CN105825198A - Pedestrian detection method and device - Google Patents
Pedestrian detection method and device Download PDFInfo
- Publication number
- CN105825198A CN105825198A CN201610188142.4A CN201610188142A CN105825198A CN 105825198 A CN105825198 A CN 105825198A CN 201610188142 A CN201610188142 A CN 201610188142A CN 105825198 A CN105825198 A CN 105825198A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- video information
- behavior characteristics
- parameter
- prospect
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/44—Event detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Closed-Circuit Television Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a pedestrian detection method. The method comprises: obtaining the video information of a detection area collected by a video monitoring device in real time; analyzing the obtained video information, and performing statistics of the pedestrian parameters in the detection area and the pedestrians' behavior characteristics; and emitting warning information when it is determined that there is an abnormal event in the detection area according to the pedestrian parameters and the pedestrians' behavior characteristics. The present invention further provides a pedestrian detection device. The pedestrian detection method and device are able to automatically analyze the monitoring videos, timely discover abnormal events and improve the detection efficiency of pedestrians.
Description
Technical field
The present invention relates to intelligent monitoring technology field, particularly relate to a kind of pedestrian detection method and device.
Background technology
In some public places; the place that particularly pedestrian is more; the places such as such as bus platform, transport hub, stadiums; the situation that pedestrian density is bigger often occurs; it is susceptible to anomalous event in this case; there is potential safety hazard, such as, pedestrian illegally enters prohibited area etc. or occurs article the problem such as to leave over.At present, it is typically all and arranges in these places Security Personnel to exercise supervision, or, it is monitored by video monitoring apparatus, then by artificial checking monitoring video, scene is analyzed the event of noting abnormalities, but, for the place that place is bigger, video monitoring equipment is the most, it is accomplished by user and checks substantial amounts of video information, not only inefficiency simultaneously, and cause some anomalous events to be found in time.
Summary of the invention
The present invention provides a kind of pedestrian detection method and device, and its main purpose is automatically to be analyzed monitor video, and note abnormalities event in time, improves the efficiency of pedestrian detection.
For achieving the above object, the present invention provides a kind of pedestrian detection method, and this pedestrian detection method includes:
Obtain the video information in the detection region collected based on video monitoring apparatus in real time;
The described video information got is resolved, adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information;
When the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, send information warning.
Alternatively, the described described video information to getting resolves, and before adding up the step of the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information, described pedestrian detection method also includes:
The described video information got is carried out pretreatment, to eliminate outside environmental elements to the impact of prospect in described detection region.
Alternatively, the described described video information to getting resolves, and the step adding up the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information includes:
Choosing with reference to background from video information, the described reference background according to choosing obtains prospect from described video information;
Using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Alternatively, described choosing with reference to background from video information, after the described step obtaining prospect with reference to background from described video information chosen, described pedestrian detection method further comprises the steps of:
Use pedestrian's grader that the described prospect in the described detection region in each two field picture of described video information is analyzed;
In conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.
Alternatively, the described described video information to getting resolves, and after adding up the step of the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information, described pedestrian detection method also includes:
When the described pedestrian's parameter according to statistics and the described behavior characteristics of pedestrian judge, in described detection region, anomalous event occurs, described video information and corresponding pedestrian's parameter of described video information and behavior characteristics are added in corresponding described pedestrian's grader.
Additionally, for achieving the above object, the present invention also provides for a kind of pedestrian detection device, and this pedestrian detection device includes:
Acquisition module, for obtaining the video information detecting region collected based on video monitoring apparatus in real time;
Parsing module, for resolving the described video information got, adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information;
Perform module, for when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, sending information warning.
Alternatively, described pedestrian detection device also includes:
Pretreatment module, for carrying out pretreatment to the described video information got, to eliminate outside environmental elements to the impact of prospect in described detection region.
Alternatively, described parsing module, it is additionally operable to choose from video information with reference to background, the described reference background according to choosing obtains prospect from described video information;
And, using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Alternatively, described parsing module, it is additionally operable to use pedestrian's grader that the described prospect in the described detection region in each two field picture of described video information is analyzed;
And, in conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.
Alternatively, described pedestrian detection device also includes:
Add module, for when the described pedestrian's parameter according to statistics and the described behavior characteristics of pedestrian judge, in described detection region, anomalous event occurs, described video information and corresponding pedestrian's parameter of described video information and behavior characteristics are added in corresponding described pedestrian's grader.
nullThe pedestrian detection method of present invention proposition and device,Obtain the video information in the detection region that video monitoring apparatus gathers in real time,And the image information got is resolved,Count pedestrian's parameter and the behavior characteristics of pedestrian in detection region,Judge whether scene has anomalous event,Such as fight、Article are left over etc. and to be occurred,When judging to have anomalous event to occur,Information warning can be sent,To remind user that the anomalous event occurred is processed,The present invention is without artificial checking monitoring video,It is capable of monitor video is automatically analyzed getting pedestrian's parameter,And then the scene state in detection region is estimated,Even if in the case of monitor video quantity is bigger,Can be analyzed processing to it equally,Can note abnormalities from substantial amounts of video information event according to pedestrian's parameter,Improve the efficiency of pedestrian detection.
