CN101916372B - Video-based smoke detection device and method according to multi-feature fusion - Google Patents
Video-based smoke detection device and method according to multi-feature fusion Download PDFInfo
- Publication number
- CN101916372B CN101916372B CN 201010275772 CN201010275772A CN101916372B CN 101916372 B CN101916372 B CN 101916372B CN 201010275772 CN201010275772 CN 201010275772 CN 201010275772 A CN201010275772 A CN 201010275772A CN 101916372 B CN101916372 B CN 101916372B
- Authority
- CN
- China
- Prior art keywords
- characteristic
- cigarette
- central motion
- prospect
- foreground
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a video-based smoke detection device and method according to multi-feature fusion. The method comprises the following steps: a shooting unit and a foreground-extracting unit are used to extract the foreground region of the current image; a foreground central computing unit and a foreground central storing unit of continuous multi-frame are used to obtain the central position information of the foreground region of the continuous multi-frame surveillance area image, then a foreground central motion range calculating unit is used to obtain the external rectangle of the foreground central position; then a foreground central motion range size calculating unit and a central motion characteristic judging unit are used to judge whether the central position has the central motion characteristic of smoke; and a noncentral motion characteristic detection unit and a noncentral motion characteristic condition judging unit are used to detect and judge the other characteristics of smoke, and a smoke judging unit is used to perform further judgment according to the obtained result and judge whether smoke exists in the surveillance area. The device and method of the invention are characterized by little computation workload, high accuracy rate, low false alarm rate and missing report rate and the like.
Description
Technical field
The present invention relates to a kind of smoke detecting apparatus and method of the many Feature Fusion based on video.
Background technology
Modern society, the use of various occasion electrical equipment is more and more, has increased potential safety hazard, and people's prevention awareness also strengthens gradually, begins to pay close attention to various security products.At present, there has been relevant monitoring security product on market, to sell, from monitoring means, can be divided into two types basically: sensor monitoring and video monitoring.Sensor monitoring is to utilize smoke sensor to survey whether the generation of cigarette is arranged and report to the police; This monitor mode is vulnerable to extraneous factor and disturbs (as: extraneous factors such as temperature, distance); Rate of false alarm is high; And the reaction of the smoke detector system of sensor class is slow, needs finite concentration and temperature, and loss causes during discovery.Video monitoring is to utilize the real-time collection site video of camera, is guarded constantly by the operator on duty, takes place in time to handle when there is the situation of cigarette, fire at the scene; Though rate of false alarm is low like this, owing to need artificial on duty; Time has been grown and has been easy to generate visual fatigue, causes careless omission.In addition, part alarm video monitoring system automatically can realize automatic warning, but because technical reason, rate of false alarm is very high.So lack at present that a kind of applied range, detectability are strong, real-time and the high smoke detection system of reliability.
In addition; Also have mechanism at the development video brainpower watch and control cigarette to be discerned, it mainly is to sum up the smoke certain law according to taking the cigarette image, as the characteristic of cigarette; Compare identification when obtaining image scene afterwards, just think have cigarette to produce when image scene satisfies above-mentioned cigarette characteristic.The normal at present cigarette characteristic of using mainly contains: cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic, color saturation degree characteristic of dark brown and cigarette morphological feature.But mostly rest on laboratory stage, and the report that does not have respective combination to use.
