CN104077591A - Intelligent and automatic computer monitoring system - Google Patents
Intelligent and automatic computer monitoring system Download PDFInfo
- Publication number
- CN104077591A CN104077591A CN201310101688.8A CN201310101688A CN104077591A CN 104077591 A CN104077591 A CN 104077591A CN 201310101688 A CN201310101688 A CN 201310101688A CN 104077591 A CN104077591 A CN 104077591A
- Authority
- CN
- China
- Prior art keywords
- human body
- membership
- domain
- fuzzy
- behavior
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses an intelligent and automatic computer monitoring system, which relates to a specific human body abnormal behavior identification method based on a joint model and belongs to the field of pattern recognition, artificial intelligence and computer vision. A camera is adopted as a video capture device, and a server is used as a video processing device. The method comprises the following steps: firstly, detecting a moving object, and obtaining a binary image of the moving object; judging whether a centre of mass hops or not; if the centre of mass hops, calculating a trajectory of the centre of mass, and otherwise, extracting a human body joint model; then, extracting an angle parameter of the human body joint model; comparing according to a degree of membership of the angle parameter and a regular database; and identifying a human body behavior to judge whether the human body behavior is abnormal or not. A specific human body abnormal behavior can be accurately identified by the method.
Description
Technical field
The invention belongs to pattern-recognition and artificial intelligence and computer video field, particularly a kind of specific human body abnormal behaviour recognition methods based on joint model.
Background technology
In recent years, video monitoring is being brought into play increasing effect in the fields such as public security, fire-fighting and traffic in city.And the high speed development of the electronic information such as computing machine, communication, pattern-recognition and artificial intelligence and signal processing technology, for the quick progress of monitoring technique provides powerful support.The manual supervisory mode of the many employings of tradition supervisory system, its all deficiency is the Main Bottleneck that video monitoring further develops.The intelligent video monitoring system of setting up based on artificial intelligence technology, for the development of video monitoring provides new approaches and new direction.
Human body behavior is understood and identification is one of important research content in intelligent video monitoring system, belongs to the advanced stage that video image is processed, and its main task is from video sequence, to extract high level information to identify.The research of most concentrates in the identification of the conventional action of human body, as walked, running etc., and fewer for the research of abnormal behaviour identification.To conventional action recognition, most methods is carried out identification behavior by training or learning sample storehouse, and abnormal behaviour is difficult to define with Sample Storehouse conventionally, compare with general conventional behavior, fight, chase, push away human body abnormal behaviour under the specific occasions such as behaviour, jump, often there is the features such as sudden large, the duration is short, it is violent, unpredictable to move, no periodic irregularities.Therefore, the identification to abnormal behaviour, need to study new method.
Conventional method has template matching method and state-space method at present.Adopt the behavior recognition methods of template matching technique first image sequence to be converted to static in shape pattern, then compare with pre-stored behavior sample.Method based on state-space model defines each static posture as a state, between these states, by certain probability, connects.Any motion sequence can be seen an ergodic process between the different conditions of these static postures as, calculates joint probability during traveling through, and its maximal value is selected as the standard of classification behavior.The advantage of template matching technique be computation complexity low, realize simple, yet it is responsive for the variation at noise and run duration interval, is difficult in actual applications the effect that reaches desirable.State-space method can overcome the shortcoming of template matches, but algorithm is realized comparatively complexity.
Summary of the invention
The object of the invention is to overcome the above-mentioned deficiency of prior art, adopt a kind of specific human body abnormal behaviour recognition methods based on joint model.The joint model that foundation forms with upper limbs, lower limb and trunk, computes joint angles parameter, determines human body attitude degree of membership according to angle parameter, contrasts with rule database, determines whether as human body abnormal behaviour.
The invention provides a kind of specific human body abnormal behaviour recognition methods based on joint model, comprising:
The present invention adopts video camera as video acquisition device, adopts server as video process apparatus.