Accompanying drawing explanation
Fig. 1 is the flow chart of pedestrian detection method first embodiment of the present invention;
Fig. 2 is the flow chart of pedestrian detection method the second embodiment of the present invention;
Fig. 3 is the high-level schematic functional block diagram of pedestrian detection device first embodiment of the present invention.
The realization of the object of the invention, functional characteristics and advantage will in conjunction with the embodiments, are described further referring to the drawings.
Detailed description of the invention
Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.
The present invention provides a kind of pedestrian detection method.With reference to shown in Fig. 1, for the flow chart of pedestrian detection method first embodiment of the present invention.
In the present embodiment, this pedestrian detection method includes:
Step S10, obtains the video information in the detection region collected based on video monitoring apparatus in real time;
Step S20, resolves the described video information got, and adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information.
Needing the place carrying out pedestrian is monitored, video monitoring apparatus is set, in real time the pedestrian in these places is monitored, such as it is arranged on some and is susceptible to place or other the places being susceptible to security incident such as the bus platform of crowd massing, transport hub or stadiums, to gather pedestrian's video information in these places, wherein, the quantity of video monitoring apparatus is arranged according to actual needs, and above-mentioned video monitoring apparatus can also be city video monitoring system.
Obtain the video information collected based on described video monitoring apparatus, due to each video monitoring apparatus, such as photographic head, the picture scope of its camera lens capture is wider, the region of detection may be needed far beyond user, if being directly analyzed the complete video shot processing, the data volume that needs can be caused to analyze and process is excessive, therefore, for the analyzing and processing realizing video information rapidly and efficiently, pre-set detection region, after getting video information, original video information can be carried out editing process, obtain detecting the video information in region.
Generally, when needing the region to monitoring bigger, need to arrange multiple photographic head, so, the picture in one detection region is accomplished by combining the video that two or more video monitoring apparatus detect and is analyzed, therefore, after the video information getting detection region, the video information got is classified, files storage, in order to subsequent analysis.
The video information in the detection region got is resolved, counts the pedestrian's parameter detecting region in video information, and obtain the behavior characteristics of pedestrian.Specifically, about pedestrian's parameter, following parameter can be included but not limited to: the parameters such as pedestrian's quantity, crowd's flow, crowd density and people's group velocity.
Parsing about video information, can be in the way of combining background modeling and foreground detection, dynamic prospect in video information is analyzed being calculated pedestrian's parameter, first choose with reference to background from video information, then from video information, prospect is obtained according to reference to background, such as, the image photographed for Still Camera in video monitoring has the situation comparing agreement context, can use Codebook algorithm that video information is carried out foreground detection, so that the dynamic prospect in video information is analyzed, to obtain pedestrian's parameter, certainly about foreground detection algorithm, in other embodiments, the algorithm that can also use other realizes.
Specifically, when video resolves, each two field picture of video information can be analyzed, then according to the result of single-frame images detection, and combine each different key states of relevant information record room and time between successive video frames, can be obtained with all environmental informations by mutually switching utilization substantially and reach detection accurate to target, tracking and counting statistics, to realize prospect, i.e. mobile object for background in video information, the behavior analysis of the article etc. that such as pedestrian and pedestrian carry.
Step S30, when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, sends information warning.
Behavior characteristics according to pedestrian's parameter and pedestrian judges to detect in region whether anomalous event occurs, judgement about pedestrian's parameter, user arranges the judgment threshold of pedestrian's quantity, crowd's flow, crowd density and people's group velocity as criterion according to factors such as detection area size, when the pedestrian's parameter detected is more than the threshold value preset, then it is assumed that be anomalous event;Substantial amounts of population surveillance video information can be gathered as sample about behavior characteristics, the human body behavior characteristics storehouse of exploitation anomalous event, classification stores the behavior characteristics of dissimilar anomalous event as reference standard, the behavior characteristics got is contrasted with the behavior characteristics in human body behavior characteristics storehouse, to judge whether anomalous event.Anomalous event includes but not limited to situations below: circumference invasion, protection article violation, article abandon behavior, gateway anisotropy etc..
About circumference intrusion detection, the most often having this kind of event, in meeting-place, unauthorized visitor walks around normal entry and illegally enters into meeting-place from non-porch;On underground railway track, someone jumps into track, illegally enters.These events entering certain area by improper channel are referred to as circumference invasion.
About protection article event detection in violation of rules and regulations, in public places during display articles, around spectacular, often delimit certain protection zone; forbid that other people enter or touch spectacular; but, usually there is also pedestrian and crossed guard rail, entered into protection zone or touch the phenomenon of display articles.
