CN107679471B - Indoor personnel air post detection method based on video monitoring platform - Google Patents

Indoor personnel air post detection method based on video monitoring platform Download PDF

Info

Publication number
CN107679471B
CN107679471B CN201710871443.1A CN201710871443A CN107679471B CN 107679471 B CN107679471 B CN 107679471B CN 201710871443 A CN201710871443 A CN 201710871443A CN 107679471 B CN107679471 B CN 107679471B
Authority
CN
China
Prior art keywords
state
duty
frames
person
video
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201710871443.1A
Other languages
Chinese (zh)
Other versions
CN107679471A (en
Inventor
王霞
张为
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN201710871443.1A priority Critical patent/CN107679471B/en
Publication of CN107679471A publication Critical patent/CN107679471A/en
Application granted granted Critical
Publication of CN107679471B publication Critical patent/CN107679471B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06KGRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
    • G06K9/00Methods or arrangements for recognising patterns
    • G06K9/62Methods or arrangements for pattern recognition using electronic means
    • G06K9/6267Classification techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/11Region-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/194Segmentation; Edge detection involving foreground-background segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10016Video; Image sequence
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20212Image combination
    • G06T2207/20224Image subtraction
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30196Human being; Person
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30232Surveillance
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30242Counting objects in image

Abstract

The invention relates to an indoor post detection method based on a video monitoring platform, which comprises the following steps: processing the preprocessed video frames by using an interframe difference method to obtain difference images of every two frames; defining a state machine, and realizing dynamic conversion of four states, wherein the state machine specifically comprises four states: no person is on duty, suspected to be off duty and someone is on duty; judging the state of the operator on duty according to the mean value of the difference image, and entering a state machine if the mean value of the difference image is not more than 200; otherwise, if the average value of the difference image is more than 200, the original state is kept unchanged, and the state machine is not entered, so that the interference caused by sudden change of the brightness of the image in the monitoring video is properly eliminated; and the state variables and the idle timer are correspondingly changed through the personnel on-duty state information obtained through the conversion of the state machine.