Summary of the invention
The present invention is directed to the proposition of above problem, and develop a kind of smoke detecting apparatus and method of the many Feature Fusion based on video.The present invention adopts technological means following:
A kind of smoke detecting apparatus is characterized in that, has:
Take the unit, obtain the present image of monitor area;
The foreground extraction unit according to present image and the background image that said shooting unit photographs arrives, extracts the foreground area of present image;
Prospect center calculation unit, the center of calculating the foreground area in the current frame image that extracts said foreground extraction unit;
The unit is preserved at the prospect center of continuous multiple frames, preserves the center information of the foreground area of the continuous multiple frames monitor area image that said prospect center calculation unit calculates;
Prospect central motion range computation unit; Preserve the center information of foreground area of the continuous multiple frames monitor area image of unit storage according to the prospect center of said continuous multiple frames, generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image;
Prospect central motion range size computing unit, the boundary rectangle that prospect central motion range computation unit is generated carries out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle;
Central motion characteristic identifying unit; Whether the ratio of width to height of judging rectangular area that said prospect central motion range size computing unit draws and/or rectangle is in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic;
Non-central motion feature detecting unit; Be used to detect the foreground area of extracting said foreground extraction unit and whether have cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic; Cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
Non-central motion feature condition criterion unit; Be used for judging whether non-central motion feature detection has said cigarette scrambling characteristic to foreground image; Cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
The cigarette identifying unit, the result who draws according to said central motion characteristic identifying unit and non-central motion feature condition criterion unit judges whether have cigarette; Judge foreground area when said central motion characteristic identifying unit and have cigarette central motion characteristic; And non-central motion feature condition criterion unit judges goes out to have said cigarette scrambling characteristic in the foreground image; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
The boundary rectangle that contains said prospect center that said prospect central motion range computation unit generates is the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
The boundary rectangle that contains said prospect center that said prospect central motion range computation unit generates, the intensive zone, center of the foreground area of the continuous multiple frames monitor area image of preserving the unit is preserved at the prospect center that is positioned at said continuous multiple frames.
A kind of cigarette detection method is characterized in that having:
Take step, obtain the present image of monitor area;
The foreground extraction step according to present image and the background image that said shooting step photographs, is extracted the foreground area of present image;
Prospect center calculation step, the center of calculating the foreground area in the current frame image that said foreground extraction step extracts;
Step is preserved at the prospect center of continuous multiple frames, preserves the center information of the foreground area of the continuous multiple frames monitor area image that said prospect center calculation step calculates;
Prospect central motion range computation step; Preserve the center information of foreground area of the continuous multiple frames monitor area image of step storage according to the prospect center of said continuous multiple frames, generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image;
Prospect central motion range size calculation procedure, the boundary rectangle that prospect central motion range computation step is generated carries out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle;
Central motion characteristic determination step; Whether the ratio of width to height of judging rectangular area that said prospect central motion range size calculation procedure draws and/or rectangle is in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic;
Non-central motion feature detects step; Be used to detect the foreground area that said foreground extraction step extracts and whether have cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic; Cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
Non-central motion feature condition criterion step; Be used for judging that non-central motion feature detection step detects foreground image and whether has said cigarette scrambling characteristic; Cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
The cigarette determination step, the result who draws according to said central motion characteristic determination step and non-central motion feature condition criterion step judges whether have cigarette; Judge foreground area when said central motion characteristic determination step and have cigarette central motion characteristic; And non-central motion feature condition criterion step is judged has said cigarette scrambling characteristic in the foreground image; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
The boundary rectangle that contains said prospect center that said prospect central motion range computation step generates is the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
The boundary rectangle that contains said prospect center that said prospect central motion range computation step generates, the prospect center that is positioned at said continuous multiple frames are preserved the intensive zone, center of the foreground area of the continuous multiple frames monitor area image that step preserves.
Owing to adopted technique scheme, the smoke detecting apparatus and the method for the many Feature Fusion based on video provided by the invention are through video image identification algorithm; Central motion rule to smog detects, and combines existing cigarette scrambling characteristic, cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, the detection method of color saturation degree characteristic of dark brown and cigarette morphological feature; Foreground image to extracting is discerned detection, then the identification testing result is judged, as the criterion that whether has cigarette to produce; Has the accuracy rate height, characteristics such as rate of false alarm, rate of failing to report are low.In addition, in cigarette central motion feature detection, introduce the judgement of boundary rectangle size and not only can reduce operand significantly, and can farthest guarantee the accuracy rate of its judgement.
Description of drawings
Fig. 1 is a structured flowchart of the present invention;
Fig. 2 is a process flow diagram of the present invention;
Fig. 3 is the process flow diagram that non-central motion feature detects step in the instance for the present invention.
