WO2014139415A1 - 一种智能门窗防入侵装置以及系统、智能门禁系统 - Google Patents

一种智能门窗防入侵装置以及系统、智能门禁系统 Download PDF

Info

Publication number
WO2014139415A1
WO2014139415A1 PCT/CN2014/073258 CN2014073258W WO2014139415A1 WO 2014139415 A1 WO2014139415 A1 WO 2014139415A1 CN 2014073258 W CN2014073258 W CN 2014073258W WO 2014139415 A1 WO2014139415 A1 WO 2014139415A1
Authority
WO
WIPO (PCT)
Prior art keywords
module
information
intrusion
video
alarm
Prior art date
Application number
PCT/CN2014/073258
Other languages
English (en)
French (fr)
Inventor
黄鹏宇
周建雄
何跃凯
彭元华
郭振中
Original Assignee
成都百威讯科技有限责任公司
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 成都百威讯科技有限责任公司 filed Critical 成都百威讯科技有限责任公司
Publication of WO2014139415A1 publication Critical patent/WO2014139415A1/zh

Links

Classifications

    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B13/00—Burglar, theft or intruder alarms
    • G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
    • G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
    • G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
    • G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
    • G08B13/19695—Arrangements wherein non-video detectors start video recording or forwarding but do not generate an alarm themselves
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B13/00—Burglar, theft or intruder alarms
    • G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
    • G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
    • G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
    • G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
    • G08B13/19617—Surveillance camera constructional details
    • G08B13/19619—Details of casing
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N7/00—Television systems
    • H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Definitions

  • the present invention relates to an anti-illegal intrusion system, and in particular to an intelligent door and window anti-intrusion device, an anti-intrusion system, and an intelligent access control system.
  • an anti-theft device such as a camera is generally installed in a residential area or an outdoor corridor, such as an intelligent video-based monitoring and alarm system for indoor personnel, mainly through the image of the camera.
  • Intelligent analysis to achieve the analysis of moving targets in indoor places, to achieve timely detection and alarm of illegal intrusion of personnel, while ignoring non-hazardous sports, reducing false alarms, there is also a home intelligent anti-theft system, which is at least by security door, machinery Locks, windows, open air terraces and balcony doors.
  • the handle position sensor in the assembly sets the balcony path detection area from the open balcony into the balcony door passage, and enters the window path detection area in the passage of the room from the window, thereby realizing automatic identification of the home and the thief in the home. system.
  • a single anti-intrusion detection technology has its own defects. In some cases, it cannot work normally. Once it is not working properly, the entire detection system is in a paralyzed state, and the stability of the system is poor.
  • Intelligent video analysis and detection may be interfered by unrelated motions such as ambient light changes and specular reflections, resulting in a large number of false alarms;
  • pyroelectric infrared detection is susceptible to interference from temperature, strong light, and complex environmental motion; vibration detection is susceptible to wind The external force causes the interference caused by the vibration of the door and window;
  • the glass breakage detection is susceptible to interference from the external environment noise;
  • the door magnetic detection requires the door and window to be strictly closed, and will not be used if the room needs ventilation. For example, in hot summer, the user often needs to open the window to ventilate. .
  • the traditional anti-intrusion system can't predict the thief's stepping behavior before committing the crime, and the system's early warning ability is poor;
  • An object of the present invention is to provide a smart door and window anti-intrusion system, which solves the problems of the prior anti-theft measures, a single false alarm rate, a high false alarm rate, an untimely alarm, and easy operation.
  • the present invention adopts the following technical solutions:
  • a smart door and window anti-intrusion device comprises a lighting module, a video collecting module, a comprehensive information collecting module, an alarm sending module, a signal processing module and a lighting control module, wherein:
  • the illumination module provides an illumination source for video capture of the video capture module;
  • the video collection module starts collecting surrounding video information when the lighting module is turned on
  • the integrated information collecting module collects the surrounding information in real time and provides the information to the signal processing module in real time; the lighting control module determines the opening of the lighting module, wherein the lighting control module compares the surrounding information according to the signal processing module The result of the processing to determine the opening;
  • the signal processing module distinguishes the intrusion behavior according to the video information and the surrounding information, thereby determining whether to activate the alarm sending module and issue an alarm signal.
  • a power management module is further included, and the anti-intrusion device is powered, and the power management module is connected to an external power source and a battery;
  • the battery When the external power source is turned off, the battery is automatically enabled to supply power to the anti-intrusion device.