(1) server carries out motion detection identification to the video image of camera acquisition in real time, takes a decision as to whether human motion;
(2), according to the result of determination of previous step, obtain human motion bianry image; By prior background modeling, and adopt background subtraction method and the average weighted strategy of frame differential method are detected and extract movement human profile from video sequence, to overcome the imperfect problem of the extraction objective contour that movement human color and background caused when comparatively close; Then use the way that merges in region to add that morphological erosion and expansion principle process, to guarantee the integrality of profile;
(3), according to continuous multiple frames human body bianry image, judge whether barycenter has saltus step; If so, calculate centroid trajectory, differentiate and whether have abnormal behavior, finish; If not, turn next step; Centroid calculation formula is as follows,
Wherein
center-of-mass coordinate,
nobject pixel sum,
it is moving target pixel;
Centroid trajectory, obtains pedestrian's speed, position and direction by human body walking track;
(4) set up human synovial model, whole model is comprised of 6 line segments, represents respectively head, trunk, left upper extremity, right upper extremity, left lower extremity and right lower extremity;
(5) human synovial model is analyzed, extracted attitude angle parameter; By following the tracks of the two-dimentional Cartesian coordinates angle in all main joints, determine attitude angle parameter, and then the normal degree of membership of judgement human body attitude.A pedestrian's final posture determines by the associating attitude of upper limbs, lower limb and trunk, and therefore in the present invention, the joint model figure of employing is comprised of two upper limbs, two lower limb and trunk.And five angle parameters that human motion is exerted an influence, front arm have been defined
, rear arm
, trunk
, foreleg
, back leg
, as shown in Figure 1;
(6) according to rule database, human synovial model obtained in the previous step is analyzed, judge that whether human body behavior is abnormal; Owing to being identified the process that the behavior of object is non-linear, the change in time and space of more complicated, be often difficult to obtain accurate mathematical model.For this problem, the present invention adopts fuzzy discrimination technology.First utilize the first angle obfuscation to pedestrian contour limbs of knowledge of rule-based knowledge base, then differentiate, finally carry out ambiguity solution, obtain differentiating result;
6.1) obfuscation, is mapped to fuzzy quantity in corresponding domain accurately inputting numerical value.In pedestrian's motion process, each closes festival-gathering and produces different angles, and the scope of activities that the present invention defines upper limbs, trunk and lower limb is 0 ° ~ 180 °, and input value is [0,180].Each input value has a corresponding domain, inputs in the text domain and is (0,1), and obfuscation is the fuzzy subset of degree of membership size after ambiguity in definition on this domain.Set up angle and the relation function of inputting domain for this reason.Set up angle and follow following rule with the relation function of input domain: (a) meet the understanding of people to the corresponding relation of angle and degree of membership variation, deviation angle is less, and the degree of membership of its normal behaviour is larger; Deviation angle is larger, and its normal behaviour degree of membership is less; In certain amplitude scope, all think normally, when exceeding people's criterion, just think that namely its degree of membership that belongs to normal behaviour sharply reduces extremely.(b) value of degree of membership should followed normal distribution distribute or anti-normal distribution.Probability event in daily, as fuzzy membership, namely, in a large amount of learning samples, using there is the degree of membership of the frequency of this event as it, is defined as:
Be wherein joint angles parameter, span is.
6.2) fuzzy discrimination, adopts trunk and four limbs behavior blur level weighted mean as people's global behavior fuzzy value.
The degree of membership that is wherein,, for health different parts is to human body attitude weighing factor.