Embodiment
The smoke detecting apparatus of the many Feature Fusion based on video of the present invention and the principle of method are, through carrying out cigarette scrambling characteristic, cigarette diffusivity characteristic to obtaining monitoring image; The cigarette color characteristic; Cigarette translucence characteristic, the detection of color saturation degree characteristic of dark brown and cigarette morphological feature combines the detection of central motion characteristic simultaneously; Draw comprehensive judgement according to the result who detects at last, to reach the purpose that cigarette detects.Implementation method is: mainly be the image that obtains monitor area through video camera; Accomplish the processing of view data through data processor (can form) then, and judge whether monitor area has the generation of cigarette by the device that server, PC or integrated processing enter etc. have a data processing function.Below in conjunction with accompanying drawing the present invention is specified:
As shown in Figure 1, the video camera in this smoke detecting apparatus can be CMOS or CCD, can be thermal camera or common camera; Shooting unit 101 wherein will obtain the present image of monitor area and give the foreground extraction unit 102 in the data processor to handle; Present image and background image that foreground extraction unit 102 photographs according to said shooting unit 101, the foreground area of extraction present image; Foreground area in the current frame image that extracts through 103 pairs of foreground extraction unit, prospect center calculation unit is then calculated acquisition prospect center; After give the prospect center of continuous multiple frames to preserve unit 104 to preserve, the center information that the foreground area of the continuous multiple frames monitor area image that said prospect center calculation unit 103 calculates is preserved in unit 104 is preserved at the prospect center of this continuous multiple frames; And, generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image through the center information of prospect central motion range computation unit 105 according to the foreground area of the continuous multiple frames monitor area image of prospect center preservation unit 104 storages of continuous multiple frames; By prospect central motion range size computing unit 106, the boundary rectangle that prospect central motion range computation unit 105 is generated carries out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle again; Whether the ratio of width to height that rectangular area that said prospect central motion range size computing unit 106 draws and/or rectangle are judged by central motion characteristic identifying unit 107 in the back is in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic.Said non-central motion feature detecting unit 108; Be used to detect the foreground area of extracting said foreground extraction unit 102 and whether have cigarette scrambling characteristic; Cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature; And give non-central motion feature condition criterion unit 109 to judge the detected result; Judge whether non-central motion feature detection has said cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic in foreground image; Cigarette translucence characteristic; At least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature has then to provide corresponding result of determination, otherwise does not make the judgement with above-mentioned characteristic.The result who is drawn according to said central motion characteristic identifying unit 107 and non-central motion feature condition criterion unit 109 by cigarette identifying unit 110 at last judges whether have cigarette; Judge foreground area when said central motion characteristic identifying unit 107 and have cigarette central motion characteristic; And non-central motion feature condition criterion unit 109 is judged has said cigarette scrambling characteristic in the foreground image; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
Said prospect central motion range computation unit 105 mainly comprises following two kinds of technical schemes that generate boundary rectangle:
1. generate the boundary rectangle contain said prospect center, be the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
2. generate the boundary rectangle that contains said prospect center, the intensive zone, center of the foreground area of the continuous multiple frames monitor area image of preserving the unit is preserved at the prospect center that is positioned at said continuous multiple frames.
Above-mentioned each detection and identifying unit all can be realized through microprocessor separately, but also can on a station server, realize through a high performance CPU for practicing thrift cost.Present embodiment adopts is that arm processor and dsp processor are used and above-mentioned each functional module are realized wherein dsp processor is realized concrete operation of data, the processing of arm processor realization overall logic and judgement.View data therebetween, intermediate data and result data all can be stored in the storer that links to each other with data processor.
In addition because the different existing different interference in field, so above-mentioned threshold value need carry out can confirming only value after the actual test according to the characteristics in place, so describe with regard to the data of no longer enumerating reality here.
As shown in Figure 2 through description to detection method, the technical scheme that the present invention proposes is done further elaboration:
S201, take and obtain the present image of monitor area through video camera.
Present image and background image that S202, data processor photograph according to said shooting step; Extract the foreground area of present image; This foreground extraction step can adopt the method for adaptive background subduction model to come the extraction prospect, through 1. extracting the monochrome information of present frame; 2. calculate binary-state threshold; 3. upgrade background; 4. background subtraction; 5. the processing of the step of binary conversion treatment completion extraction foreground image is to obtain the foreground area of image.
The center of the foreground area in the current frame image that S203, the above-mentioned foreground extraction step of calculating are extracted.
The center information of the foreground area of the continuous multiple frames monitor area image that S204, the above-mentioned prospect center calculation step of preservation calculate.