  • the integrated information collection module includes one or more modules of an audio collection module, a heat release infrared detection alarm module, a vibration detection module, a glass break detection module, and a door magnetic detection module;
  • the surrounding information includes voice information, infrared information, vibration information, glass broken sound information, and door magnetic information.
  • the anti-intrusion device as described above is characterized in that:
  • the glass breakage detection module includes a glass breakage intrusion detector, and the detector effectively detects high frequency glass broken sound, and has a strong inhibitory effect on low frequency ordinary sound signals.
  • the illumination lamp in the illumination module is in a ring shape, and the heat release infrared detection alarm module, the vibration detection module, the video collection module, the audio collection module and the alarm transmission module are distributedly installed on an outer edge of the illumination lamp, and the signal A processing module, a power management module, and a switch control module are mounted inside the lighting.
  • the glass breakage detection module and the door magnetization detection module respectively convert the detected information into a voltage signal to output to the signal processing module by wired or wireless communication.
  • the alarm sending module sends out an alarm signal sent by the signal processing module by using wired or wireless communication.
  • the anti-intrusion device as described above is characterized in that:
  • the wireless communication method includes 3G, WIFK Zigbee. Bluetooth or other wireless communication methods.
  • the signal processing module collects signals transmitted by the heat release infrared detection alarm module, the vibration detection module, the glass break detection module, the door magnetic detection module, the video acquisition module, and the audio collection module, and performs comprehensive analysis to complete the resolution. Intrusion behavior, and an alarm signal is sent through the alarm sending module.
  • the signal processing module includes a main processing chip that employs a dual core architecture mode in which the main core processor and the slave core processor are combined.
  • the main core processor receives the video information and the surrounding information collected by the video collection module and the integrated information detection module;
  • the video information and the surrounding information constitute a multi-source signal transmitted to the slave core processor
  • the slave processor resolves the intrusion behavior to form a final decision signal.
  • the distinguishing intrusion behavior mainly includes running an audio and video intelligent analysis algorithm, and simultaneously distinguishing with other information in the surrounding information.
  • the present invention also provides an intelligent anti-intrusion system, the system comprising a storage module and an anti-intrusion device according to any of the above; the storage module establishes a communication connection with the anti-intrusion device by means of wired or wireless communication, Storing information collected by the anti-intrusion device;
  • the distinguishing the intrusion behavior includes identifying a person close to the anti-intrusion device, and the identity recognition is completed by face recognition or other manners; While performing the identification, the surrounding information collected by the integrated information collection module when the person approaches the anti-intrusion device is further analyzed;
  • the signal processing module analyzes the surrounding information based on the identity and forms a decision signal to distinguish whether the intrusion behavior is established.
  • the other manner of identification includes identifying by a mobile phone that is carried around, the intrusion prevention device communicating with the mobile phone to determine the identity of the person.
  • the face registration is further included to form a face database
  • the face recognition is performed by using the HAAR feature combined with the adaboost algorithm for face detection.
  • the two-layer human eye locator is used to obtain the human eye positioning by the adaboost algorithm, and the face positioning result is used to correct the binoculars by image rotation.
  • Feature extraction is performed by two-dimensional Gabor filter, and the covariance distance between vectors is used as the matching metric.
  • the nearest neighbor classification method is used to match the face image to be recognized with the face image in the face database.
  • the distinguishing the intrusion behavior further includes detecting and analyzing the action in the video information.
  • the detection and analysis of the action includes background modeling and motion detection, target confirmation, target tracking, and behavior analysis.
  • the background modeling and motion detection are performed by multi-frame image averaging to obtain a background mean, and the background variance. Initialize to a constant, by formula
  • the background pixel is separated from the foreground pixel by foreground detection, where 1 represents the foreground pixel, 0 represents the background pixel, and f(x, y) represents the gray value of the pixel at (x, y) in the image, ⁇ ( x, y) represents the corresponding background mean, ⁇ (x, y) represents the variance of the corresponding background;
  • the target confirms that the foreground area that does not satisfy the size requirement is filtered according to the set size by the target size filter;
  • the target filter filters out the highlighted foreground region according to the set threshold;
  • the foreground region is gradient-differentiated with the corresponding background, and the motion continuity, the motion saliency, and the area change continuity are respectively filtered by the time domain filter;
  • the target tracking estimating the position of the target in the next frame according to the current frame and the position of the previous target, by using a formula
  • the target position prediction is achieved for the estimated target speed, where and v respectively represent the speed of the target in the X direction and the y direction at time t, and N is the time window, ⁇ . And ⁇ respectively represent the abscissa and ordinate of the center of the target outer rectangle at t-n time, and ⁇ and ⁇ respectively represent the abscissa and ordinate of the center of the target outer rectangle at time t-n-1;
  • the behavior analysis determines the specific behavior by the user setting the alarm area and the corresponding rules.