6.3) ambiguity solution, contrary with obfuscation, ambiguity solution is to the conversion process of accurately measuring by fuzzy quantity.Output domain is converted into the variation range of output physical quantity, the point on the output domain that ambiguity solution is tried to achieve when operation is converted into the value of the physical quantity of output.Output controlled quentity controlled variable
uquantification domain be [0,1].By fuzzy value, be converted into an accurate value, it is a point on output domain.Output quantity
ubasic domain be { behavior normal, abnormal behavior }.
accompanying drawing explanation
Fig. 1 human synovial angle parameter figure
Fig. 2 the inventive method process flow diagram
Fig. 3 background subtraction process flow diagram
Fig. 4 frame differential method process flow diagram
embodiment
Fig. 2 is the process flow diagram of the inventive method, with reference to Fig. 2, the invention provides a kind of embodiment that identifies human body abnormal behaviour.
Camera acquisition video sequence image, as the input of server, server carries out motion detection identification to the video image of input in real time, takes a decision as to whether human motion;
1. according to the result of determination of server, obtain human motion bianry image; By prior background modeling, employing detects and extracts movement human profile by background subtraction and the average weighted strategy of frame differential method from video sequence, to overcome, movement human color and background caused when comparatively close, extracts the imperfect of objective contour.
The ultimate principle of background subtraction is to build a relative background according to current scene, each frame and background in video sequence made comparisons, and by the diversity judgement between current video frame and background, whether be prospect, method flow is as shown in Figure 3.Frame differential method is exactly to utilize the difference between the several consecutive frame images in front and back in image sequence to extract the moving region in image, and its rudimentary algorithm process flow diagram as shown in Figure 4.
Use the way that merges in region to add that morphological erosion and expansion principle process, to guarantee the integrality of profile thereafter;
2. set up coordinate, adopt the method for circulation searching, maximal value and the minimum value vertical and horizontal direction of calculating every frame contour images mid point are designated as
thereby, find the minimum and maximum point of ordinate and horizontal ordinate
with
, the line of these two points of take is diagonal line, determines the position of single people in every two field picture;
3. according to continuous multiple frames human body bianry image, judge whether barycenter has saltus step; If so, calculate centroid trajectory, differentiate and whether have abnormal behavior, finish; If not, turn next step; Centroid calculation formula is as follows,
Wherein
center-of-mass coordinate,
nobject pixel sum,
it is moving target pixel;
4. set up human synovial model, whole model is comprised of 6 line segments, represents respectively head, trunk, left upper extremity, right upper extremity, left lower extremity and right lower extremity;
5. human synovial model is analyzed, extracted attitude angle parameter; By following the tracks of the angle of the two-dimentional Cartesian coordinates in all main joints, determine attitude angle parameter, and then the normal degree of membership of judgement human body attitude.A pedestrian's final posture determines by the associating attitude of upper limbs, lower limb and trunk, and the joint model figure therefore adopting is in the present invention comprised of two upper limbs, two lower limb and trunk.And definition five angle parameters that human motion is exerted an influence, front arm
, rear arm
, trunk
, foreleg
, back leg
;
6. according to rule database, human synovial model obtained in the previous step is classified, judge that whether human body behavior is abnormal; First utilize the first angle obfuscation to pedestrian contour limbs of knowledge of rule-based knowledge base, then differentiate, finally carry out ambiguity solution, obtain differentiating result; Fuzzy membership computing formula
，
Be wherein joint angles parameter, span is
.
Carry out fuzzy discrimination, adopt trunk and four limbs behavior blur level weighted mean as people's global behavior fuzzy value.
Wherein
for
degree of membership,
,
for health different parts is to human body attitude weighing factor.In the embodiment of the present invention, trunk and four limbs are defined as the weighing factor of action: { left upper extremity, left lower extremity, trunk, left lower extremity, right lower extremity }={ 0.15,0.15,0.3,0.2,0.2}.In embodiments of the present invention, will creep, discontinuous jump, falls down, squats down, and fights etc. and to differentiate for human body abnormal behaviour;
If the fuzzy membership of trunk is less than threshold value
, abnormal behavior;
If its comprehensive fuzzy membership is less than threshold value
, abnormal behavior.
Wherein
with
value is respectively the fuzzy membership of trunk and global behavior standard variance.