S205, the center information of foreground area of preserving the continuous multiple frames monitor area image of step storage according to the prospect center of above-mentioned continuous multiple frames generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image.
S206, the boundary rectangle that prospect central motion range computation step is generated carry out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle.
S207, judge rectangular area that above-mentioned prospect central motion range size calculation procedure draws and/or rectangle the ratio of width to height whether in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic; The threshold range of not stipulating again then returns step S201 and detects again.
Whether the foreground area that S208, the above-mentioned foreground extraction step of detection are extracted has cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
S209, the above-mentioned non-central motion feature of judgement detect step and detect whether have said cigarette scrambling characteristic in the foreground image; Cigarette diffusivity characteristic; The cigarette color characteristic; Cigarette translucence characteristic, at least a characteristic (decision procedure is as shown in Figure 3, can adopt parallel judgment mode) in color saturation degree characteristic of dark brown and the cigarette morphological feature; Satisfied then further judge, do not satisfy and then return step S201 and detect again.
S210, the result who draws according to above-mentioned central motion characteristic determination step S207 and non-central motion feature condition criterion step S209 judge whether have cigarette; Judge foreground area when said central motion characteristic determination step and have cigarette central motion characteristic; And non-central motion feature condition criterion step is judged has said cigarette scrambling characteristic in the foreground image; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
Implementation procedure same steps as S205 with above-mentioned device mainly comprises following two kinds of technical schemes that generate boundary rectangle:
1. generate the boundary rectangle contain said prospect center, be the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
2. generate the boundary rectangle that contains said prospect center, the intensive zone, center of the foreground area of the continuous multiple frames monitor area image of preserving the unit is preserved at the prospect center that is positioned at said continuous multiple frames.
The principle of cigarette central motion characteristic: the smog in the fire can constantly spread in production process, and volume increases gradually, sees that from image then showing as area constantly becomes big; And because the effect of air-flow, the trend of motion can orientedly just be gone up, tiltedly gone up to smog usually; Above-mentioned two characteristics can embody through the central motion rule of smog on image, thereby the motor area of smog with other objects separated, and reach the purpose of Smoke Detection.
Following illustrative example is elaborated to step S203 to S207:
If the area threshold of prospect central motion scope is S=18, the number of continuous multiple frames is n=22.
1, calculate average coordinates Xavg and the average coordinates Yavg on the Y direction on the X-direction of all foreground points in the foreground image, (Xavg Yavg) is designated as the center of this foreground area with point.
2, the foreground area center of record and preservation continuous multiple frames.
3, to the foreground area center of the continuous multiple frames that write down; Calculate the minimum value Xmin and the maximal value Xmax of horizontal ordinate, and the minimum value Ymin of ordinate and maximal value Ymax, with (Xmin; Ymin) (Xmin; Ymax) (Xmax, Ymax) (Xmax, Ymin) 4 is exactly the boundary rectangle at the prospect center of a continuous multiple frames for the rectangle on summit.
4, calculate the size of above-mentioned boundary rectangle, present embodiment only limits rectangle size with rectangular area, in addition can also limit the size of rectangle with the ratio of width to height of rectangle, and perhaps area is used in combination with the ratio of width to height and limits rectangle size.
If the area of 5 boundary rectangles, then meets the central motion rule of smog greater than threshold value S, be judged as smog.
Content to claim 3 and 6; The influence that testing result is caused for the prospect center of getting rid of indivedual malpositions; Can when the prospect of calculating center, append a qualifications; If the distance between the prospect center of the prospect center of present image and the continuous multiple frames that write down is excessive, think that then the prospect center of present image is unusual, do not carry out record.The practical implementation method can adopt following method: after above-mentioned steps 1, carry out following steps, calculate the horizontal ordinate mean value Xcent and the ordinate mean value Ycent at the prospect center of the continuous multiple frames that has write down; Prospect centre coordinate (Xavg according to the present image that calculates in the above-mentioned steps 1; Yavg), calculate (Xcent-Xavg) and absolute value (Ycent-Yavg) respectively, as long as an absolute value is arranged greater than defined threshold; Then do not write down the prospect center of present image; Continue to detect the next frame image, otherwise carry out above-mentioned steps 2, the prospect center of record present image.