Description

Indoor personnel air post detection method based on video monitoring platform
Technical Field
The invention belongs to the field of intelligent video monitoring, and particularly relates to an indoor personnel air post detection system and method based on an existing video monitoring platform.
Background
In the current society developing at a high speed, unstable factors in various industries are increasing day by day, and safety problems are paid more and more attention by people, particularly in the field of security protection. In order to ensure the safety of the important safety departments and the important unit facilities of the country and maintain the long and permanent security of the society, the duty on the post becomes a necessary choice. If the person on duty is left off duty, it is likely to cause an immeasurable loss. In order to prevent the situation, many enterprises and departments adopt a supervision mode to carry out idle work detection so as to take countermeasures in time and eliminate potential safety hazards.
With the continuous development of video monitoring technology, the method of monitoring the on-duty information by means of video monitoring gradually replaces the manual on-duty checking mode. The traditional video monitoring system needs a specially-assigned person to check the monitoring video of the duty room in real time, has long monitoring time and needs to be highly concentrated in spirit, and forms higher labor cost. Moreover, when people watch the monitoring display for a certain time, the situation of inattention occurs, so that key video clips are easily missed, and the purpose of supervision cannot be achieved. The intelligent video monitoring technology comes up. Compared with the traditional monitoring means, the intelligent video monitoring has higher reliability, quicker response and lower cost, the video images are collected through the camera, the information characteristics of the video frame images are intelligently analyzed, the judgment and the response are made according to the preset criterion, and for the idle work detection, whether the idle work behavior occurs or not is judged according to the preset criterion, and the alarm is given and corresponding measures are taken.
The air station detection based on the monitoring platform mainly comprises camera video acquisition, system background intelligent analysis and display end real-time display, wherein the intelligent analysis part relates to target detection, tracking, analysis and recognition and the like. At present, the methods for detecting targets in video images mainly include: the interframe difference method, the background difference method, and the optical flow method. The interframe difference method is simplest and most direct, and is mostly used for the conditions of simple background and small environmental interference; the background difference method is sensitive to the dynamic change of the scene; optical flow methods can detect objects as the camera moves, but are less suitable for real-time processing. In general, the idle guard detection based on the video monitoring platform obtains the state information of the person on duty through a video sequence obtained by a real-time intelligent analysis monitoring camera, and immediately makes a response when the idle guard condition occurs, so that potential safety hazards are eliminated.
The existing method for solving the problem of idle post detection is generally an idle post detection method based on a video monitoring platform. For example, in patent CN104021653A, the functions of video analysis and alarm at the level crossing are realized by running a video dynamic analysis camera, a video analysis alarm and a central video monitoring server for early warning, alarm and short message alarm, and it is mainly determined whether a person is on duty according to whether a video frame image changes dynamically; in patent CN102740059A, an image analysis and tracking algorithm is used to supervise an operator on duty, an identification algorithm is used to identify whether there is a person in an image, and when the accumulated unattended time reaches a preset time parameter, an unattended supervision signal is sent to a control center, where there is no specific process mentioned for a human body detection algorithm, and there are only two states of unattended and occupied. In general, if a human body detection algorithm with high accuracy and a set of more perfect state transformation processes can be designed, the indoor air post detection function based on the monitoring platform can be better realized.
Disclosure of Invention
The invention aims to provide a method for monitoring indoor idle post behaviors in real time and processing the indoor idle post behaviors in time based on the existing video monitoring platform, which has the following technical scheme:
an indoor post detection method based on a video monitoring platform comprises the following steps:
1) setting idle post detection parameters including a scale transformation parameter and an idle post time threshold;
2) inputting a video, reading the video frame by frame, scaling all video frames into a uniform size by multiplying the scale conversion parameters, and preprocessing the video frames for further processing in the subsequent steps;
3) processing the preprocessed video frames by using an interframe difference method to obtain difference images of every two frames, performing threshold value binarization processing and morphological processing, calculating pixel brightness mean values and standard deviations of the difference images, and then performing the process in a circulating manner to detect the dynamic foreground in real time;
4) defining a state machine, and realizing dynamic conversion of four states, wherein the state machine specifically comprises four states: no person is on duty, suspected to be off duty and someone is on duty;
5) judging the state of the operator on duty according to the difference image mean value obtained by the calculation, and entering a state machine if the difference image mean value is not more than 200; otherwise, if the average value of the difference image is more than 200, the original state is kept unchanged, and the state machine is not entered, so that the interference caused by sudden change of the brightness of the image in the monitoring video is properly eliminated;
6) realizing the initialization of state machine parameters, defining an idle state timer initialized to 0 and a state variable, and correspondingly changing the state variable and the idle state timer through the personnel on duty state information obtained by the transformation of the state machine;
7) running a state machine, and judging state change once every 10 frames to realize dynamic conversion of four states; when the video image is in any one of the states, firstly loading a pre-trained human body detection classifier to operate a human body detection process, then correspondingly obtaining personnel on-duty information statistics in every 10 frames of video images, judging the current state of personnel according to a corresponding judgment criterion and analyzing a state change trend; wherein the content of the first and second substances,