  • the analyzing the surrounding information comprises detecting and analyzing the voice information collected by the audio collecting module, and extracting the voice information helpful for distinguishing the intrusion as a foreground signal from the background audio information which is a background signal.
  • the detection and analysis of the voice information includes:
  • the subband energy ratio Ai(t) is defined as the subband energy and The ratio of the total energy, where U(i) and L(i) represent the upper and lower boundaries of subband i, where the frequency band is divided into M subbands by logarithm;
  • mixed Gaussian background modeling assumes that the probability distribution of signal changes can be fitted with K Gaussian distributions, and the probability distribution of the audio signal is expressed as
  • Each Gaussian distribution 1 ⁇ , 4 1 , (7 1) in the model is given a weight of 0 ⁇ , where cj i represents the mean and standard deviation of the Gaussian distribution, respectively, and multiple Gaussian distributions obtain the probability distribution of the signal by linear combination.
  • the first B Gaussian distributions are taken as b
  • the current signal amplitude is matched with the previous Gaussian distribution. If any of the Gaussian distributions match the success, the signal is the background signal, otherwise it is the foreground signal.
  • the system performs the identification, analyzes the surrounding information, forms a decision signal, and determines whether the intrusion behavior is established includes the following steps:
  • the digital video signal is collected by the video collection module
  • the digital audio signal is collected by the audio collection module
  • the external heat temperature change information is collected by the heat release infrared detection alarm module and converted into a voltage signal, which is collected by the vibration detection module.
  • the external vibration generates information of deformation or force transformation and is converted into a voltage signal
  • the glass breaking detection and the door magnetic detection module collect information of the glass breaking and the door magnetic change and convert into a voltage signal
  • the signal processing module receives the voltage signals of the heat release infrared detection alarm module, the vibration detection module, the glass break detection module, and the door magnetic detection module, and combines audio and video intelligent analysis, completes identification, and identifies the identity. To comprehensively determine the occurrence of an intrusion, and when there is an intrusion, an alarm signal is issued;
  • the alarm sending module receives the alarm signal and sends an alarm message to the outside through a wired or wireless manner.
  • the system as described above is characterized by:
  • the alarm sending module receives the alarm signal and sends an alarm message to the outside, including:
  • An alarm transmitting module receives the digital video signal and the digital audio signal and transmits the digital video signal and the digital audio signal together with an alarm signal.
  • the present invention also provides an intelligent access control system, comprising the intelligent anti-intrusion system according to any one of the above, wherein the smart access control system uses the intelligent anti-intrusion system to complete the identification,
  • the surrounding information is analyzed to identify whether the guest is visiting and automatically sends the information to the visitor to the mobile or fixed terminal.
  • Whether the identification is a guest visit includes the following steps:
  • the digital video signal is collected by the video collection module
  • the digital audio signal is collected by the audio collection module
  • the external heat temperature change information is collected by the heat release infrared detection alarm module and converted into a voltage signal, which is collected by the vibration detection module.
  • the external vibration generates information of deformation or force transformation and is converted into a voltage signal
  • the glass breaking detection and the door magnetic detection module collect information of the glass breaking and the door magnetic change and convert into a voltage signal
  • the signal processing module receives the voltage signals of the heat release infrared detection alarm module, the vibration detection module, the glass break detection module and the door magnetic detection module, and combines the audio and video intelligent analysis to determine whether the guest is visiting;
  • the third step if there is a guest visit behavior, wirelessly send the video or photo information of the visitor outside the door to a mobile or fixed terminal such as the home owner's mobile phone, and activate the terminal to give a ring, vibration or Voice prompts.
  • a mobile or fixed terminal such as the home owner's mobile phone
  • the invention has the following beneficial effects: the intelligent door and window anti-intrusion system of the invention adopts pyroelectric infrared detection, vibration detection, audio detection, glass break detection and door magnetic detection technology, and discovers the door and window area through multi-dimensional detection.
  • the possible illegal intrusion behavior can accurately distinguish whether the illegal intrusion, the guest visit or the normal movement of the head of the household, and the real-time and visual alarm signal of the audio, video or picture as the carrier is sent by wired or wireless communication means.