In addition; To what detected in the non-cigarette central motion feature detection step described in above-mentioned steps S208 and the step 209: cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic; Cigarette translucence characteristic, the concrete detection means of color saturation degree characteristic of dark brown and cigarette morphological feature is following:
Cigarette scrambling characteristic detection method: extract the foreground area edge, calculate the girth square of foreground area and the ratio of area, if in the threshold range of stipulating, then present frame meets the scrambling characteristic; Specifically be to extract the prospect edge with Canny operator method; That calculates the foreground area girth square multiply by the ratio of 4 π with the foreground area area; If ratio is in the threshold range of regulation (for example: greater than 20 and less than 180), then present frame satisfies the scrambling characteristic of smog.The interference that the moving object of regular shape brings can be effectively got rid of in the detection of scrambling, reduces rate of false alarm.
Cigarette diffusivity characteristic detection method: calculate the ratio of the foreground area area of present frame and former frame, if satisfy the threshold condition of regulation, then present frame meets the diffusivity characteristic; It specifically is the area of representing present frame and previous frame foreground area respectively with m_curarea and m_prearea; If m_curarea/m_prearea satisfies the threshold range (for example: between 0.76 and 1.05) of regulation, then present frame satisfies the diffusivity characteristic of smog.Diffusible detection can be got rid of the interference that the rigid motion object of fixed in shape brings.
Cigarette color characteristic detection method: calculate the relation between the foreground image each point rgb value,, then meet color characteristic if satisfy defined terms, if the quantity of foreground point that meets color characteristic in the defined threshold scope, then present frame meets color characteristic; Concrete calculating also judges whether the each point rgb value in the prospect satisfies following three conditions: the difference of condition 1:RGB maximal value and minimum value less than the threshold value T1 that stipulates (for example: T1=30) condition 2:RGB mean value [Tmin in the threshold range of regulation; Tmax] (for example: Tmin=110; When Tmax=220) condition 3:B value is maximum; The difference of RGB maximal value and minimum value less than certain threshold value T2 (for example: if T2=80) certain point satisfy condition 1 with condition 2; Perhaps satisfy condition 2 with condition 3, then this point satisfies the color characteristic of smog.If the pixel that satisfies the smog color characteristic in prospect proportion (for example: T3=0.6), then present frame satisfies the color characteristic of smog greater than defined threshold T3.The detection of smog color can be got rid of the interference that differs bigger with the smog color moving object and cause.
Cigarette translucence characteristic detection method: calculate the foreground area edge energy and reduce degree, if greater than defined threshold, then present frame meets the translucence characteristic; Perhaps calculate the relative attenuation coefficient of discharge of introducing after the high-pass filtering, if satisfy the threshold condition of regulation, then present frame meets the translucence characteristic; Specifically can adopt two kinds of methods, method 1: judge translucence according to prospect edge energy minimizing degree.If the difference of the prospect edge energy of the prospect edge energy of background frames and present frame then satisfies the translucence characteristic in the threshold range of regulation (for example: greater than 0.025, less than 0.036).The computing formula of edge energy is: w=v1*v2+h1*h2+d1*d2; Wherein, v1 changes for the background image horizontal direction, and v2 changes for the present image horizontal direction, and h1 changes for the background image vertical direction, and h2 changes for the present image vertical direction, and d1 changes for the background image diagonal, and d2 changes for the present image diagonal.
Method 2: the corresponding point of establishing high-pass filtering rear backdrop frame and prospect frame be respectively B (x, y), F (x, y), the relative attenuation coefficient of discharge be C (x, y), if B (x, y)＞F (x, y), then C (x, y)=(B (and x, y)-F (x, y))/B (x, y), otherwise C (x, y)=0; If (then this is potential smog point to the relative attenuation coefficient of discharge C of the marginal point of prospect for x, y) in certain threshold range (for example: greater than 0.06, less than 0.33).With the number of potential smog point on the method statistics prospect edge, and the number of potential smog point in the whole prospect.If in the prospect marginal point potential smog point proportion the regulation threshold range in (for example: greater than 0.25; Less than 0.6); Potential smog point proportion is greater than the threshold value (for example: greater than 0.08) of regulation in whole prospect simultaneously, and then present frame satisfies the translucence characteristic of smog.