Abandoning behavioral value about article, detect the people in current video and thing, the definition of an abandon that is object is abandoned on one side by a people for a long time, and people also leaves this object simultaneously.In public restaurants, airport or market, this behavior is probably the hidden danger of the attack of terrorism, needs to detect it.
Or other some anomalous events, such as, a large amount of crowds illegally assemble, and need to carry out in time evacuation process etc., will not enumerate at this, and user can be configured as required.When by human body behavioral value, it is judged that when scene state is above-mentioned anomalous event, send a warning, show testing result, to notify that user has anomalous event to occur.
And, carrying out showing and the monitor terminal of dissection process about above-mentioned video information, user can be by monitor terminal to detection region, and some places are imported and exported and the parameter such as whether allowed to pass through and be configured.
Further, before resolving video information, the method is further comprising the steps of:
The described video information got is carried out pretreatment, to eliminate outside environmental elements to the impact of prospect in described detection region.
Owing to Same Scene can be affected by illumination factor, environment tone factor and weather conditions etc., the result that video information is analyzed is caused to produce impact, therefore, before video information is resolved, first video information is carried out pretreatment, eliminate the impact on prospect of the above-mentioned outside environmental elements, wherein, to need the target being tracked analyzing in detection region, the article that such as pedestrian and pedestrian carry etc. are referred to as prospect.
About the pretreatment of video, different pretreatment modes can be used for different influence factors.
Such as, for not fogging clear video, use the mode of image enhancement processing, part interested in prominent image or clarification of objective, some unwanted feature in suppression image, improve the definition of image, that improves image views and admires quality, use histogram equalization, the mode of greyscale transformation, dynamic range of images can be made to increase, contrast is expanded, make image clear, feature is obvious, wherein, greyscale transformation specifically can include that the method that greyscale transformation comprises is a lot, such as converse process, threshold transformation, gray scale stretches, gray scale cutting, gray level correction, the processing modes such as dynamic range adjustment.
In video often due to the existence of gamma characteristic, the luminance distortion of picture signal can be caused, affect picture quality.Therefore this distortion is compensated, i.e. carry out gamma correction;Smothing filtering technology is used to be mainly used in eliminating the noise in image.
In the case of greasy weather, rainy day or haze do not dissipate, owing to the visibility of scene reduces, in image, the feature such as target contrast and color is attenuated, cause the image blurring unclear of life outdoor videos system, the normal work of impact, it is thus desirable to eliminate rainwater, the haze impact on scene image before Video coding, set up fog image degradation model, then by the method for image restoration, greasy weather degraded image is restored, compensate the distortion that degenerative process causes, it is thus achieved that to the optimal estimation without mist image, thus improve Misty Image quality.
When reality is applied, the prospect obtained has usually contained a lot of noise, illumination variation, background object block, mutually blocking between target, the attitude angle change of target, all become the key issue affecting tracking performance, these influence factors from the perspective of signal, it all can regard the noise jamming to target as, is therefore accomplished by the impact eliminating these noises to prospect.In order to eliminate noise, foreground image can be carried out the noise that Connected component analysis understands in foreground image, specifically, carry out opening operation and noise is contracted to zero, closed operation is utilized to rebuild the edge lost, the outline portion using polygon approach to exist, obtain the boundary rectangle of profile.
In actual monitored scene, difference and the change of ambient lighting due to camera angles, in order to obtain the behavior analysis of more accurate prospect, employing can eliminate the texture description operator of illumination interference and video image is carried out background modeling, make to split the prospect obtained more accurate, eliminate the ambient lighting change impact on prospect.
nullThe pedestrian detection method that the present embodiment proposes,Obtain the video information in the detection region that video monitoring apparatus gathers in real time,And the image information got is resolved,Count pedestrian's parameter and the behavior characteristics of pedestrian in detection region,Judge whether scene has anomalous event,Such as fight、Article are left over etc. and to be occurred,When judging to have anomalous event to occur,Information warning can be sent,To remind user that the anomalous event occurred is processed,The present invention is without artificial checking monitoring video,It is capable of monitor video is automatically analyzed getting pedestrian's parameter,And then the scene state in detection region is estimated,Even if in the case of monitor video quantity is bigger,Can be analyzed processing to it equally,Can note abnormalities from substantial amounts of video information event according to pedestrian's parameter,Improve the efficiency of pedestrian detection.
Second embodiment of pedestrian detection method of the present invention is proposed based on first embodiment.With reference to shown in Fig. 2, in the present embodiment, step S20 includes following refinement step:
Step S21, chooses with reference to background from video information, and the described reference background according to choosing obtains prospect from described video information;
Step S22, using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Numerous embodiments is had with foreground detection analysis, such as, by the way of frame difference about with reference to choosing of background, piece image in designated is with reference to background, comparing with reference to background with present frame, filter less difference as required, the result obtained is exactly prospect;Or, by the way of background statistical model, background statistical model is that the background to a period of time is added up, and then calculates its statistical data, such as meansigma methods, average difference, standard deviation, average drifting value etc., using statistical data as the method with reference to background;Or, use the mode of encoding background model, for pixel each in image variation on a timeline, set up multiple (or one) and contain the mobility scale of recent all changes, when detection, current pixel goes to compare with mobility scale, if current pixel falls in the range of any mobility scale, then it it is background.After obtaining with reference to background, being compared with reference to background by present image, the difference drawn is prospect.