If pedestrian's behavior is judged into extremely by system, safety alarm system is by automatic alarm.
The foregoing is only embodiments of the invention, be not intended to limit protection scope of the present invention.Within the spirit and principles in the present invention all, to do any modifications, are equal to replacement and improvement etc., within all should being included in protection scope of the present invention.
1, the specific human body abnormal behaviour recognition methods based on joint model, the hardware platform of the method based on video camera and server formation, its feature comprises the following steps:
(1) server carries out motion detection identification to the video image of camera acquisition in real time, takes a decision as to whether human motion;
(2), according to the result of determination of previous step, obtain human motion bianry image;
(3) according to human body bianry image, judge whether barycenter has saltus step, if calculate centroid trajectory, differentiate and whether have abnormal behavior, finish; Turn if not next step;
(4) set up human synovial model, whole model is comprised of 6 line segments, represents respectively head, trunk, left upper extremity, right upper extremity, left lower extremity and right lower extremity;
(5) human synovial model is analyzed, extracted attitude angle parameter; By following the tracks of the angle of the two-dimentional Cartesian coordinates in all main joints, determine attitude angle parameter, and then the normal degree of membership of judgement human body attitude.A pedestrian's final posture determines by the associating attitude of upper limbs, lower limb and trunk, so in the present invention, joint model is comprised of two upper limbs, two lower limb and trunk.And definition five angle parameters that human motion is exerted an influence, front arm, rear arm, trunk, foreleg
, back leg;
(6) according to rule database and database, human synovial model obtained in the previous step is analyzed, determine whether abnormal; Owing to being identified the process that the behavior of object is non-linear, the change in time and space of more complicated, be often difficult to obtain accurate mathematical model.For this problem, the present invention adopts fuzzy discrimination technology.First utilize the knowledge of rule-based knowledge base first to obtaining the angle obfuscation of pedestrian contour limbs, then differentiate, finally carry out ambiguity solution, obtain differentiating result.
6.1) obfuscation, is mapped to fuzzy quantity in corresponding domain accurately inputting numerical value.In pedestrian's motion process, each closes festival-gathering and produces different angles, and the scope of activities that the present invention defines arm, trunk and leg is 0 ° ~ 180 °, and input value is [0,180].Each input value has a corresponding domain, inputs in the text domain and is (0,1), and obfuscation is the fuzzy subset of degree of membership size after ambiguity in definition on this domain.Set up angle and the relation function of inputting domain for this reason;
6.2) fuzzy discrimination, adopts trunk and four limbs behavior blur level weighted mean as people's global behavior fuzzy value;
6.3) ambiguity solution, contrary with obfuscation, ambiguity solution is to the conversion process of accurately measuring by fuzzy quantity.Output domain is converted into the variation range of output physical quantity, the point on the output domain that ambiguity solution is tried to achieve when operation is converted into the value of the physical quantity of output.Output controlled quentity controlled variable
uquantification domain be [0,1].By fuzzy value, be converted into an accurate value, it is a point on output domain.Output quantity
ubasic domain be { behavior normal, abnormal behavior }.
2, a kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
The method of extracting bianry image is the weighted mean of background subtraction and frame differential method.
3, a kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Fuzzification process in described claim 1 step (6), set up angle and follow following rule with the relation function of input domain: (a) meet the understanding of people to the corresponding relation of angle and degree of membership variation, be that deviation angle is less, the degree of membership of its normal behaviour is larger; Deviation angle is larger, and its normal behaviour degree of membership is less; In certain amplitude scope, all think normally, when exceeding people's criterion, just think that namely its degree of membership that belongs to normal behaviour sharply reduces extremely.(b) value of degree of membership should followed normal distribution distribute or anti-normal distribution.