the human body upper half body sample is used as a training sample to train a human body detection classifier, and the human body detection process is as follows: detecting the current preprocessed video frame by using a pre-trained human body detection classifier, scanning the video frame image in a multi-scale mode to obtain a target detected as a human body, identifying the target by using a rectangular frame, storing the target in a rectangular frame queue, and drawing a human body boundary rectangular frame on the original video image.
The transformation process for the four states is as follows:
when no person is on duty, the off duty timer starts to time, 1 is accumulated each time, the human body detection process is operated, if a video frame detects a person, the state of 'suspected on duty' is entered, otherwise, the state is not changed;
when the person is suspected to be on duty, the human body detection process is operated, if the mean value of difference images of at least 6 frames in every 10 frames is more than 2, the person is considered to be in a motion state, otherwise, the person is in a static state; if the video frame is in the motion state, judging to be a dynamic frame, continuously counting the number of video frames with detected people, and if at least 6 frames in 10 frames have detected people, considering the state to be 'people on duty'; otherwise, considering that no one is on duty, adding 10 to the off duty timer; if the person is in a static state, namely the person is judged to be a static frame, the overlapping area between every two rectangular frames of the detected person in every two frames of 10 frames is calculated, if the ratio of the overlapping area to the area of a single rectangular frame is not less than 0.2, the person is considered to be the same person basically in a static sitting posture, if at least 6 frames in the 10 frames are the person in the static sitting posture, the person is considered to be on duty, and if the person is not on duty, the person is considered to be on duty;
when the people are on duty, the idle duty timer restarts counting from 0, the human body detection process is operated, if no people are detected, the state of suspected idle duty is entered, otherwise, the state is not changed;
when the 'suspected idle guard' state is reached, the initial value of the idle guard timer is still 0, the human body detection process is operated, the judgment is carried out once every 10 frames, if at least 8 frames in every 10 frames are the condition that the difference image mean value is not more than 2, the frame is a static frame, and at least 6 frames do not detect the human body, the state is considered to be 'no person is on guard', the idle guard timer is increased by 10, otherwise, the 'person is on guard';
by analogy, the state machine process is operated once every 10 frames to finish one judgment;
8) after the state machine conversion process is finished, checking the idle timer, if the accumulated time of the idle timer exceeds a preset idle time threshold value, for example, no one is on duty within 5min, judging that an idle behavior occurs, giving an alarm in real time, displaying alarm state information, resetting the idle timer to 0, and restarting to time; otherwise, judging that no idle post behavior occurs;
9) and at the end of each detection of the on-duty state of the personnel, emptying the queue for storing the human body boundary rectangular frame in the single-frame image for the next round of idle post detection.
The invention does not depend on the color information of the video frame image, the monitoring video at night can be processed, the monitoring video sequence is analyzed in real time by establishing a complete indoor personnel on duty detection system based on a video monitoring platform, adopting a relatively accurate human body detection algorithm and a relatively perfect state machine conversion process, the state information of the person on duty is obtained, and the person on duty immediately reacts and takes corresponding measures when the on duty condition occurs, thereby eliminating the potential safety hazard. By adopting the method, the supervision level of related departments is improved, related personnel can be assisted to complete work tasks better, the labor cost is reduced, the production and life safety is guaranteed, and the method has great significance for the field of intelligent security.
Drawings
FIG. 1 is a block diagram of a video monitoring system carried by the method of the present invention
FIG. 2 is a block diagram of a human body detection system designed by the method of the present invention
FIG. 3 is a transformation block diagram of a state machine designed by the method of the present invention
FIG. 4 is a flow chart of the method of the present invention
Detailed Description
The general processing architecture of a video monitoring system in the existing security field is as follows: the picture shot by the analog camera is directly transmitted to the monitor to be displayed through one part of the cable, and the other part of the picture is transmitted to the hard disk video recorder. Converting the analog signal entering the hard disk video recorder into a digital code stream, on one hand, coding the digital code stream, and storing the digital code stream in the hard disk video recorder in a file form; on the other hand, the digital video recorder can be connected with the digital video recorder through a network at any time, and code streams are extracted to be displayed and analyzed. As shown in fig. 1.
The indoor air post detection system software based on the video monitoring platform formed by the method provided by the invention can be connected with a hard disk video through a network, collect video data and analyze the video data in real time, or extract a transcoded video file stored in a hard disk video recorder. The method comprises the following steps: firstly, setting idle post parameters including a scale transformation parameter and an idle post time threshold parameter; reading the video frame by frame and preprocessing the video; processing the video by an interframe difference method to obtain a difference image; defining a state machine, wherein the state machine comprises four states: "no people are on duty", "suspected to be off duty", "someone is on duty"; entering different states by analyzing the mean value of the difference image; in each state, firstly, running a human body detection process on the preprocessed video frame, and detecting whether a human body object exists in the video frame by using a pre-trained human body detection classifier; the conversion of the four states is realized through the conversion of the state machine, the states of 'no person on duty', 'suspected off duty' and 'someone on duty' can be judged more accurately, and the values of the off duty timer are correspondingly counted; if the idle time timer exceeds the preset idle time threshold value, alarming, resetting the idle time timer to zero and then restarting timing; by analogy, judging once every 10 frames; and finally storing the processed result video.
By monitoring video images in real time, a human body detection algorithm with high accuracy and a relatively perfect state change judgment mechanism are adopted, the duty state of an analyst is detected and analyzed in real time, effective supervision on indoor idle post behaviors is realized, and potential safety hazards are eliminated. The invention is shown in detail in figure 4.