  • FIG. 1 is a schematic diagram showing the connection of an embodiment of a smart door and window anti-intrusion system according to the present invention.
  • FIG. 2 is a schematic diagram of hardware connection of the smart door and window anti-intrusion system of the present invention.
  • FIG. 3 is a software structural diagram of a smart door and window anti-intrusion system according to the present invention.
  • FIG. 4 is a schematic flow chart of a main processor end in a signal processing module of the smart door and window anti-intrusion system of the present invention.
  • FIG. 5 is a schematic flow chart of a slave processor in a signal processing module of the smart door and window anti-intrusion system of the present invention.
  • FIG. 6 is a schematic diagram of a face registration and face recognition process of the smart door and window anti-intrusion system of the present invention.
  • FIG. 7 is a schematic diagram of a video-based moving target detection and tracking process of the smart door and window anti-intrusion system of the present invention.
  • FIG. 8 is a schematic diagram of an audio detection process of the smart door and window anti-intrusion system of the present invention.
  • FIG. 9 is a schematic diagram of the face image analysis of the face-up in the face normalization of the smart door and window anti-intrusion system of the present invention.
  • FIG. 1 shows an embodiment of a smart door and window anti-intrusion system of the present invention: a smart door and window anti-intrusion system, including
  • the video acquisition module adopts a video sensor to collect a video stream signal, completes digitization of the image signal, preprocesses the image signal, obtains a digital video signal that satisfies the requirements of the signal processing module, and outputs the specific detection mode to the signal processing module, which may be a regional mode or curtain mode
  • the audio acquisition module completes the digital-to-analog conversion of the speech signal, the sampling and encoding of the speech signal, and the filtering process to obtain a digital audio signal required by the signal processing module and output to the signal processing module;
  • the heat release infrared detection alarm module senses the difference between the temperature of the moving object and the background object.
  • the pyroelectric infrared can sense the difference between the human body temperature and the background temperature, and converts it into a voltage signal and outputs it to the signal processing module;
  • the vibration detecting module converts the deformation or force information generated by the external vibration into a voltage signal, and outputs the signal to the signal processing module;
  • the glass break detection module and the door magnetic detection module convert the corresponding information into a voltage signal to output to the signal processing module by wired or wireless communication;
  • the signal processing module collects the signals transmitted by the heat release infrared detection alarm module, the vibration detection module, the video acquisition module and the audio collection module, performs comprehensive analysis, distinguishes the illegal intrusion or the guest visit, and sends an alarm signal to the alarm sending module or The home owner sends a photo of the visitor and other information;
  • the alarm sending module adopts wired or wireless communication mode, and is mainly used for receiving an alarm signal sent by the signal processing module and sending an alarm message to the household mobile phone or the cell management center, and can adopt wired, 3G, WIFL Zigbee, Bluetooth or other wireless communication methods.
  • 3G, WIFL Zigbee, Bluetooth or other wireless communication methods One;
  • the power management module is connected to the external power supply and the battery.
  • the external power supply When the external power supply is normally powered, the whole product uses an external power supply.
  • the internal backup battery When the external power supply is cut off, the internal backup battery is automatically enabled to ensure that the above modules continue to work normally after the external power supply is interrupted. It is good to ensure that the camera continues to work for 2-24 hours after the external power supply is interrupted.
  • the invention is disposed on the ring-shaped luminaire, and the heat-release infrared detection alarm module, the vibration detection module, the video acquisition module, the audio collection module, and the alarm transmission module are installed on an outer edge of the ring-shaped luminaire, Letter
  • the number processing module, the power management module, and the switch light control module of the simple lamp itself are installed in the middle of the simple lamp.
  • the switch light control module can combine the vibration, sound, infrared and other signals to determine whether someone is entering. If someone enters, turn on the light illumination to provide ambient light for subsequent video capture.
  • the system and the light collection not only can hide the system well, prevent the criminals from destroying the intelligent door and window anti-intrusion system, but also the video capture in the system.
  • the camera of the module fills the light to make the collected video information clearer.
  • the video acquisition module is provided with an auxiliary light source to ensure that it can still collect valid image data in a low illumination environment.
  • the signal processing module uses a dual core architecture mode in which the main core processor and the slave core processor are combined.
  • the main core processor mainly completes the collection of digital video signals and digital audio signals, audio and video coding, collecting the signals output by the heat release infrared detection alarm module and the signals output by the vibration detection module, and receiving the glass break detection module and the door magnetic detection module.