4, a kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Described membership function, the probability event in daily, as fuzzy membership, namely, in a large amount of learning samples, using there is the degree of membership of the frequency of this event as it, is defined as:
5, a kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Fuzzy discrimination formula in described claim 1 step (6), is defined as:
6, a kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Rule base in described claim 1 step (6), rule base is deposited fuzzy discrimination rule.Fuzzy discrimination rule is the experience of the long-term accumulation of operating personnel and domain expert's relevant knowledge, and it is a knowledge base model of being differentiated differentiating object.
Claims (18)
1. the specific human body abnormal behaviour recognition methods based on joint model, the hardware platform of the method based on video camera and server formation, its feature comprises the following steps:
Server carries out motion detection identification to the video image of camera acquisition in real time, takes a decision as to whether human motion;
According to the result of determination of previous step, obtain human motion bianry image;
According to human body bianry image, judge whether barycenter has saltus step, if calculate centroid trajectory, differentiate and whether have abnormal behavior, finish; Turn if not next step;
Set up human synovial model, whole model is comprised of 6 line segments, represents respectively head, trunk, left upper extremity, right upper extremity, left lower extremity and right lower extremity;
Human synovial model is analyzed, extracted attitude angle parameter; By following the tracks of the angle of the two-dimentional Cartesian coordinates in all main joints, determine attitude angle parameter, and then the normal degree of membership of judgement human body attitude.
2. a pedestrian's final posture is determined by the associating attitude of upper limbs, lower limb and trunk, so in the present invention, joint model is comprised of two upper limbs, two lower limb and trunk.
3. five angle parameters that also definition exerts an influence to human motion, front arm
, rear arm
, trunk
, foreleg
, back leg
;
According to rule database and database, human synovial model obtained in the previous step is analyzed, determine whether abnormal; Owing to being identified the process that the behavior of object is non-linear, the change in time and space of more complicated, be often difficult to obtain accurate mathematical model.
4. for this problem, the present invention adopts fuzzy discrimination technology.
5. first utilize the knowledge of rule-based knowledge base first to obtaining the angle obfuscation of pedestrian contour limbs, then differentiate, finally carry out ambiguity solution, obtain differentiating result.
6.1) obfuscation, is mapped to fuzzy quantity in corresponding domain accurately inputting numerical value.
7. in pedestrian's motion process, each closes festival-gathering and produces different angles, and the scope of activities that the present invention defines arm, trunk and leg is 0 ° ~ 180 °, and input value is [0,180].
8. each input value has a corresponding domain, inputs in the text domain and is (0,1), and obfuscation is the fuzzy subset of degree of membership size after ambiguity in definition on this domain.
9. set up angle and the relation function of inputting domain for this reason; 6.2) fuzzy discrimination, adopts trunk and four limbs behavior blur level weighted mean as people's global behavior fuzzy value; 6.3) ambiguity solution, contrary with obfuscation, ambiguity solution is to the conversion process of accurately measuring by fuzzy quantity.
10. output domain is converted into the variation range of output physical quantity, the point on the output domain that ambiguity solution is tried to achieve when operation is converted into the value of the physical quantity of output.
11. output controlled quentity controlled variables
uquantification domain be [0,1].
12. are converted into an accurate value by fuzzy value, and it is a point on output domain.
13. output quantities
ubasic domain be { behavior normal, abnormal behavior }.
14. a kind of human body abnormal behaviour recognition methodss based on joint model according to claim 1, is characterized in that:
The method of extracting bianry image is the weighted mean of background subtraction and frame differential method.
15. a kind of human body abnormal behaviour recognition methodss based on joint model according to claim 1, is characterized in that:
Fuzzification process in described claim 1 step (6), set up angle and follow following rule with the relation function of input domain: (a) meet the understanding of people to the corresponding relation of angle and degree of membership variation, be that deviation angle is less, the degree of membership of its normal behaviour is larger; Deviation angle is larger, and its normal behaviour degree of membership is less; In certain amplitude scope, all think normally, when exceeding people's criterion, just think that namely its degree of membership that belongs to normal behaviour sharply reduces extremely.