The following describes each part in detail:
1. air post parameter setting
Setting up scaling parameters, such as 0.75, for post-processing; the other parameter is a time threshold of the idle post, for example, 5min, so as to be used for the detection alarm judgment of the idle post in the later period.
2. Video frame pre-processing
Reading the input video frame by frame, multiplying each video frame by a scale conversion parameter, such as 0.75, reducing the video frame to 0.75 time of the original size, improving the calculation efficiency, and preprocessing the scaled video frame: since some surveillance videos are night videos, the video frames are converted from RGB signals into a gray-scale image format, so that the color information can be independent.
3. Dynamic foreground extraction
And processing the preprocessed video frames by using an interframe difference method to obtain a difference image of every two frames, performing threshold value binarization processing and morphological processing, calculating a pixel brightness mean value and a standard deviation of the difference image, and then performing the process in a circulating manner to detect the dynamic foreground in real time.
4. Definition state machine
Defining a state machine, and realizing dynamic conversion of four states, wherein the state machine specifically comprises four states: no person is on duty, suspected to be off duty and someone is on duty.
5. Enter state machine
Judging the state of the operator on duty according to the difference image mean value obtained by the calculation, and entering a state machine if the difference image mean value is not more than 200; on the contrary, if the average value of the difference image is more than 200, the original state is kept unchanged, and the state machine is not entered, so that the interference caused by sudden change of the brightness of the image in the monitoring video can be eliminated.
6. State machine parameter initialization
Defining an idle guard timer with an initial value of 0 and a state variable, and correspondingly adjusting the values of the state variable and the idle guard timer through the personnel on-duty state information obtained by the transformation of the state machine.
7. State machine transition
Running a state machine, detecting the duty state of personnel, judging state change every 10 frames, and realizing dynamic conversion of four states; when the video image is in any one of the states, firstly loading a pre-trained human body detection classifier to operate a human body detection process, then correspondingly obtaining personnel on-duty information statistics in every 10 frames of video images, judging the current state of personnel according to a corresponding judgment criterion and analyzing a state change trend; the following is a detailed description:
the human body detection process is shown in fig. 2, and comprises the following steps:
considering that a specific environment for people to be checked in a duty room, most people in a sitting posture sometimes walk around, but the integrity of the upper half of the human body can be basically ensured, and the upper half of the human body comprises a head and shoulder part and corresponding structural features with high human body recognition rate, based on such consideration, the human body detection classifier is trained by using the upper half of the human body sample as a training sample. The main human body detection steps are as follows: detecting a current preprocessed video frame by using a pre-trained human body detection classifier, scanning the video frame image in a multi-scale mode until the whole video frame image is traversed to obtain a target detected as a human body, identifying the target by using a rectangular frame, storing the target in a rectangular frame queue, and then drawing an enlarged boundary rectangular frame on an original video image (the size of an original image is adapted by scaling the parameter of the rectangular frame by correspondingly dividing the parameter of a predefined scale transformation parameter);
in addition, as shown in fig. 3, the flow diagram of the state machine is roughly as follows:
when no person is on duty, the off duty timer starts to time, 1 is accumulated each time, the human body detection process is operated, if a video frame detects a person, the state of 'suspected on duty' is entered, otherwise, the state is not changed;
when the person is suspected to be on duty, the human body detection process is operated, if the mean value of difference images of at least 6 frames in every 10 frames is more than 2, the person is considered to be in a motion state, otherwise, the person is in a static state; if the video frame is in the motion state, judging to be a dynamic frame, continuously counting the number of video frames with detected people, and if at least 6 frames in 10 frames have detected people, considering the state to be 'people on duty'; otherwise, the user is considered to be unattended, and the idle timer is added with 10; if the person is in a static state, namely the person is judged to be a static frame, the overlapping area between every two rectangular frames of the detected person in every two frames of 10 frames is calculated, if the ratio of the overlapping area to the area of a single rectangular frame is not less than 0.2, the person is considered to be the same person basically in a static sitting posture, if at least 6 frames in the 10 frames are the person in the static sitting posture, the person is considered to be on duty, and if the person is not on duty, the person is considered to be on duty;
when the people are on duty, the idle duty timer counts from 0 again, the human body detection process is operated, if the people are not detected suddenly, the suspected idle duty state is entered, otherwise, the state is not changed;
when the 'suspected idle guard' state is reached, the initial value of the idle guard timer is still 0, the human body detection process is operated, the judgment is carried out once every 10 frames, if at least 8 frames in every 10 frames are the condition that the difference image mean value is not more than 2, the frame is a static frame, and at least 6 frames do not detect the human body, the state is considered to be 'no person is on guard', the idle guard timer is increased by 10, otherwise, the 'person is on guard';
by analogy, the state machine process is operated once every 10 frames to finish one judgment;
8. idle post behavior determination
After the whole conversion process of the state machine is finished, checking the idle timer, if the accumulated time of the idle timer exceeds a preset idle time threshold value, for example, no one is on duty within 5min, judging that an idle behavior occurs, giving an alarm in real time, displaying alarm state information, resetting the idle timer to 0, and restarting to time; otherwise, judging that no idle post behavior occurs.
9. Cyclically performing idle post detection
And at the end of each detection of the on-duty state of the personnel, emptying the queue for storing the human body boundary rectangular frame in the single-frame image for the next round of idle post detection.
10. Saving result video
After the whole system engineering is finished, all the video processing processes can be stored in a new video mode for workers to check in the future.