The color saturation degree characteristic detection method of dark brown: in the hsv color space; The S value of corresponding point is poor in the S value of calculating present frame prospect each point and the background frames; If satisfy the threshold condition of regulation; Then meet the color saturation characteristic, if the quantity of foreground point that meets the color saturation characteristic in the defined threshold scope, then present frame meets the color saturation characteristic; Specifically in the hsv color space, if in the prospect of present frame (x, and in y) the S value and background frames (x, the difference of S value y) is in the threshold range of stipulating (for example: greater than 6, less than 8), and then this point satisfies the color saturation characteristic.If the pixel number that satisfies color saturation reach regulation threshold value (for example: account for whole foreground points 80%), then present frame satisfies smog color saturation characteristic.
Cigarette morphological feature detection method: the rectangular area that a two field picture on average is divided into identical size; If comprise the foreground point of defined amount in certain rectangular area then be labeled as the prospect fringe region; The coincidence degree of the prospect fringe region of statistics present frame and former frame; If the coincidence degree satisfies the threshold condition of regulation, then present frame meets morphological feature; Significant change can take place in smog form in the process of flying away, and this is a characteristic that is different from the rigid motion object.Detection method: the detected prospect of current frame image border is overlapped with the prospect border that the former frame image detection goes out, describe the variation of form according to the coincidence degree.Concrete steps are following:
A) the rectangular pixels zone that image segmentation is become some identical sizes (for example: the 4*4 size), for each zone a zone bit is set simultaneously.
B) detect the interior foreground point number in each zone, if foreground point number (for example: greater than 5 and less than 11) in the defined threshold scope in this zone was 1 (representing that this zone is the border of prospect) with this area flag position then, otherwise is 0.
C) zone bit of two two field pictures of the front and back after the computed segmentation is the coincidence degree in 1 zone; Represent that with Same_num collimation mark will position, same spatial location top is 1 quantity in two two field pictures of front and back; Use Area_num to represent that square frame mark position in the current frame image is 1 quantity; If the ratio of the two (Same_num/Area_num) is in the threshold range of regulation (for example: between 0.3～0.7), then present frame satisfies the smog morphological change characteristics.The smog form detects the interference that the rigid motion object can get rid of fixed in shape brings.
This cigarette detection method can well be distinguished smog and other moving object, and the object of cyclical movement is by a small margin distinguished better effects if.
The above; Be merely the preferable embodiment of the present invention; But protection scope of the present invention is not limited thereto; Any technician who is familiar with the present technique field is equal to replacement or change according to technical scheme of the present invention and inventive concept thereof in the technical scope that the present invention discloses, all should be encompassed within protection scope of the present invention.
Claims (6)
1. smoke detecting apparatus is characterized in that having:
Take the unit, obtain the present image of monitor area;
The foreground extraction unit according to present image and the background image that said shooting unit photographs arrives, extracts the foreground area of present image;
Prospect center calculation unit, the center of calculating the foreground area in the current frame image that extracts said foreground extraction unit;
The unit is preserved at the prospect center of continuous multiple frames, preserves the center information of the foreground area of the continuous multiple frames monitor area image that said prospect center calculation unit calculates;
Prospect central motion range computation unit; Preserve the center information of foreground area of the continuous multiple frames monitor area image of unit storage according to the prospect center of said continuous multiple frames, generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image;
Prospect central motion range size computing unit, the boundary rectangle that prospect central motion range computation unit is generated carries out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle;
Central motion characteristic identifying unit; Whether the ratio of width to height of judging rectangular area that said prospect central motion range size computing unit draws and/or rectangle is in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic;
Non-central motion feature detecting unit; Be used to detect the foreground area of extracting said foreground extraction unit and whether have cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic; Cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
Non-central motion feature condition criterion unit; Be used for judging whether non-central motion feature detection has said cigarette scrambling characteristic to foreground area; Cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
The cigarette identifying unit, the result who draws according to said central motion characteristic identifying unit and non-central motion feature condition criterion unit judges whether have cigarette; Judge foreground area when said central motion characteristic identifying unit and have cigarette central motion characteristic; And non-central motion feature condition criterion unit judges goes out to have said cigarette scrambling characteristic in the foreground area; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
2. a kind of smoke detecting apparatus according to claim 1; The boundary rectangle that it is characterized in that the center of containing said foreground area that said prospect central motion range computation unit generates is the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
3. a kind of smoke detecting apparatus according to claim 1; The boundary rectangle that it is characterized in that the center of containing said foreground area that said prospect central motion range computation unit generates, the intensive zone, center of the foreground area of the continuous multiple frames monitor area image of preserving the unit is preserved at the prospect center that is positioned at said continuous multiple frames.