The value of 16. (b) degree of membership should followed normal distribution distribute or anti-normal distribution.
17. a kind of human body abnormal behaviour recognition methodss based on joint model according to claim 1, is characterized in that:
Described membership function, the probability event in daily, as fuzzy membership, namely, in a large amount of learning samples, using there is the degree of membership of the frequency of this event as it, is defined as:
A kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Fuzzy discrimination formula in described claim 1 step (6), is defined as:
A kind of human body abnormal behaviour recognition methods based on joint model according to claim 1, is characterized in that:
Rule base in described claim 1 step (6), rule base is deposited fuzzy discrimination rule.
18. fuzzy discrimination rules are the experience of the long-term accumulation of operating personnel and domain expert's relevant knowledge, and it is a knowledge base model of being differentiated differentiating object.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310101688.8A CN104077591A (en) | 2013-03-27 | 2013-03-27 | Intelligent and automatic computer monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310101688.8A CN104077591A (en) | 2013-03-27 | 2013-03-27 | Intelligent and automatic computer monitoring system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104077591A true CN104077591A (en) | 2014-10-01 |
Family
ID=51598837
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310101688.8A Pending CN104077591A (en) | 2013-03-27 | 2013-03-27 | Intelligent and automatic computer monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104077591A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108134922A (en) * | 2017-12-25 | 2018-06-08 | 李锐 | A kind of behavior label and retroactive method based on artificial intelligence |
| CN108898108A (en) * | 2018-06-29 | 2018-11-27 | 炬大科技有限公司 | A kind of user's abnormal behaviour monitoring system and method based on sweeping robot |
| CN109991979A (en) * | 2019-03-29 | 2019-07-09 | 华中科技大学 | A kind of lower limb robot gait planing method towards complex environment |
| CN111183460A (en) * | 2017-11-17 | 2020-05-19 | 英国奥科斯国际有限公司 | Fall detector and improvement of fall detection |
| CN111460855A (en) * | 2019-01-20 | 2020-07-28 | 青岛海尔智能技术研发有限公司 | Method and device for monitoring behavior of article terminal user and computer storage medium |
| CN113536950A (en) * | 2021-06-22 | 2021-10-22 | 河南农道智能科技有限公司 | Personnel electric shock detection method and system based on artificial intelligence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739557A (en) * | 2010-01-26 | 2010-06-16 | 重庆大学 | Motion cycle analysis-based method and device for identifying abnormal human behavior |
| US20110175625A1 (en) * | 2008-07-17 | 2011-07-21 | Nederlandse Organisatie Voor Toegepast- Natuurwetenschappelijk Onderzoek Tno | Electromagnetic body scanning system, a method and a computer program product |
| CN102214309A (en) * | 2011-06-15 | 2011-10-12 | 北京工业大学 | Special human body recognition method based on head and shoulder model |
| CN102917207A (en) * | 2012-10-24 | 2013-02-06 | 沈阳航空航天大学 | Motion sequence based abnormal motion vision monitoring system |
-
2013
- 2013-03-27 CN CN201310101688.8A patent/CN104077591A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110175625A1 (en) * | 2008-07-17 | 2011-07-21 | Nederlandse Organisatie Voor Toegepast- Natuurwetenschappelijk Onderzoek Tno | Electromagnetic body scanning system, a method and a computer program product |
| CN101739557A (en) * | 2010-01-26 | 2010-06-16 | 重庆大学 | Motion cycle analysis-based method and device for identifying abnormal human behavior |
| CN102214309A (en) * | 2011-06-15 | 2011-10-12 | 北京工业大学 | Special human body recognition method based on head and shoulder model |
| CN102917207A (en) * | 2012-10-24 | 2013-02-06 | 沈阳航空航天大学 | Motion sequence based abnormal motion vision monitoring system |
Non-Patent Citations (1)
| Title |
|---|