Further, between step S21 and step S22, the method is further comprising the steps of:
Use pedestrian's grader that the described prospect in the described detection region in each two field picture of described video information is analyzed;In conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.It is analyzed obtaining characteristic parameter to the behavior of prospect according to grader, substantial amounts of pedestrian's video information can be gathered as sample, exploitation human body behavior analysis algorithms library, set up pedestrian's grader, it both can run on the intelligent high-definition video camera of front end, the monitor terminal on backstage can be run on again, about pedestrian's grader, multiple pedestrian's feature calculation of Adaboost algorithm video information based on magnanimity and setting can be used to obtain.Further, when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, corresponding pedestrian's parameter of described video information and described video information and behavior characteristics are added in described pedestrian's grader of correspondence, as sample for reference, to increase the accuracy of pedestrian's grader classification.
The present invention also proposes a kind of pedestrian detection device.
With reference to shown in Fig. 3, for the high-level schematic functional block diagram of pedestrian detection device first embodiment of the present invention.
In this embodiment, this pedestrian detection device includes:
Acquisition module 10, for obtaining the video information detecting region collected based on video monitoring apparatus in real time;
Parsing module 20, for resolving the described video information got, adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information.
Needing the place carrying out pedestrian is monitored, video monitoring apparatus is set, in real time the pedestrian in these places is monitored, such as it is arranged on some and is susceptible to place or other the places being susceptible to security incident such as the bus platform of crowd massing, transport hub or stadiums, to gather pedestrian's video information in these places, wherein, the quantity of video monitoring apparatus is arranged according to actual needs, and above-mentioned video monitoring apparatus can also be city video monitoring system.
Acquisition module 10 obtains the video information collected based on described video monitoring apparatus, due to each video monitoring apparatus, such as photographic head, the picture scope of its camera lens capture is wider, the region of detection may be needed far beyond user, if being directly analyzed the complete video shot processing, the data volume that needs can be caused to analyze and process is excessive, therefore, for the analyzing and processing realizing video information rapidly and efficiently, pre-set detection region, after getting video information, original video information can be carried out editing process, obtain detecting the video information in region.
Generally, when needing the region to monitoring bigger, need to arrange multiple photographic head, so, the picture in one detection region is accomplished by combining the video that two or more video monitoring apparatus detect and is analyzed, therefore, after the video information getting detection region, the video information got is classified, files storage, in order to subsequent analysis.
The video information in the parsing module 20 detection region to getting resolves, count the pedestrian's parameter detecting region in video information, and obtain the behavior characteristics of pedestrian, specifically, about pedestrian's parameter, following parameter can be included but not limited to: the parameters such as pedestrian's quantity, crowd's flow, crowd density and people's group velocity.
Parsing about video information, parsing module 20 can be in the way of combining background modeling and foreground detection, dynamic prospect in video information is analyzed being calculated pedestrian's parameter, first choose with reference to background from video information, then from video information, prospect is obtained according to reference to background, such as, the image photographed for Still Camera in video monitoring has the situation comparing agreement context, can use Codebook algorithm that video information is carried out foreground detection, so that the dynamic prospect in video information is analyzed, to obtain pedestrian's parameter, certainly about foreground detection algorithm, in other embodiments, the algorithm that can also use other realizes.
Specifically, parsing module 20 is when video resolves, each two field picture of video information can be analyzed, then according to the result of single-frame images detection, and combine each different key states of relevant information record room and time between successive video frames, can be obtained with all environmental informations by mutually switching utilization substantially and reach detection accurate to target, tracking and counting statistics, to realize prospect, i.e. mobile object for background in video information, the behavior analysis of the article etc. that such as pedestrian and pedestrian carry.
Perform module 30, for when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, sending information warning.
Behavior characteristics according to pedestrian's parameter and pedestrian judges to detect in region whether anomalous event occurs, judgement about pedestrian's parameter, user arranges the judgment threshold of pedestrian's quantity, crowd's flow, crowd density and people's group velocity as criterion according to factors such as detection area size, when the pedestrian's parameter detected is more than the threshold value preset, then it is assumed that be anomalous event;Substantial amounts of population surveillance video information can be gathered as sample about behavior characteristics, the human body behavior characteristics storehouse of exploitation anomalous event, classification stores the behavior characteristics of dissimilar anomalous event as reference standard, the behavior characteristics got is contrasted with the behavior characteristics in human body behavior characteristics storehouse, to judge whether anomalous event.Anomalous event includes but not limited to situations below: circumference invasion, protection article violation, article abandon behavior, gateway anisotropy etc..