Claims (1)

1. An indoor post detection method based on a video monitoring platform comprises the following steps:
1) setting idle post detection parameters including a scale transformation parameter and an idle post time threshold;
2) inputting a video, reading the video frame by frame, scaling all video frames into a uniform size by multiplying the scale conversion parameters, and preprocessing the video frames for further processing in the subsequent steps;
3) processing the preprocessed video frames by using an interframe difference method to obtain difference images of every two frames, performing threshold value binarization processing and morphological processing, calculating pixel brightness mean values and standard deviations of the difference images, and then performing the process in a circulating manner to detect the dynamic foreground in real time;
4) defining a state machine, and realizing dynamic conversion of four states, wherein the state machine specifically comprises four states: no person is on duty, suspected to be off duty and someone is on duty;
5) judging the state of the operator on duty according to the difference image mean value obtained by the calculation, and entering a state machine if the difference image mean value is not more than 200; otherwise, if the average value of the difference image is more than 200, the original state is kept unchanged, and the state machine is not entered, so that the interference caused by sudden change of the brightness of the image in the monitoring video is properly eliminated;
6) realizing the initialization of state machine parameters, defining an idle state timer initialized to 0 and a state variable, and correspondingly changing the state variable and the idle state timer through the personnel on duty state information obtained by the transformation of the state machine;
7) running a state machine, and judging state change once every 10 frames to realize dynamic conversion of four states; when the video image is in any one of the states, firstly loading a pre-trained human body detection classifier to operate a human body detection process, then correspondingly obtaining personnel on-duty information statistics in every 10 frames of video images, judging the current state of personnel according to a corresponding judgment criterion and analyzing a state change trend; wherein the content of the first and second substances,
the human body upper half body sample is used as a training sample to train a human body detection classifier, and the human body detection process is as follows: detecting a current preprocessed video frame by using a pre-trained human body detection classifier, scanning the video frame image in a multi-scale mode to obtain a target detected as a human body, identifying the target by using a rectangular frame, storing the target in a rectangular frame queue, and drawing a human body boundary rectangular frame on an original video image;
the transformation process for the four states is as follows:
when no person is on duty, the off duty timer starts to time, 1 is accumulated each time, the human body detection process is operated, if a video frame detects a person, the state of 'suspected on duty' is entered, otherwise, the state is not changed;
when the person is suspected to be on duty, the human body detection process is operated, if the mean value of difference images of at least 6 frames in every 10 frames is more than 2, the person is considered to be in a motion state, otherwise, the person is in a static state; if the video frame is in the motion state, judging to be a dynamic frame, continuously counting the number of video frames with detected people, and if at least 6 frames in 10 frames have detected people, considering the state to be 'people on duty'; otherwise, considering that no one is on duty, adding 10 to the off duty timer; if the person is in a static state, namely the person is judged to be a static frame, the overlapping area between every two rectangular frames of the detected person in every two frames of 10 frames is calculated, if the ratio of the overlapping area to the area of a single rectangular frame is not less than 0.2, the person is considered to be the same person in the static sitting posture, if at least 6 frames in the 10 frames are the person in the static sitting posture, the person is considered to be on duty, and if the person is not detected to be on duty, the person is considered to be on duty;
when the people are on duty, the idle duty timer restarts counting from 0, the human body detection process is operated, if no people are detected, the state of suspected idle duty is entered, otherwise, the state is not changed;
when the 'suspected idle guard' state is reached, the initial value of the idle guard timer is still 0, the human body detection process is operated, the judgment is carried out once every 10 frames, if at least 8 frames in every 10 frames are the condition that the difference image mean value is not more than 2, the frame is a static frame, and at least 6 frames do not detect the human body, the state is considered to be 'no person is on guard', the idle guard timer is increased by 10, otherwise, the 'person is on guard';
by analogy, the state machine process is operated once every 10 frames to finish one judgment;
8) after the state machine conversion process is finished, checking the idle timer, if the accumulated time of the idle timer exceeds a preset idle time threshold value, judging that idle behavior occurs, giving an alarm in real time, displaying alarm state information, resetting the idle timer to 0, and restarting timing; otherwise, judging that no idle post behavior occurs;
9) and at the end of each detection of the on-duty state of the personnel, emptying the queue for storing the human body boundary rectangular frame in the single-frame image for the next round of idle post detection.
CN201710871443.1A 2017-09-24 2017-09-24 Indoor personnel air post detection method based on video monitoring platform Expired - Fee Related CN107679471B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710871443.1A CN107679471B (en) 2017-09-24 2017-09-24 Indoor personnel air post detection method based on video monitoring platform