| 张军: "基于视频的运动人体异常行为分析识别研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111183460A (en) * | 2017-11-17 | 2020-05-19 | 英国奥科斯国际有限公司 | Fall detector and improvement of fall detection |
| CN108134922A (en) * | 2017-12-25 | 2018-06-08 | 李锐 | A kind of behavior label and retroactive method based on artificial intelligence |
| CN108134922B (en) * | 2017-12-25 | 2020-09-25 | 李锐 | Behavior marking and tracing method based on artificial intelligence |
| CN108898108A (en) * | 2018-06-29 | 2018-11-27 | 炬大科技有限公司 | A kind of user's abnormal behaviour monitoring system and method based on sweeping robot |
| CN108898108B (en) * | 2018-06-29 | 2022-04-26 | 炬大科技有限公司 | User abnormal behavior monitoring system and method based on sweeping robot |
| CN111460855A (en) * | 2019-01-20 | 2020-07-28 | 青岛海尔智能技术研发有限公司 | Method and device for monitoring behavior of article terminal user and computer storage medium |
| CN111460855B (en) * | 2019-01-20 | 2023-11-24 | 青岛海尔智能技术研发有限公司 | Method and device for monitoring user behavior of article terminal and computer storage medium |
| CN109991979A (en) * | 2019-03-29 | 2019-07-09 | 华中科技大学 | A kind of lower limb robot gait planing method towards complex environment |
| CN109991979B (en) * | 2019-03-29 | 2020-10-30 | 华中科技大学 | Lower limb robot anthropomorphic gait planning method oriented to complex environment |
| CN113536950A (en) * | 2021-06-22 | 2021-10-22 | 河南农道智能科技有限公司 | Personnel electric shock detection method and system based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103279737B (en) | A kind of behavioral value method of fighting based on space-time interest points | |
| CN109492581A (en) | A kind of human motion recognition method based on TP-STG frame | |
| CN104134077B (en) | A kind of gait recognition method unrelated with visual angle based on the determination theories of learning | |
| CN101867699B (en) | Real-time tracking method of nonspecific target based on partitioning | |
| CN104077591A (en) | Intelligent and automatic computer monitoring system | |
| CN105404847B (en) | A kind of residue real-time detection method | |
| CN106204586B (en) | A kind of moving target detecting method under complex scene based on tracking | |
| CN103310444B (en) | A kind of method of the monitoring people counting based on overhead camera head | |
| JP2007128513A (en) | Scene analysis | |
| CN107247946B (en) | Behavior recognition method and device | |
| CN101271527A (en) | Exception action detecting method based on athletic ground partial statistics characteristic analysis | |
| CN113011367A (en) | Abnormal behavior analysis method based on target track | |
| CN105513053B (en) | One kind is used for background modeling method in video analysis | |
| CN109145696B (en) | Old people falling detection method and system based on deep learning | |
| CN106210634A (en) | A kind of wisdom gold eyeball identification personnel fall down to the ground alarm method and device | |
| CN110415268A (en) | A kind of moving region foreground image algorithm combined based on background differential technique and frame difference method | |
| CN108509938A (en) | A kind of fall detection method based on video monitoring | |
| CN109460704A (en) | A kind of fatigue detection method based on deep learning, system and computer equipment | |
| CN103150572A (en) | On-line type visual tracking method | |
| CN106909879A (en) | A kind of method for detecting fatigue driving and system | |
| CN102142085A (en) | Robust tracking method for moving flame target in forest region monitoring video | |
| CN109118516A (en) | A kind of target is from moving to static tracking and device | |
| CN103049748B (en) | Behavior monitoring method and device | |
| Subaweh et al. | Implementation of pixel based adaptive segmenter method for tracking and counting vehicles in visual surveillance | |
| CN114782988A (en) | Construction environment-oriented multi-stage safety early warning method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20141001 |
|
| WD01 | Invention patent application deemed withdrawn after publication |