal, wherein, described S is Smoke Detection sensitivity, and the value of S preestablishes.

9. the video detecting device of tunnel smog according to claim 8 is characterized in that, described detection window is evenly arranged in the top of video image.

10. the video detecting device of tunnel smog according to claim 8 is characterized in that, the video detecting device of described tunnel smog also comprises:

The validity mark module, be used for the time window average window during less than Lower Threshold or greater than Upper threshold long, be labeled as invalid window when long, with the time window average window more than or equal to described Lower Threshold and when being less than or equal to described Upper threshold long, be labeled as effective window when long;

When described window average and computing module calculate window when long the time during window average sum, calculate the time window average sum of window when effectively long.

11. the video detecting device of tunnel smog according to claim 10 is characterized in that, described Xiamen is limited to 1000, describedly is limited to 3000 to the doorstep.

12. the video detecting device of tunnel smog according to claim 8, it is characterized in that, described detection module also is used for when described condition is satisfied in judgement, whether the interval time of judging current time and the last moment of triggering the smog alert signal is greater than presetting duration, trigger the smog alert signal again when presetting duration judging.

13. the video detecting device of tunnel smog according to claim 12, it is characterized in that, described detection module also is used in described judgement when presetting duration, whether the high light area of judging the track is less than the 1/N of track area, described N is predefined value, and during less than the 1/N of track area, trigger the smog alert signal again at the high light area of judging the track.

14. the video detecting device of tunnel smog according to claim 8 is characterized in that, described variance computing module calculates the variance s of the gradation of image of detection window according to following formula ²:

$s^2=\frac{1}{n-1}{\mathrm{\Σ}}_{i=1}^n{(y_i-\stackrel{\‾}{y})}^2$

Wherein, y _iBe the gray-scale value of detection window interior pixel point, n is the quantity of pixel.

Images