4. cigarette detection method is characterized in that having:
Take step, obtain the present image of monitor area;
The foreground extraction step according to present image and the background image that said shooting step photographs, is extracted the foreground area of present image;
Prospect center calculation step, the center of calculating the foreground area in the current frame image that said foreground extraction step extracts;
Step is preserved at the prospect center of continuous multiple frames, preserves the center information of the foreground area of the continuous multiple frames monitor area image that said prospect center calculation step calculates;
Prospect central motion range computation step; Preserve the center information of foreground area of the continuous multiple frames monitor area image of step storage according to the prospect center of said continuous multiple frames, generate the boundary rectangle of the center of the foreground area that contains said continuous multiple frames monitor area image;
Prospect central motion range size calculation procedure, the boundary rectangle that prospect central motion range computation step is generated carries out the ratio of width to height size calculating of calculating of rectangular area size and/or rectangle;
Central motion characteristic determination step; Whether the ratio of width to height of judging rectangular area that said prospect central motion range size calculation procedure draws and/or rectangle is in the defined threshold scope, if in the defined threshold scope then judge that said foreground area has cigarette central motion characteristic;
Non-central motion feature detects step; Be used to detect the foreground area that said foreground extraction step extracts and whether have cigarette scrambling characteristic, cigarette diffusivity characteristic, cigarette color characteristic; Cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
Non-central motion feature condition criterion step; Be used for judging that non-central motion feature detection step detects foreground area and whether has said cigarette scrambling characteristic; Cigarette diffusivity characteristic; The cigarette color characteristic, cigarette translucence characteristic, at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature;
The cigarette determination step, the result who draws according to said central motion characteristic determination step and non-central motion feature condition criterion step judges whether have cigarette; Judge foreground area when said central motion characteristic determination step and have cigarette central motion characteristic; And non-central motion feature condition criterion step is judged has said cigarette scrambling characteristic in the foreground area; Cigarette diffusivity characteristic, cigarette color characteristic, cigarette translucence characteristic; Under the situation of at least a characteristic in color saturation degree characteristic of dark brown and the cigarette morphological feature, judging has cigarette.
5. a kind of cigarette detection method according to claim 4; The boundary rectangle that it is characterized in that the center of containing said foreground area that said prospect central motion range computation step generates is the boundary rectangle of the center of the foreground area that comprises all continuous multiple frames monitor area images.
6. a kind of cigarette detection method according to claim 4; The boundary rectangle that it is characterized in that the center of containing said foreground area that said prospect central motion range computation step generates, the prospect center that is positioned at said continuous multiple frames are preserved the intensive zone, center of the foreground area of the continuous multiple frames monitor area image that step preserves.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010275772 CN101916372B (en) | 2010-09-08 | 2010-09-08 | Video-based smoke detection device and method according to multi-feature fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010275772 CN101916372B (en) | 2010-09-08 | 2010-09-08 | Video-based smoke detection device and method according to multi-feature fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101916372A CN101916372A (en) | 2010-12-15 |
| CN101916372B true CN101916372B (en) | 2012-12-26 |
Family
ID=43323881
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010275772 Expired - Fee Related CN101916372B (en) | 2010-09-08 | 2010-09-08 | Video-based smoke detection device and method according to multi-feature fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101916372B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567722B (en) * | 2012-01-17 | 2013-09-25 | 大连民族学院 | Early-stage smoke detection method based on codebook model and multiple features |