About circumference intrusion detection, the most often having this kind of event, in meeting-place, unauthorized visitor walks around normal entry and illegally enters into meeting-place from non-porch;On underground railway track, someone jumps into track, illegally enters.These events entering certain area by improper channel are referred to as circumference invasion.
About protection article event detection in violation of rules and regulations, in public places during display articles, around spectacular, often delimit certain protection zone; forbid that other people enter or touch spectacular; but, usually there is also pedestrian and crossed guard rail, entered into protection zone or touch the phenomenon of display articles.
Abandoning behavioral value about article, detect the people in current video and thing, the definition of an abandon that is object is abandoned on one side by a people for a long time, and people also leaves this object simultaneously.In public restaurants, airport or market, this behavior is probably the hidden danger of the attack of terrorism, needs to detect it.
Or other some anomalous events, such as, a large amount of crowds illegally assemble, and need to carry out in time evacuation process etc., will not enumerate at this, and user can be configured as required.When by human body behavioral value, it is judged that when scene state is above-mentioned anomalous event, send a warning, show testing result, to notify that user has anomalous event to occur.
And, carrying out showing and the monitor terminal of dissection process about above-mentioned video information, user can be by monitor terminal to detection region, and some places are imported and exported and the parameter such as whether allowed to pass through and be configured.
Further, this device also includes pretreatment module, for the described video information got being carried out pretreatment, to eliminate outside environmental elements to the impact of prospect in described detection region.
Owing to Same Scene can be affected by illumination factor, environment tone factor and weather conditions etc., the result that video information is analyzed is caused to produce impact, therefore, before video information is resolved, first pretreatment module carries out pretreatment to video information, eliminates the impact on prospect of the above-mentioned outside environmental elements, wherein, to need the target being tracked analyzing in detection region, the article that such as pedestrian and pedestrian carry etc. are referred to as prospect.
About the pretreatment of video, different pretreatment modes can be used for different influence factors.
Such as, for not fogging clear video, use the mode of image enhancement processing, part interested in prominent image or clarification of objective, some unwanted feature in suppression image, improve the definition of image, that improves image views and admires quality, use histogram equalization, the mode of greyscale transformation, dynamic range of images can be made to increase, contrast is expanded, make image clear, feature is obvious, wherein, greyscale transformation is a lot of in the way of specifically can comprising to include greyscale transformation, such as converse process, threshold transformation, gray scale stretches, gray scale cutting, gray level correction, the processing modes such as dynamic range adjustment.
In video often due to the existence of gamma characteristic, the luminance distortion of picture signal can be caused, affect picture quality.Therefore this distortion is compensated, i.e. carry out gamma correction;Smothing filtering technology is used to be mainly used in eliminating the noise in image.
In the case of greasy weather, rainy day or haze do not dissipate, owing to the visibility of scene reduces, in image, the feature such as target contrast and color is attenuated, cause the image blurring unclear of life outdoor videos system, the normal work of impact, it is thus desirable to eliminate rainwater, the haze impact on scene image before Video coding, set up fog image degradation model, then by the mode of image restoration, greasy weather degraded image is restored, compensate the distortion that degenerative process causes, it is thus achieved that to the optimal estimation without mist image, thus improve Misty Image quality.
When reality is applied, the prospect obtained has usually contained a lot of noise, illumination variation, background object block, mutually blocking between target, the attitude angle change of target, all become the key issue affecting tracking performance, these influence factors from the perspective of signal, it all can regard the noise jamming to target as, is therefore accomplished by the impact eliminating these noises to prospect.In order to eliminate noise, foreground image can be carried out the noise that Connected component analysis understands in foreground image, specifically, carry out opening operation and noise is contracted to zero, closed operation is utilized to rebuild the edge lost, the outline portion using polygon approach to exist, obtain the boundary rectangle of profile.
In actual monitored scene, difference and the change of ambient lighting due to camera angles, in order to obtain the behavior analysis of more accurate prospect, employing can eliminate the texture description operator of illumination interference and video image is carried out background modeling, make to split the prospect obtained more accurate, eliminate the ambient lighting change impact on prospect.
nullThe pedestrian detection device that the present embodiment proposes,Obtain the video information in the detection region that video monitoring apparatus gathers in real time,And the image information got is resolved,Count pedestrian's parameter and the behavior characteristics of pedestrian in detection region,Judge whether scene has anomalous event,Such as fight、Article are left over etc. and to be occurred,When judging to have anomalous event to occur,Information warning can be sent,To remind user that the anomalous event occurred is processed,The present invention is without artificial checking monitoring video,It is capable of monitor video is automatically analyzed getting pedestrian's parameter,And then the scene state in detection region is estimated,Even if in the case of monitor video quantity is bigger,Can be analyzed processing to it equally,Can note abnormalities from substantial amounts of video information event according to pedestrian's parameter,Improve the efficiency of pedestrian detection.
Second embodiment of pedestrian detection device of the present invention is proposed based on first embodiment.In the present embodiment, parsing module 20 is additionally operable to choose from video information with reference to background, and the described reference background according to choosing obtains prospect from described video information;And, using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Numerous embodiments is had with foreground detection analysis, such as, by the way of frame difference about with reference to choosing of background, piece image in designated is with reference to background, comparing with reference to background with present frame, filter less difference as required, the result obtained is exactly prospect;Or, by the way of background statistical model, background statistical model is that the background to a period of time is added up, and then calculates its statistical data, such as meansigma methods, average difference, standard deviation, average drifting value etc., using statistical data as the mode with reference to background;Or, use the mode of encoding background model, for pixel each in image variation on a timeline, set up multiple (or one) and contain the mobility scale of recent all changes, when detection, current pixel goes to compare with mobility scale, if current pixel falls in the range of any mobility scale, then it it is background.After obtaining with reference to background, being compared with reference to background by present image, the difference drawn is prospect.
Further, parsing module 20 is additionally operable to: use pedestrian's grader to be analyzed the described prospect in the described detection region in each two field picture of described video information;And, in conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.It is analyzed obtaining characteristic parameter to the behavior of prospect according to grader, substantial amounts of pedestrian's video information can be gathered as sample, exploitation human body behavior analysis algorithms library, set up pedestrian's grader, it both can run on the intelligent high-definition video camera of front end, the monitor terminal on backstage can be run on again, about pedestrian's grader, multiple pedestrian's feature calculation of Adaboost algorithm video information based on magnanimity and setting can be used to obtain.Further, this device also includes adding module, for when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, corresponding pedestrian's parameter of described video information and described video information and behavior characteristics are added in described pedestrian's grader of correspondence, as sample for reference, to increase the accuracy of pedestrian's grader classification.
These are only the preferred embodiments of the present invention; not thereby the scope of the claims of the present invention is limited; every equivalent structure utilizing description of the invention and accompanying drawing content to be made or equivalence flow process conversion; or directly or indirectly it is used in other relevant technical fields, the most in like manner it is included in the scope of patent protection of the present invention.
Claims (10)
1. a pedestrian detection method, it is characterised in that described pedestrian detection method includes:
Obtain the video information in the detection region collected based on video monitoring apparatus in real time;
The described video information got is resolved, adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information;
When the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, send information warning.
Pedestrian detection method the most according to claim 1, it is characterized in that, the described described video information to getting resolves, and before adding up the step of the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information, described pedestrian detection method also includes:
The described video information got is carried out pretreatment, to eliminate outside environmental elements to the impact of prospect in described detection region.
Pedestrian detection method the most according to claim 1 and 2, it is characterised in that the described described video information to getting resolves, the step adding up the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information includes:
Choosing with reference to background from video information, the described reference background according to choosing obtains prospect from described video information;
Using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Pedestrian detection method the most according to claim 3, it is characterized in that, described choosing with reference to background from video information, after the described step obtaining prospect with reference to background from described video information chosen, described pedestrian detection method further comprises the steps of:
Use pedestrian's grader that the described prospect in the described detection region in each two field picture of described video information is analyzed;
In conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.
Pedestrian detection method the most according to claim 4, it is characterized in that, the described described video information to getting resolves, and after adding up the step of the behavior characteristics detecting pedestrian's parameter in region and pedestrian described in described video information, described pedestrian detection method also includes:
When the described pedestrian's parameter according to statistics and the described behavior characteristics of pedestrian judge, in described detection region, anomalous event occurs, described video information and corresponding pedestrian's parameter of described video information and behavior characteristics are added in corresponding described pedestrian's grader.
6. a pedestrian detection device, it is characterised in that described pedestrian detection device includes:
Acquisition module, for obtaining the video information detecting region collected based on video monitoring apparatus in real time;
Parsing module, for resolving the described video information got, adds up pedestrian's parameter and the behavior characteristics of pedestrian detecting region described in described video information;
Perform module, for when the described behavior characteristics of the described pedestrian's parameter according to statistics and pedestrian judges, in described detection region, anomalous event occurs, sending information warning.
Pedestrian detection device the most according to claim 6, it is characterised in that described pedestrian detection device also includes:
Pretreatment module, for carrying out pretreatment to the described video information got, to eliminate outside environmental elements to the impact of prospect in described detection region.
Pedestrian detection device the most according to claim 6, it is characterised in that described parsing module, is additionally operable to choose from video information with reference to background, and the described reference background according to choosing obtains prospect from described video information;
And, using the characteristic parameter of described prospect as described pedestrian's parameter, using the behavior characteristics of described prospect as the described behavior characteristics of pedestrian.
Pedestrian detection device the most according to claim 6, it is characterised in that described parsing module, is additionally operable to use pedestrian's grader to be analyzed the described prospect in the described detection region in each two field picture of described video information;
And, in conjunction with the described pedestrian's grader analysis result to described each two field picture, obtain the behavior characteristics of described prospect.
Pedestrian detection device the most according to claim 6, it is characterised in that described pedestrian detection device also includes:
Add module, for when the described pedestrian's parameter according to statistics and the described behavior characteristics of pedestrian judge, in described detection region, anomalous event occurs, described video information and corresponding pedestrian's parameter of described video information and behavior characteristics are added in corresponding described pedestrian's grader.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610188142.4A CN105825198A (en) | 2016-03-29 | 2016-03-29 | Pedestrian detection method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610188142.4A CN105825198A (en) | 2016-03-29 | 2016-03-29 | Pedestrian detection method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105825198A true CN105825198A (en) | 2016-08-03 |
Family
ID=56525038
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610188142.4A Pending CN105825198A (en) | 2016-03-29 | 2016-03-29 | Pedestrian detection method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105825198A (en) |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106778619A (en) * | 2016-12-16 | 2017-05-31 | 合肥寰景信息技术有限公司 | A kind of pedestrian detection method combined based on laser and vision signal |
CN107346415A (en) * | 2017-06-08 | 2017-11-14 | 小草数语(北京)科技有限公司 | Method of video image processing, device and monitoring device |
CN107920224A (en) * | 2016-10-08 | 2018-04-17 | 杭州海康威视系统技术有限公司 | A kind of abnormality alarming method, equipment and video monitoring system |
CN107977646A (en) * | 2017-12-19 | 2018-05-01 | 北京博睿视科技有限责任公司 | A kind of jube passs quality testing method of determining and calculating |
CN108154108A (en) * | 2017-12-22 | 2018-06-12 | 触景无限科技(北京)有限公司 | The monitoring method and device of a kind of emergency |
CN108491773A (en) * | 2018-03-12 | 2018-09-04 | 中国工商银行股份有限公司 | A kind of recognition methods and system |
CN108665487A (en) * | 2017-10-17 | 2018-10-16 | 国网河南省电力公司郑州供电公司 | Substation's manipulating object and object localization method based on the fusion of infrared and visible light |
CN108846980A (en) * | 2018-06-27 | 2018-11-20 | 肖鑫茹 | A kind of monitoring and alarming system |
CN109087478A (en) * | 2018-08-22 | 2018-12-25 | 徐自远 | A kind of early warning of the anti-swarm and jostlement of intelligence and method of river diversion and system |
CN111294554A (en) * | 2018-12-10 | 2020-06-16 | 丰田自动车株式会社 | Behavior monitoring device, behavior monitoring system, and behavior monitoring program |
CN111435538A (en) * | 2019-01-14 | 2020-07-21 | 上海欧菲智能车联科技有限公司 | Positioning method, positioning system, and computer-readable storage medium |
CN113269144A (en) * | 2021-06-21 | 2021-08-17 | 临沂边锋自动化设备有限公司 | Traffic video monitoring system |
CN114913655A (en) * | 2022-06-16 | 2022-08-16 | 慧之安信息技术股份有限公司 | Museum intelligent security method based on edge calculation |
CN115273154A (en) * | 2022-09-26 | 2022-11-01 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Thermal infrared pedestrian detection method and system based on edge reconstruction and storage medium |
CN115826477A (en) * | 2023-01-31 | 2023-03-21 | 鲸服科技有限公司 | Water area monitoring system and method based on data visualization |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859436A (en) * | 2010-06-09 | 2010-10-13 | 王巍 | Large-amplitude regular movement background intelligent analysis and control system |
CN102164270A (en) * | 2011-01-24 | 2011-08-24 | 浙江工业大学 | Intelligent video monitoring method and system capable of exploring abnormal events |
CN103646257A (en) * | 2013-12-30 | 2014-03-19 | 中国科学院自动化研究所 | Video monitoring image-based pedestrian detecting and counting method |
US20140279740A1 (en) * | 2013-03-15 | 2014-09-18 | Nordic Technology Group Inc. | Method and apparatus for detection and prediction of events based on changes in behavior |
-
2016
- 2016-03-29 CN CN201610188142.4A patent/CN105825198A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859436A (en) * | 2010-06-09 | 2010-10-13 | 王巍 | Large-amplitude regular movement background intelligent analysis and control system |
CN102164270A (en) * | 2011-01-24 | 2011-08-24 | 浙江工业大学 | Intelligent video monitoring method and system capable of exploring abnormal events |
US20140279740A1 (en) * | 2013-03-15 | 2014-09-18 | Nordic Technology Group Inc. | Method and apparatus for detection and prediction of events based on changes in behavior |
CN103646257A (en) * | 2013-12-30 | 2014-03-19 | 中国科学院自动化研究所 | Video monitoring image-based pedestrian detecting and counting method |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107920224B (en) * | 2016-10-08 | 2020-02-18 | 杭州海康威视系统技术有限公司 | Abnormity warning method, abnormity warning equipment and video monitoring system |
CN107920224A (en) * | 2016-10-08 | 2018-04-17 | 杭州海康威视系统技术有限公司 | A kind of abnormality alarming method, equipment and video monitoring system |
CN106778619A (en) * | 2016-12-16 | 2017-05-31 | 合肥寰景信息技术有限公司 | A kind of pedestrian detection method combined based on laser and vision signal |
CN107346415A (en) * | 2017-06-08 | 2017-11-14 | 小草数语(北京)科技有限公司 | Method of video image processing, device and monitoring device |
CN108665487B (en) * | 2017-10-17 | 2022-12-13 | 国网河南省电力公司郑州供电公司 | Transformer substation operation object and target positioning method based on infrared and visible light fusion |
CN108665487A (en) * | 2017-10-17 | 2018-10-16 | 国网河南省电力公司郑州供电公司 | Substation's manipulating object and object localization method based on the fusion of infrared and visible light |
CN107977646A (en) * | 2017-12-19 | 2018-05-01 | 北京博睿视科技有限责任公司 | A kind of jube passs quality testing method of determining and calculating |
CN107977646B (en) * | 2017-12-19 | 2021-06-29 | 北京博睿视科技有限责任公司 | Partition delivery detection method |
CN108154108A (en) * | 2017-12-22 | 2018-06-12 | 触景无限科技(北京)有限公司 | The monitoring method and device of a kind of emergency |
CN108491773A (en) * | 2018-03-12 | 2018-09-04 | 中国工商银行股份有限公司 | A kind of recognition methods and system |
CN108846980A (en) * | 2018-06-27 | 2018-11-20 | 肖鑫茹 | A kind of monitoring and alarming system |
CN109087478A (en) * | 2018-08-22 | 2018-12-25 | 徐自远 | A kind of early warning of the anti-swarm and jostlement of intelligence and method of river diversion and system |
CN111294554A (en) * | 2018-12-10 | 2020-06-16 | 丰田自动车株式会社 | Behavior monitoring device, behavior monitoring system, and behavior monitoring program |
CN111294554B (en) * | 2018-12-10 | 2022-03-08 | 丰田自动车株式会社 | Behavior monitoring device, behavior monitoring system, and non-transitory readable storage medium |
CN111435538A (en) * | 2019-01-14 | 2020-07-21 | 上海欧菲智能车联科技有限公司 | Positioning method, positioning system, and computer-readable storage medium |
CN113269144A (en) * | 2021-06-21 | 2021-08-17 | 临沂边锋自动化设备有限公司 | Traffic video monitoring system |
CN114913655A (en) * | 2022-06-16 | 2022-08-16 | 慧之安信息技术股份有限公司 | Museum intelligent security method based on edge calculation |
CN115273154A (en) * | 2022-09-26 | 2022-11-01 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | Thermal infrared pedestrian detection method and system based on edge reconstruction and storage medium |
CN115826477A (en) * | 2023-01-31 | 2023-03-21 | 鲸服科技有限公司 | Water area monitoring system and method based on data visualization |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105825198A (en) | Pedestrian detection method and device | |
US7382898B2 (en) | Method and apparatus for detecting left objects | |
Albiol et al. | Detection of parked vehicles using spatiotemporal maps | |
US9020261B2 (en) | Video segmentation using statistical pixel modeling | |
US9390333B2 (en) | System and method for robust motion detection | |
US20070122000A1 (en) | Detection of stationary objects in video | |
US20090297023A1 (en) | Video segmentation using statistical pixel modeling | |
US20130162867A1 (en) | Method and system for robust scene modelling in an image sequence | |
US20050168574A1 (en) | Video-based passback event detection | |
CN107229894A (en) | Intelligent video monitoring method and system based on computer vision analysis technology | |
CN105046682A (en) | Video monitoring method based on local computing | |
JP2009517740A (en) | Object density estimation in video | |
KR101949676B1 (en) | syntax-based method of providing intrusion detection in compressed video | |
Ketcham et al. | The intruder detection system for rapid transit using CCTV surveillance based on histogram shapes | |
Husein et al. | Motion detect application with frame difference method on a surveillance camera | |
Aksay et al. | Camera tamper detection using wavelet analysis for video surveillance | |
Huang et al. | Rapid detection of camera tampering and abnormal disturbance for video surveillance system | |
CN1988653A (en) | Night target detecting and tracing method based on visual property | |
KR100920937B1 (en) | Apparatus and method for detecting motion, and storing video within security system | |
Wixson | Illumination assessment for vision-based traffic monitoring | |
CN106921846A (en) | Video mobile terminal legacy detection means | |
Wixson et al. | Improved illumination assessment for vision-based traffic monitoring | |
Chan | A robust target tracking algorithm for FLIR imagery | |
Beghdadi et al. | A perceptual quality-driven video surveillance system | |
Kim et al. | Statistical pattern based real-time smoke detection using DWT energy |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160803 |