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710871443.1A CN107679471B (en) 2017-09-24 2017-09-24 Indoor personnel air post detection method based on video monitoring platform

Publications (2)

Publication Number Publication Date
CN107679471A CN107679471A (en) 2018-02-09
CN107679471B true CN107679471B (en) 2020-03-06

Family

ID=61138000

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710871443.1A Expired - Fee Related CN107679471B (en) 2017-09-24 2017-09-24 Indoor personnel air post detection method based on video monitoring platform

Country Status (1)

Country Link
CN (1) CN107679471B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108564050A (en) * 2018-04-23 2018-09-21 西安安邦鼎立智能科技有限公司 A kind of two visitors one based on the video enterprise supervision personnel that endanger inspect the sentries method and system
CN109190710B (en) * 2018-09-13 2022-04-08 东北大学 off-Shift detection method based on Haar-NMF characteristics and cascade Adaboost classifier
CN109492620A (en) * 2018-12-18 2019-03-19 广东中安金狮科创有限公司 Monitoring device and its control device, post monitoring method and readable storage medium storing program for executing
CN109867186B (en) * 2019-03-18 2020-11-10 浙江新再灵科技股份有限公司 Elevator trapping detection method and system based on intelligent video analysis technology
CN110443179B (en) * 2019-07-29 2021-11-12 思百达物联网科技(北京)有限公司 Off-post detection method and device and storage medium
CN110708823A (en) * 2019-10-29 2020-01-17 佛山科学技术学院 Classroom automatic power-off control method and system based on image recognition
CN111970495B (en) * 2020-08-21 2021-12-21 湖南工学院 Remote automatic light control system
CN113065386A (en) * 2020-12-25 2021-07-02 泰州可以信息科技有限公司 Preset duration worker on-duty state detection system