| CN103310206B (en) * | 2013-07-06 | 2016-05-25 | 中国科学技术大学 | A kind of vehicle using motor detection method based on many features and multiframe information fusion |
| CN104978733B (en) * | 2014-04-11 | 2018-02-23 | 富士通株式会社 | Smog detection method and device |
| CN108140291A (en) * | 2015-11-20 | 2018-06-08 | 富士通株式会社 | Mist detecting device, method and image processing equipment |
| CN107633212A (en) * | 2017-08-30 | 2018-01-26 | 清华大学苏州汽车研究院（吴江） | A kind of firework detecting method and device based on video image |
| CN111353334A (en) | 2018-12-21 | 2020-06-30 | 富士通株式会社 | Smoke detection method and device |
| CN112634577B (en) * | 2019-09-24 | 2022-07-26 | 中科智云科技有限公司 | Method and apparatus for smoke alarm |
| CN111652184B (en) * | 2020-06-19 | 2023-06-20 | 成都通甲优博科技有限责任公司 | Smoke identification method, device, storage medium and data processing equipment |
| CN112288886B (en) * | 2020-09-15 | 2022-02-15 | 陈学鹏 | Tooth position arrangement method of accurate digital tooth model |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101059435A (en) * | 2007-05-17 | 2007-10-24 | 西北工业大学 | Method for detecting forest fire fog based on colorful CCD image analysis |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006165666A (en) * | 2004-12-02 | 2006-06-22 | Matsushita Electric Ind Co Ltd | Surveillance camera, surveillance image monitoring apparatus, surveillance image storage apparatus, and surveillance system |
| WO2008037293A1 (en) * | 2006-09-25 | 2008-04-03 | Siemens Schweiz Ag | Detection of smoke with a video camera |
-
2010
- 2010-09-08 CN CN 201010275772 patent/CN101916372B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101059435A (en) * | 2007-05-17 | 2007-10-24 | 西北工业大学 | Method for detecting forest fire fog based on colorful CCD image analysis |
Non-Patent Citations (3)
| Title |
|---|
| JP特开2006-165666A 2006.06.22 |
| 汪锦 等.面向红外视频图像的火焰识别.《上海交通大学学报》.2008,第42卷(第12期), * |
| 程鑫 等.图像型火灾火焰探测原理.《火灾科学》.2005,第14卷(第4期), * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101916372A (en) | 2010-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101916372B (en) | Video-based smoke detection device and method according to multi-feature fusion | |
| CN101944267B (en) | Smoke and fire detection device based on videos | |
| CN109886130B (en) | Target object determination method and device, storage medium and processor | |
| CN107256225B (en) | Method and device for generating heat map based on video analysis | |
| US9070023B2 (en) | System and method of alerting a driver that visual perception of pedestrian may be difficult | |
| EP2118862B1 (en) | System and method for video detection of smoke and flame | |
| US9025875B2 (en) | People counting device, people counting method and people counting program | |
| CN102122390B (en) | Method for detecting human body based on range image | |
| CN112669344B (en) | Method and device for positioning moving object, electronic equipment and storage medium | |
| JP6764481B2 (en) | Monitoring device | |
| CN103578116A (en) | Apparatus and method for tracking object | |
| US20100201871A1 (en) | Temporally consistent caption detection on videos using a 3D spatiotemporal method | |
| WO2001039498A2 (en) | Method and apparatus for detecting moving objects in video conferencing and other applications | |
| CN111091098A (en) | Training method and detection method of detection model and related device | |
| CN101916380B (en) | Video-based device and method for detecting smog | |
| CN105957300B (en) | A kind of wisdom gold eyeball identification is suspicious to put up masking alarm method and device | |
| CN114202646A (en) | Infrared image smoking detection method and system based on deep learning | |
| KR20140029069A (en) | Method and apparatus for monitoring objects from video | |
| CN101930540A (en) | Video-based multi-feature fusion flame detecting device and method | |
| CN111428600A (en) | Smoking detection method, system and device and thermal infrared image processor | |
| CN111723656A (en) | Smoke detection method and device based on YOLO v3 and self-optimization | |
| Brunner et al. | Visual metrics for the evaluation of sensor data quality in outdoor perception | |
| CN107256566B (en) | Forest fires detection method based on radiation energy | |
| CN101930541A (en) | Video-based flame detecting device and method | |
| KR102150661B1 (en) | Method and Apparatus for Preprocessing Image for Detecting Objects |
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: 20121226 Termination date: 20190908 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |