CN105578126A - Monitoring camera automatic alarm system - Google Patents
Monitoring camera automatic alarm system Download PDFInfo
- Publication number
- CN105578126A CN105578126A CN201410639997.5A CN201410639997A CN105578126A CN 105578126 A CN105578126 A CN 105578126A CN 201410639997 A CN201410639997 A CN 201410639997A CN 105578126 A CN105578126 A CN 105578126A
- Authority
- CN
- China
- Prior art keywords
- picture
- report
- pictures
- abnormal
- send
- 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
Abstract
The present invention relates to a monitoring camera automatic alarm system and belongs to the pattern recognition and intelligent monitoring field. The monitoring camera automatic alarm system can be used for automatically reporting various kinds of abnormal conditions, such as fires, crimes, natural disasters as well as abnormal natural phenomena and social phenomena, and can be also widely applied to the fields such as scientific research, remote sensing, fight against crimes, disaster prediction, medical treatment, military, astronomy and marine. According to the principles of the monitoring camera automatic alarm system, sample (template) pictures which reflect normal conditions and (or) abnormal conditions are stored in the database of the system in advance, the sample (template) pictures are compared with pictures (screenshots) obtained by a monitoring camera based on a pattern recognition method, and whether to emit an abnormity alarm or report can be decided; if the pictures which are stored in advance are normal pictures, and the pictures are not matched with the pictures obtained by the monitoring camera, an abnormity alarm or report is emitted, otherwise, no reports are emitted; and if the pictures which are stored in advance are abnormal pictures, and the pictures are matched with the pictures obtained by the monitoring camera, an abnormity alarm or report is emitted, otherwise, no reports are emitted.
Description
Technical field
The present invention relates to the fields such as artificial intelligence, pattern recognition and monitoring security protection, object will load onto brain and mouth to current monitored video camera system, makes it not only can observe and record, and can also analyze, report to the police and point out, speak.
Background technology
As everyone knows, camera and electronic eyes everywhere at present, from remote sensing, the unmanned plane of the sky, to mine and the tunnel of underground, they can constantly monitor for 24 hours in real time, and see farther than the mankind, more clear, its observation ability is superman's class far, but these monitored video camera systems only have the function of observing and recording, there is no the function of analysis and report exceptions, its role is just observation, record and deterrent effect also, can only be used for obtain evidence afterwards, can not in time, effectively prevent and prevent the generation of the bad accident such as fire, crime.So prison often can occur and to escape from prison event, near automatic teller machine, crime dramas occur, and forest fire, the flood adverse events such as make river course inrush, if utilize this technology, just can prevent and stop the generation of these hazardous events effectively.
Current video monitoring system uses video recording to add manual supervisory lower grade mode, can only to give a clue or floor manager arrests suspect afterwards, the intelligent monitoring " potentiality " of its early warning is in advance far from being excavated out, automatically by causing warning to the analysis of video image content, key message can not can not be stored selectively.Video monitoring system " not intelligence " will cause the problem of two aspects: one is the expansion along with video monitoring camera scale, and the operating pressure of dig-inning video pictures operator on duty for 24 hours will increase, and also can waste more human resources; Two is that the storage of garbage will consume a large amount of computer storage space, makes the time shorten of useful data storage.The present invention runs into abnormal conditions energy and alarm, prompting, can do special sign, longer-term storage to the video information of this period simultaneously, and emphasis is preserved, thus effectively solves the problem.
In addition, current supervisory control system is except automatically can not reporting above-mentioned known abnormal conditions, more can not report unknown abnormal conditions, therefore can't be used for helping the mankind to find new thread, new situation, new phenomenon, scientific research and scientific discovery can not be used for, " not mating report " in the present invention can notify unknown new thread, new situation at once, can be used for scientific research field, helps people to make scientific discovery.
Summary of the invention
Main purpose of the present invention is: make the first-class watch-dog of original monitoring camera have the functions (intelligence) such as analysis, warning and voice message, as electronic doorkeeper, 24 hours on duty, unusual circumstance can be reported to the police or voice reminder owner (user) immediately, meanwhile, for the video information of abnormal period, automatically make emphasis and indicate, and can preserve for a long time, do not delete.
Principle, method and step.The present invention is as the automatic abnormal alarm of one, reporting system, its principle is mainly: can need the various abnormal conditions that send warnings, report by User Defined, and it is made into sample (template) picture in advance stored in system database, again by the method for pattern recognition compared with picture that monitoring camera obtains (screenshot picture), if the match is successful, then automatically send exception reporting, unsuccessful, do not send report, this is " key player on a team's method ".In addition, user can not need with " method of exhaustion " is self-defined the various normal conditions sending exception reporting yet, and make samples pictures and deposit in the database of system, again by the method for pattern recognition compared with the monitoring institute picture that obtains, if the match is successful, do not send exception reporting, if mate unsuccessful, then automatically send exception reporting, preserve the video data relevant with exception reporting or picture simultaneously, for analysis below, investigation, evidence obtainings etc. are movable to be used, this is " counter select method ", above-mentioned key player on a team's method and anti-method of selecting can simultaneously and use, also can only with one, in general, if be easy to the picture of exhaustive various normal condition, just select method with counter, be suitable for and do not mate warning, report, otherwise, just use key player on a team's method, be suitable for coupling to report to the police, report.Anti-method (" not mating exception reporting ") of selecting wherein can be used for scientific research and scientific discovery, because in this case, the picture of a large amount of reflection normal conditions is stored in advance in database, if the monitoring picture that obtains, picture do not mate with these normal pictures, just automatically can send exception reporting, so just can Timeliness coverage new phenomenon, fresh information, new thread, thus cause new scientific discovery.Except above-mentioned user-defined picture, system self also can carry the general picture of a part, as fire, steal, fight, plunder, the picture of the aspect such as smoking, in advance stored in database, as samples pictures for all types of user such as individual or units.
The method realizing above-mentioned purpose is, in advance in " the samples pictures database " of system stored in normal picture and abnormal picture as samples pictures, if set in advance, prestored normal picture, so, when the picture that camera obtains does not mate with the samples pictures that prestores, just belong to abnormal picture, abnormal situation, just should send abnormal alarm or prompting; Otherwise, if to set in advance, the abnormal picture that prestored does samples pictures, so, when obtaining abnormal picture in camera and prestoring that the match is successful for samples pictures, just belong to abnormal conditions, should send and report to the police or remind.These two kinds of pictures can set simultaneously, be pre-stored in " samples pictures database ", and it is qualitative for mating warning and not mating warning respectively, also can only set, prestore a kind of picture, set a kind of type of alarm, which kind of is applied to prestore pictorial manner and type of alarm actually, this will determine according to the concrete feature in different industries and place, if situation is simple, normal picture and fewer, be easy to exhaustive, so just pre-deposit normal picture, application does not mate warning, such as, in astronomical observation field, in most cases, star is all actionless, occur that the situation of meteor is little, so just there is no the picture of meteor as normal picture, stored in database, application does not mate warning, like this, once there is the picture of meteor, do not mate with there is no the template picture of meteor, just report immediately, coupling is not just reported, otherwise, if situation is complicated, normal pictures is a lot, be difficult to exhaustive, so just only application coupling report manner, pre-deposits abnormal picture, such as fire, steal, fight, plunder, the abnormal picture such as smoking, application coupling is reported to the police, and just sends abnormal alarm or prompting, do not mate, do not report to the police when the picture obtained in camera and these abnormal pictures match.Not many at normal picture, can exhaustive when, also can simultaneously stored in normal picture and abnormal picture, and application coupling is simultaneously reported to the police and is not mated warning two kinds of methods, in this case wrong report can't be produced, this is because, if what monitoring camera photographed is a normal picture, so no matter it and the normal picture in database carry out mating or mating with abnormal picture, its result all can not cause warning, otherwise, if what monitoring camera photographed is an abnormal picture, so, no matter it mates with the normal picture in database or abnormal picture, capital causes warning.Because the picture in database only has two classes---normal picture and abnormal picture, normal picture is, you and I match and just do not report to the police, and do not match and just report to the police; Then, you and I match and just report to the police abnormal picture, do not match with regard to not reporting to the police.Like this, normal picture, no matter encounter the normal picture in database or abnormal picture, all can not cause warning, otherwise, abnormal picture, encounter the normal picture in database or abnormal picture all can cause warning, while warning, corresponding video also also can be made distinctive mark, longer-term storage, can not be deleted automatically by system.
Application.Application of the present invention widely, every place having monitoring camera, all can use this technology, can be widely used in the places such as bank, prison, market, warehouse, mine, tunnel, station safety check, and the field such as traffic, school, public security, scientific research, medical treatment, military affairs, national defence, unmanned plane, robot, astronomy, ocean, universe space, radio telescope.
From spatially, every place having camera, the place of all genus human viewable is all the range of application of its present and the future from the whole earth to universe space.Specifically, from seabed to ocean surface, from Plain to high mountain, from low latitude to space, all can there is the application of this technology again to cosmic space.
Say that from object the object of observation of every mankind is all applicable to this technology, because what the mankind observe, just need what is monitored, therefore can find oneself requirement from monitoring.And from the atom of microcosmic, microbe to the celestial body of macroscopic view and universe, all belong to the range of observation of the mankind.
From the time, along with high-tech development such as unmanned plane, robot, microrobot and space technologys, the mankind observe and the scope of monitoring can be increasing, in addition, along with the development of urbanization, originally also can not come into operation more and more in the remote districts, rural area of usage monitoring camera, therefore, the range of application of this invention technology also can be more and more wider, and user can get more and more.
Below several emphasis application:
1, scientific research institution.Scientific discovery starts from observation, observes the beginning that anomaly is scientific discovery, and the present invention can be used for finding, record in reporter assay room and various anomalies in production process, automatically provide valuable information, and therefrom make scientific discovery.
2, unmanned plane and robot, the two is movable monitoring camera, and therefore, the scope of observation is larger, and angle is wider, and therefore, the exception reporting that can make is also more valuable.
3, satellite remote sensing.Carried out the observation of macroscopic view by satellite remote sensing, note abnormalities phenomenon, sends report in time, contributes to the information finding that there is substantial worth, and therefrom make scientific discovery, also contributes to Timeliness coverage and some natural calamities of prevention.
4, oceanographic observation.Can be used for the observation in ocean surface and seabed, monitoring, unusual circumstance notifies at once, thus helps lend some impetus to the development of Marine Sciences, and the natural calamities such as prevention submarine earthquake, tsunami, typhoon are to the harmful effect of the mankind.
5, medical health system.Can be used for health check-up and medical diagnosis on disease etc., Timeliness coverage disease risk, when microrobot also can be sent future to enter human body, row is observed, and makes exception reporting, the pathology in Timeliness coverage human body.
6; prison system; comprise the prison of various places, detention house, the house of detention etc.; all in the urgent need to this technology, because often to fight event in the room closed and courtyard wall, also to prevent the situation that some people commits suiside because taking things too hard for the moment simultaneously; to climb over the walls in addition the situation etc. of escaping from prison; if there is this exception reporting system, Timeliness coverage potential safety hazard can be contributed to, prevent in time in prison system various illegal, obtain crime and be.
7, banking system.Often can plunder near the ATM of bank, steal and illegally install and steal the behavior that user cipher even copies bank card device, these abnormal conditions all need Timeliness coverage and are reported to associated safety department.Such as, CCTV's Focus program on October 26th, 2014 reports the stolen case of bank card betiding Xi'an on December 17th, 2013, and Xi'an one by one 14 banks all reports a case.The amount of money of stolen brush amounts to 7,400,000, the masses that are injured reach 168 people, reason is that offender has installed the electronic installation stolen user cipher He copy bank card brazenly on automatic teller machine, and the monitoring camera equipment of bank performs practically no function, can not make and reporting to the police timely, can only for transferring videograph afterwards, report to the police or prompting if above-mentioned monitored video camera system can send in time, so the moon of offender would not be succeeded.
Embodiment.
One, software development.Develop a mode identificating software system with image data base, this system can be connected with the first-class monitoring hardware equipment of monitoring camera by computer.
Two, set picture group to be identified (samples pictures group) and corresponding to report to the police, voice prompt.The general samples pictures of a part can be carried by system, as fire, theft, plunder, the picture of the hazardous accidents aspect such as to fight, in advance stored in systematic sample picture database, in addition, also should have the samples pictures that user independently sets, also it stored in samples pictures database, the method for above-mentioned two kinds of main bodys setting picture can be, 1, from the monitor video record in the past preserved, transfer warning, prompting picture.Checking monitoring record, therefrom find out the one section of video needing to be set as that dangerous scene or needs are reminded in time, then be a picture group sheet by computer its automatic screenshotss in chronological order, usually this picture group sheet is the decomposition of a sequence, such as, the action picture of first second, the action picture of the 2nd second, 3rd second, the action picture etc. of the 4th second, and be every picture group sheet a kind of common alarm sound of setting or voice prompt, what sound concrete independently can be set by user, such as, the gate monitoring of a school, when photographing car that principal opens distinctive mark to school gate, after image recognition success (such picture will take in advance and stored in computer picture library), will send " principal drive come, please open the door " preset sound be prompted to school entrance guard, and for example, the monitoring camera of one family, when having photographed relative or friend to entrance, after pattern recognition success, can automatically send " so-and-so (name) to doorway, please open the door " default voice message.Also can be that one group of similar or coherent picture (as fire picture, theft, robbery picture) sets same suggestion voice or audio warning.Such as, " event of fighting ", " someone will commit suiside " " having caught fire " etc., often the picture of group (different groups) can set different warnings, suggestion content, and it also can be set to type of alarm corresponding to multiple, different rank stored in the rank reported to the police in " the samples pictures data Kuku " of system by the order of severity; 2, warning, prompting picture is found from other video data, comprise the video data that public security department and other unit or individual provide, and the video data downloaded from the Internet (associated video data is found in available keyword search), the video data etc. searched out in film, after finding these videos, therefrom select the fragment oneself needed again, obtain one group of coherent screenshot picture with screenshotss software, then set warning, voice prompt, stored in database as samples pictures group; 3, look for expert to paint, draw and oneself need setting, and the picture never occurred in monitoring record, also can draw multiple coherent or similar picture and form one group, and set common warning, voice prompt for it, and stored in sample database; 4, by the method for true man's manoeuvre, look for performer to play the part of the warning picture of needs setting, take into one section of video, and thus screenshotss obtain one group of coherent picture, setting is reported to the police, voice prompt, stored in systematic sample picture library; 5, outdoor scene manoeuvre shooting method.Be mainly used to manoeuvre or shooting fire, earthquake, mud-rock flow, typhoon, etc. the natural video frequency that do not have personage to occur such as natural calamity and traffic congestion, and therefrom screenshotss, obtain one group of screenshot picture, formulate report to the police, voice prompt, stored in database, as samples pictures.6, in Baidu's picture searching, the picture oneself needed is gone out with corresponding keyword search, as, search out some pictures relevant with fire with " fire ", search out some with " robbery " and plunder relevant picture, forming a picture group by these similar pictures, and set common warning, voice prompt for this picture group sheet, then stored in systematic sample picture database.7, above-mentioned 6 kinds source and method in, also can set the single picture do not distributed in groups, stored in database as samples pictures.
In addition, samples pictures all in said system database also all will indicate it is " normal picture " or " abnormal picture ", because the two is identified, mate rear caused result is different, normal picture sends exception reporting when being and mating unsuccessful, abnormal picture is then send exception reporting when the match is successful, and report or the mode of reporting to the police can by user's sets itself.
Three, to samples pictures mark emphasis recognition site, and provide feature interpretation, determine the essential condition identified.In order to improve the success rate of identification, some samples pictures in database needs to mark out emphasis characteristic portion, if you would the face (crime scene) of black cloth, hold the hand making case instrument, the black smoke of emerging (fire scenario), contain the mouth of cigarette, clip the finger (smoking scene) etc. of cigarette, contribute to like this reducing the difficulty identified, efficiency and the success rate of identification can be improved simultaneously.The mode of mark can be with red line, the corresponding site of picture is enclosed.In addition, also need to make feature interpretation to samples pictures, determine the essential condition identified, this is because, the essence identified determines whether a certain things belongs to a certain concept and range, any one concept has its feature, the i.e. difference of difference and other concept, in feature interpretation, should write exactly, which can be different with samples pictures, and which must be identical with samples pictures, namely to catch essential condition (as people, some action etc. of people), ignore non-essential condition (background etc. as personage).Essential condition is the standard and key that identify, such as, in fighting, strike must be had, collide, fall down to the ground, " cigarette end smokily " and " mouth " in smoking, must be had, thief must be sneaking, the action such as to peep all round, preferably these standards are demonstrated out with the form of video, then screenshotss are one group of coherent picture, stored in database, implement for during identification.
Four, the method for pattern recognition.Mode identification method mainly contains and following four kinds: Zheng Xuanfa, instead selects method, free hand drawing (identification) matching method, many figure (identification) matching methods.So-called key player on a team's method refers to, in advance in a database stored in abnormal picture (pictures of reflection abnormal conditions), then identify mutually with screenshot picture that monitoring camera obtains, mate, automatically send when the match is successful and report to the police or report, this is key player on a team's method; The anti-method of selecting of what is called refers to, in advance in a database stored in normal picture (picture of reflection normal condition), then identify mutually with screenshot picture that monitoring camera obtains, mate, when mating unsuccessful, send and report to the police or report, this is anti-method of selecting; Put it briefly, key player on a team's method is exactly send report when the match is successful, counter select rule be coupling unsuccessful time send report.Normal picture refers to the picture of reflection normal condition, and abnormal picture refers to the picture of reflection abnormal conditions.Key player on a team's method and anti-method of selecting can be carried out simultaneously, in database also can in advance simultaneously stored in normal picture and abnormal picture as samples pictures.So-called free hand drawing matching method refers to, removes all single picture in one by one matching database by any one picture in screenshot picture group, comprises each picture in " the samples pictures group " in database and single picture that is scattered, that do not distribute in groups, called number map matching method is named again " Corresponding matching method ", specifically refer to, screenshot picture group and samples pictures group are carried out corresponding coupling as a whole removing respectively, specifically, be exactly from left to right, go to mate the 1st picture in " samples pictures " with the picture of the 1st in " screenshot picture group ", the 2nd picture in matched sample picture group is removed with the picture of the 2nd in screenshot picture group, the 3rd picture in matched sample picture group is removed with the picture of the 3rd in screenshot picture group, the like, until mated, if there is the matching degree of multipair youngster's picture all very high continuously, so just can determine that the match is successful.This matching process is similar to the left justify two pictures, lower overlap, so cry again " positive peak match ", just like overlapping from left to right for two wavy curve figure, primary peak below primary peak correspondence above, the secondary peak below secondary peak correspondence above, the like.Many figure method of identification is equivalent to " video identification ", or " action, Activity recognition ", and its essence goes to identify, mate another section of video with one section of video, and therefore, the accuracy rate of many figure method of identification and success rate are higher than free hand drawing method of identification.If in free hand drawing matching method, matching degree will reach 80% and just can judge to identify successfully, so, in many figure matching method, as long as matching degree reaches 60%, can judge to identify successfully.Above-mentioned free hand drawing coupling and many figure couplings can only choosing one wherein identify, also can select two kinds of identifications that hocket.Above-mentioned " samples pictures group " with the difference of " screenshot picture group " is, the former obtains video data through artificial editing, then with the automatic screenshotss of computer, it is the decomposition of a complete action, and the latter is computer (as 1 minute, 2 minutes) editing automatically in units of the time, then automatic screenshotss, are exactly not necessarily the decomposition of a complete action, but are also likely the decomposition of a complete action.Like this, when carrying out above-mentioned " positive peak match ", the correspondence likely made the mistake, namely the correspondence automatically completed by computer is not being corresponding in turn to during realistic operation decomposes, but, this can by changing original corresponding order, repeatedly attempt error correction, as, originally upper 1 to lower 1, upper 2 to lower 2, upper 3 to lower 3, change now 2 into lower 3, upper 3 to lower 4, upper 4 to lower 5, Deng, the like, this cries " matching method of avoiding the peak hour ", in this matching method, the left end of two pictures does not also line up, but after stagger one or several picture, start to align successively, such thing, if come by people, very slow, very difficult, but come not difficult by computer, because the arithmetic speed of computer is very fast.Avoiding the peak hour in coupling, if there is the matching degree of multipair youngster's picture all very high continuously, so just can determine that the match is successful.
A threshold value can be set when identifying in addition, when similarity exceedes this threshold value, identify, the match is successful, do not reach this threshold value, identify, mate unsuccessful, this threshold value also by the concrete condition of user according to oneself, independently can adjust, because threshold value is too high, the abnormal conditions of this report can be missed, otherwise threshold value is too low, then can bring unnecessary warning, the bothering of report.In many figure couplings, if there is the similarity threshold (matching degree) of multipair youngster's picture all very high continuously, so just can determine that the match is successful.The accuracy of many figure method of identification higher than free hand drawing method of identification, so the threshold value of many figure method of identification can lower than the threshold value of free hand drawing method of identification.Relation between above-mentioned four kinds of recognition methodss is, key player on a team's method and anti-choosing in method can comprise free hand drawing method of identification and Duo Tu method of identification, namely the two can use free hand drawing method of identification and Duo Tu method of identification to go to identify, and the one that both can be used alone in free hand drawing method of identification or many figure method of identification, also can the two simultaneously and use.
Five, the process after identification.1, if the samples pictures in database is normal picture, so, identify and do not need report successfully, do not send any sound, bring unnecessary bothering with exempt from customs examination user.Otherwise, identify unsuccessful, report be sent, namely send alarm sound or suggestion voice, concrete alarm sound and prompting voice, can by the picture sets itself of user according to different content, the rank of warning also can be set to multiple as the case may be, such as, one-level is reported to the police, directly can be reported to public security department, secondary alarm, internal institution personnel can be offered, three grades of warnings, then by loudspeaker to abnormal behaviour person with voice warning etc.; 2, if the samples pictures in database is abnormal picture, so, identify and will send report successfully, otherwise, identify unsuccessful, just do not send any report.3, in above-mentioned situation, as run into urgent, serious situation, while sending warning, also will be sent on the mobile phone that an information and screenshot picture specify to user by wireless network, that section of warning video that the match is successful also will provide emphasis and indicate, emphasis is preserved, and longer-term storage is not deleted; 4, in above-mentioned situation, as run into not too serious but needing to provide the situation of prompting, then send voice prompt, as " so-and-so guest has come, is waiting for outdoors, please before go enabling " etc.
Claims (9)
1. the intelligent monitor system allowing monitoring camera possess analysis ability and automatic alarm (announcement) ability, as one automatically " exception reporting system ", it is characterized in that, in advance in the database of system stored in sample (template) picture of reflection normal condition and (or) abnormal conditions, then method for distinguishing is known in mode, compared with picture that monitoring camera obtains (screenshot picture), can determine whether send abnormal alarm, report, if in advance stored in be normal picture, abnormal alarm is sent when then mating unsuccessful, report, any report is not sent when the match is successful, otherwise, if in advance stored in be abnormal picture, warning, report is sent when then the match is successful, when mating unsuccessful, do not send any report, send and report to the police or report, also can automatically preserve the video data relevant with exception reporting simultaneously, for analysis below, investigation, collect evidence, to solve a case etc. activity.
2. according to claim 1, pattern-recongnition method wherein can be divided into " key player on a team's method " and " counter select method " two kinds of recognition methodss, so-called key player on a team's method refers to, in advance in a database stored in abnormal picture, then compared with screenshot picture that monitoring camera obtains, automatically send when the match is successful and report to the police or report, this is key player on a team's method, the anti-method of selecting of what is called refers to, in advance in a database stored in normal picture, then compared with screenshot picture that monitoring camera obtains, when mating unsuccessful, send and report to the police or report, this is anti-method of selecting, put it briefly, key player on a team's method is exactly send report when the match is successful, counter select rule be coupling unsuccessful time send report, normal picture refers to the picture of reflection normal condition, abnormal picture refers to that the picture of reflection abnormal conditions is (as fire, obtaining crime is), key player on a team's method and anti-method of selecting can be carried out simultaneously, also can pre-deposit normal picture and abnormal picture as samples pictures in database simultaneously.
3. according to claim 1, identification wherein can be divided into again " free hand drawing coupling " and " scheming coupling " good for both sides recognition methods more, so-called free hand drawing coupling refers to, go all samples pictures in one by one matching database by any one picture in screenshot picture group, comprise each picture in " the samples pictures group " in database and the single picture do not distributed in groups, called number map coupling refers to, " screenshot picture group " and " samples pictures group " is carried out corresponding coupling as a whole removing respectively, specifically, be exactly from left to right, the 1st picture in matched sample picture group is removed with the picture of the 1st in screenshot picture group, the 2nd picture in matched sample picture group is removed with the picture of the 2nd in screenshot picture group, the 3rd picture in matched sample picture group is removed with the picture of the 3rd in screenshot picture group, the like, until mated, if there is the matching degree of multipair youngster's picture all very high continuously, reach a certain similarity threshold, so just can determine that the match is successful, this is " positive peak match " in many figure couplings, can have in addition " coupling of avoiding the peak hour ", namely two pictures left end and do not line up, but it is initial from left end, stagger the coupling starting after one or several picture to align successively, this is also a kind of many figure matching method of entirety, above-mentioned free hand drawing coupling and many figure couplings can only choosing one wherein identify, also optional two kinds of identifications that hocket.
4. according to claim 3, source and the difference of samples pictures group and screenshot picture group are, the former (samples pictures group) obtains video data through artificial editing, then with the automatic screenshotss of computer, be the decomposition of a complete action, and the latter's (screenshot picture group) is computer (as 1 minute, 2 minutes) automatically editing in units of the time, then screenshotss automatically, so " screenshot picture group " is exactly not necessarily the decomposition of a complete action.
5. according to claim 1, a threshold value can be set during identification, when similarity exceedes this threshold value, identify, the match is successful, do not reach this threshold value, identify, mate unsuccessful, this threshold value also independently can be adjusted according to the concrete condition of oneself by user, because threshold value is too high, the abnormal conditions of this report can be missed, otherwise threshold value is too low, then can bring unnecessary warning, the bothering of report.
6. according to claim 1, samples pictures in database all will mark emphasis recognition site, if you would the face (crime scene) of black cloth, hold the hand making case instrument, the black smoke of emerging (fire scenario), contain the mouth of cigarette, clip the finger (smoking scene) etc. of cigarette, contribute to like this reducing the difficulty identified, accuracy and the success rate of identification can be improved simultaneously, the mode of mark can be with red line, the corresponding site of picture is enclosed, also to provide feature interpretation simultaneously, determine the essential condition identified.Essential condition is identified as the necessary condition of merit, is the standard and key that identify, such as, in fighting, collision must be had, fall down to the ground, " cigarette end smokily " and " mouth " in smoking, must be had, thief must have the essential conditions such as sneaking, the action of peeping all round.
7. according to claim 2, sample (template) picture in said system database, one of normal picture or abnormal picture two can be only had, also can deposit these two kinds of pictures simultaneously, must indicate it is " normal picture " or " abnormal picture " in this case, because result caused after the two is identified, the match is successful is different, send exception reporting when normal picture is and mates unsuccessful, abnormal picture is then send exception reporting when the match is successful.
8. according to claim 1, send warning, the mode of report can by the picture sets itself of user according to different content, and the different stage of warning can be set, such as, one-level is reported to the police, directly can be reported to public security department, secondary alarm, internal institution personnel can be offered, three grades of warnings, then by loudspeaker to abnormal behaviour person with voice warning etc., send serious to report to the police, also to send on the mobile phone that an information and screenshot picture specify to user by wireless network simultaneously, the match is successful or mate unsuccessful that section of anomalous video also will provide emphasis sign, emphasis is preserved, longer-term storage.As run into not too serious but needing to provide the situation of prompting, then send voice prompt, as " so-and-so guest has come, is waiting for outdoors, please before go enabling " etc.
9., according to claim 1, the samples pictures in system database both can carry the general picture of a part by system, as fire, steal, fight, the samples pictures of the aspect such as robbery, also independently can be set according to the special circumstances of oneself by user; Warning after identification, report manner, equally also can carry the general type of alarm of a part (as conventional alarm sound and voice message sound etc.) by system, also independently can be set according to the special circumstances of oneself by user, therefore, samples pictures in database and type of alarm all can be formed by User Defined, dynamic, variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410639997.5A CN105578126A (en) | 2014-11-11 | 2014-11-11 | Monitoring camera automatic alarm system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410639997.5A CN105578126A (en) | 2014-11-11 | 2014-11-11 | Monitoring camera automatic alarm system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105578126A true CN105578126A (en) | 2016-05-11 |
Family
ID=55887728
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410639997.5A Pending CN105578126A (en) | 2014-11-11 | 2014-11-11 | Monitoring camera automatic alarm system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105578126A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106408854A (en) * | 2016-11-02 | 2017-02-15 | 北京弘恒科技有限公司 | Garden invasion monitoring and early-warning system |
| CN106534784A (en) * | 2016-11-22 | 2017-03-22 | 苏州航天系统工程有限公司 | Acquisition analysis storage statistical system for video analysis data result set |
| CN107169448A (en) * | 2017-05-12 | 2017-09-15 | 甘肃政法学院 | A kind of fusion treatment effective information is simultaneously dynamically caught and method for early warning |
| CN107240213A (en) * | 2017-07-11 | 2017-10-10 | 合肥弘恩机电科技有限公司 | A kind of warehousing management intelligent early-warning method |
| CN107613256A (en) * | 2017-09-26 | 2018-01-19 | 珠海市领创智能物联网研究院有限公司 | A kind of monitoring implementation method of smart home |
| CN109040718A (en) * | 2018-10-16 | 2018-12-18 | 陈桂苹 | Intelligent monitor system based on web camera |
| CN109413369A (en) * | 2017-08-17 | 2019-03-01 | 孟思宏 | A kind of monitor video intellectual analysis early warning platform |
| CN109697989A (en) * | 2018-12-26 | 2019-04-30 | 深圳创维数字技术有限公司 | Audio data processing method, server and readable storage medium storing program for executing |
| CN109857114A (en) * | 2019-02-26 | 2019-06-07 | 温州市职业中等专业学校 | Campus Mobile Intelligent Robot and campus intelligent management system |
| CN109905644A (en) * | 2017-12-08 | 2019-06-18 | 孟思宏 | A kind of sewage treatment plant's monitor video intellectual analysis early warning platform |
| CN109991655A (en) * | 2019-04-26 | 2019-07-09 | 上海天诚通信技术股份有限公司 | A kind of earthquake monitoring device and seismic monitoring method |
| CN110851329A (en) * | 2019-11-12 | 2020-02-28 | 黑龙江电力调度实业有限公司 | Machine room state diagnosis system and method |
| CN111126153A (en) * | 2019-11-25 | 2020-05-08 | 北京锐安科技有限公司 | Safety monitoring method, system, server and storage medium based on deep learning |
| CN111246175A (en) * | 2020-01-20 | 2020-06-05 | 深圳英飞拓科技股份有限公司 | Rule-based face snapshot camera voice processing method, system and device |
| CN111523797A (en) * | 2020-04-21 | 2020-08-11 | 深圳中科慧据科技有限公司 | Rail transit early warning event scheduling method and device, computer equipment and storage medium |
| CN111768583A (en) * | 2019-04-02 | 2020-10-13 | 上海博泰悦臻网络技术服务有限公司 | Vehicle fuel oil anti-theft method and system and vehicle |
| CN112732965A (en) * | 2019-10-28 | 2021-04-30 | 中国银联股份有限公司 | Abnormal event detection method and detection system based on image description generation |
| CN113992893A (en) * | 2021-10-26 | 2022-01-28 | 福建云知声智能科技有限公司 | Park inspection method and device, storage medium and electronic device |
| CN116187651A (en) * | 2022-12-14 | 2023-05-30 | 深圳技术大学 | Method and device for generating supervision report of cargo state on arrival ship |
-
2014
- 2014-11-11 CN CN201410639997.5A patent/CN105578126A/en active Pending
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106408854A (en) * | 2016-11-02 | 2017-02-15 | 北京弘恒科技有限公司 | Garden invasion monitoring and early-warning system |
| CN106534784A (en) * | 2016-11-22 | 2017-03-22 | 苏州航天系统工程有限公司 | Acquisition analysis storage statistical system for video analysis data result set |
| CN107169448A (en) * | 2017-05-12 | 2017-09-15 | 甘肃政法学院 | A kind of fusion treatment effective information is simultaneously dynamically caught and method for early warning |
| CN107240213A (en) * | 2017-07-11 | 2017-10-10 | 合肥弘恩机电科技有限公司 | A kind of warehousing management intelligent early-warning method |
| CN109413369A (en) * | 2017-08-17 | 2019-03-01 | 孟思宏 | A kind of monitor video intellectual analysis early warning platform |
| CN107613256A (en) * | 2017-09-26 | 2018-01-19 | 珠海市领创智能物联网研究院有限公司 | A kind of monitoring implementation method of smart home |
| CN109905644A (en) * | 2017-12-08 | 2019-06-18 | 孟思宏 | A kind of sewage treatment plant's monitor video intellectual analysis early warning platform |
| CN109040718A (en) * | 2018-10-16 | 2018-12-18 | 陈桂苹 | Intelligent monitor system based on web camera |
| CN109040718B (en) * | 2018-10-16 | 2020-07-03 | 广州市信时通科技有限公司 | Intelligent monitoring system based on network camera |
| CN109697989A (en) * | 2018-12-26 | 2019-04-30 | 深圳创维数字技术有限公司 | Audio data processing method, server and readable storage medium storing program for executing |
| CN109697989B (en) * | 2018-12-26 | 2021-03-16 | 深圳创维数字技术有限公司 | Audio data processing method, server and readable storage medium |
| CN109857114A (en) * | 2019-02-26 | 2019-06-07 | 温州市职业中等专业学校 | Campus Mobile Intelligent Robot and campus intelligent management system |
| CN111768583A (en) * | 2019-04-02 | 2020-10-13 | 上海博泰悦臻网络技术服务有限公司 | Vehicle fuel oil anti-theft method and system and vehicle |
| CN109991655A (en) * | 2019-04-26 | 2019-07-09 | 上海天诚通信技术股份有限公司 | A kind of earthquake monitoring device and seismic monitoring method |
| CN112732965A (en) * | 2019-10-28 | 2021-04-30 | 中国银联股份有限公司 | Abnormal event detection method and detection system based on image description generation |
| CN110851329A (en) * | 2019-11-12 | 2020-02-28 | 黑龙江电力调度实业有限公司 | Machine room state diagnosis system and method |
| CN111126153A (en) * | 2019-11-25 | 2020-05-08 | 北京锐安科技有限公司 | Safety monitoring method, system, server and storage medium based on deep learning |
| CN111246175A (en) * | 2020-01-20 | 2020-06-05 | 深圳英飞拓科技股份有限公司 | Rule-based face snapshot camera voice processing method, system and device |
| CN111523797A (en) * | 2020-04-21 | 2020-08-11 | 深圳中科慧据科技有限公司 | Rail transit early warning event scheduling method and device, computer equipment and storage medium |
| CN113992893A (en) * | 2021-10-26 | 2022-01-28 | 福建云知声智能科技有限公司 | Park inspection method and device, storage medium and electronic device |
| CN116187651A (en) * | 2022-12-14 | 2023-05-30 | 深圳技术大学 | Method and device for generating supervision report of cargo state on arrival ship |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105578126A (en) | Monitoring camera automatic alarm system | |
| Paraskevas | Aligning strategy to threat: a baseline anti‐terrorism strategy for hotels | |
| Ogunleye et al. | A computer-based security framework for crime prevention in Nigeria | |
| Strom et al. | Building on clues: Examining successes and failures in detecting US terrorist plots, 1999–2009 | |
| Manning | A view of surveillance | |
| Nobo et al. | Natural disasters and crime: Criminological lessons from Hurricane Katrina | |
| Bahir et al. | Real‐Time M ajor E vents M onitoring and A lert S ystem through Social Networks | |
| Pethő-Kiss | Countering Terrorist Acts against Christian Places of Worship | |
| Auyero | Clandestine connections: the political and relational makings of collective violence | |
| Hernandez et al. | Improving School Safety in the E-Learning Era | |
| Piątek | Technology “democratization”. Peacetech–new quality of security management | |
| Arnett | Black Lives Monitored | |
| Norton | Primary Law Enforcement Mistakes during Initial Critical Incident Response and Timeline of These Events Anatomy of the First 60 | |
| Rich | Machines as Crime Fighters-Are You Ready | |
| Monyeki | Data mining to analyse recurrent crime in South Africa | |
| Collingburn et al. | CT QUICK LOOK | |
| Townsend et al. | Intelligence-Led Mitigation | |
| Matuszek | CCTV systems–technological and legal aspects. The present and the prospects for future | |
| Hunt | Technology and the American criminal justice system | |
| Flynn et al. | Report of the Haiti Accompaniment Project, June 29-July 9, 2004 | |
| Budisalich et al. | MAKING SCHOOL SAFETY AFFORDABLE: APPLYING TECHNOLOGY AND COMMONSENSE TO PROTECT STUDENTS IN ECONOMICALLY CHALLENGED SCHOOL SYSTEMS | |
| Schauer et al. | Detecting a Complex Attack Scenario in an Airport: The PRAETORIAN Framework | |
| Welsh et al. | Is CCTV Effective in Preventing Crime in Public Places? | |
| ZB et al. | Document Version: D3. 8 LETSCROWD ESM implementation guidelines for crowd protection 1.00 | |
| Oyer et al. | More words, less action: A framing analysis of FEMA public relations communications during hurricanes Katrina and Gustav |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: Du Xiangyang Document name: Notification of Publication of the Application for Invention |
|
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160511 |