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101883257A (en) * 2009-05-08 2010-11-10 上海弘视通信技术有限公司 Guard detection system and detection method thereof
CN102647580A (en) * 2012-04-27 2012-08-22 浙江晨鹰科技有限公司 Video monitoring method and system
CN104408406A (en) * 2014-11-03 2015-03-11 安徽中科大国祯信息科技有限责任公司 Staff off-post detection method based on frame difference method and background subtraction method
CN106570467A (en) * 2016-10-25 2017-04-19 南京南瑞集团公司 Convolutional neutral network-based worker absence-from-post detection method
CN107103300A (en) * 2017-04-22 2017-08-29 高新兴科技集团股份有限公司 One kind is left the post detection method and system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7424175B2 (en) * 2001-03-23 2008-09-09 Objectvideo, Inc. Video segmentation using statistical pixel modeling

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101883257A (en) * 2009-05-08 2010-11-10 上海弘视通信技术有限公司 Guard detection system and detection method thereof
CN102647580A (en) * 2012-04-27 2012-08-22 浙江晨鹰科技有限公司 Video monitoring method and system
CN104408406A (en) * 2014-11-03 2015-03-11 安徽中科大国祯信息科技有限责任公司 Staff off-post detection method based on frame difference method and background subtraction method
CN106570467A (en) * 2016-10-25 2017-04-19 南京南瑞集团公司 Convolutional neutral network-based worker absence-from-post detection method
CN107103300A (en) * 2017-04-22 2017-08-29 高新兴科技集团股份有限公司 One kind is left the post detection method and system

Also Published As

Publication number Publication date
CN107679471A (en) 2018-02-09

Similar Documents

Publication Publication Date Title
CN107679471B (en) Indoor personnel air post detection method based on video monitoring platform
KR101942808B1 (en) Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN
US9420236B2 (en) Monitoring system and monitoring method
CN102521578B (en) Method for detecting and identifying intrusion
US20060170769A1 (en) Human and object recognition in digital video
CN104966304B (en) Multi-target detection tracking based on Kalman filtering and nonparametric background model
CN109117827B (en) Video-based method for automatically identifying wearing state of work clothes and work cap and alarm system
CN102750709B (en) Video is utilized to detect the method and apparatus of behavior of fighting
Zin et al. Unattended object intelligent analyzer for consumer video surveillance
CN104680145B (en) The on off state change detecting method and device of a kind of
CN105844659B (en) The tracking and device of moving component
CN109409289A (en) A kind of electric operating safety supervision robot security job identifying method and system
CN110096945B (en) Indoor monitoring video key frame real-time extraction method based on machine learning
CN111767823A (en) Sleeping post detection method, device, system and storage medium
CN112163572A (en) Method and device for identifying object
KR100887942B1 (en) System for sensing abnormal phenomenon on realtime and method for controlling the same
CN101930540A (en) Video-based multi-feature fusion flame detecting device and method
CN107948465A (en) A kind of method and apparatus for detecting camera and being disturbed
KR102194499B1 (en) Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN and Driving Method Thereof
CN111079694A (en) Counter assistant job function monitoring device and method
CN107909599A (en) A kind of object detecting and tracking system
CN104392201A (en) Human fall identification method based on omnidirectional visual sense
CN113158752A (en) Intelligent safety management and control system for electric power staff approach operation
CN112633157A (en) AGV working area safety real-time detection method and system
CN112287816A (en) Dangerous working area accident automatic detection and alarm method based on deep learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20200306

Termination date: 20200924