  • the signal transmits the multi-source signal together with the audio and video data to the slave core processor; the intrusion behavior judgment and the account owner identity verification from the core processor, wherein the intrusion behavior judgment mainly includes running the audio and video intelligent analysis algorithm, and determining whether the video and the sound are used There is an intrusion behavior, and at the same time, combined with the signal output by the heat release infrared detection alarm module, the signal output by the vibration detection module, the glass break detection module and the signal sent by the door magnetic detection module, the final decision signal is formed.
  • the software is mainly composed of signal acquisition, signal analysis, alarm transmission and linkage control.
  • the signal acquisition and alarm transmission are run on the main core processing end, and the signal analysis is run on the slave core processing end.
  • the signal acquisition interface includes an audio and video acquisition module, a pyroelectric infrared alarm signal acquisition module, a vibration alarm signal acquisition module, a glass break detection module, a door magnetic detection module, and a wireless alarm signal acquisition module.
  • the main core processing end completes the acquisition of the switching alarm signal and the delay synchronization, and transmits the audio and video signals to the slave core processing end.
  • the initialization of the main chip main processing terminal is completed first, and then the peripheral acquisition device is completed, the communication device is initialized, and the main core processing program is started. Create audio and video capture threads, pyroelectric infrared detection threads, wireless receive threads, communication threads, and alarm send threads.
  • Audio and video collection thread When receiving audio and video data, combine one frame of multi-source data, including video data, audio data, various sensor alarm strengths and the presence or absence of the user's mobile phone, and put them into FrameBuffer for communication thread to call and complete at the same time. Pre-recorded audio and video to provide alarm information.
  • Alarm signal acquisition thread Receives a variety of wired or wireless access alarm sensor switch alarm signal Ai, where Ai can be pyroelectric infrared alarm signal, vibration alarm signal, glass break alarm signal, door magnetic alarm signal.
  • Communication thread Receives an alarm signal sent from the core processor and notifies the alarm sending thread. If there is one frame of multi-source data in the FrameBuffer, the data is sent to the DSP side.
  • Alarm sending thread After receiving the alarm signal, organize relevant alarm information, such as pre-recorded audio and video, pictures, etc., to send to the external terminal.
  • the multitasking kernel and the application are loaded into the slave processor memory, and a series of initializations are completed from the core processor, and the communication thread and the face recognition thread are automatically created.
  • the communication thread first determines whether it needs to send an alarm/warning signal. If yes, it sends an alarm/warning signal to the main processing terminal. Otherwise, it determines whether one frame of multi-source information is received.
  • the multi-source information includes: audio and video data, various sensor alarms.
  • the signal and the identifier of the user's mobile phone are placed in the frame buffer FrameBuffer, and the receiving acknowledgement signal is sent to the main processing end, indicating that the processing is working normally.
  • the face recognition thread extracts the video data from the FrameBuffer to complete the face recognition.
  • the intrusion detection thread reads a frame of data from the FrameBuffer, through audio analysis, vibration Dynamic information, infrared information to determine whether someone is about to enter the surveillance area, if it is, then turn on the illumination, provide ambient lighting for the next video analysis, and facilitate the door open. After the illumination is turned on, the motion detection is started. Since only the behavior of entering the door is concerned, it is necessary to judge the motion direction, and when it is judged to be the door entrance direction, the target is tracked. When the target existence time is greater than the set time, the description is suspicious. If the WIFI mode identity fails, the system will actively cooperate with the voice recognition to perform face recognition. If the identity verification fails, the intrusion behavior alarm is triggered, and the alarm is sent to the ARM. Early warning signal.
  • the identity determination of the head of the household includes face recognition or installation of WIFI communication software on the user's mobile phone, and the video collection module and the user's mobile phone perform identity verification through WIFI communication.
  • the method for face recognition mainly includes the following parts:
  • Face registration collecting frontal photos of different angles of the same person, finding the position of the face in the image through face positioning, and normalizing face size, face angle and illumination, extracting features of standardized face images, entering Register a user database;
  • Face recognition using HAAR feature combined with adaboost algorithm to achieve face detection for face localization, using a two-layer human eye locator, all obtained eye positioning by adaboost algorithm, using face positioning results to correct both eyes by image rotation
  • the face normalization is realized horizontally, and the feature extraction is performed by the two-dimensional Gabor filter.
  • the covariance distance between the vectors is used as the matching metric, and the face image of the query is matched with the database face image by the nearest neighbor classification method.
  • the Adaboost algorithm combines a large number of weak classifiers with general classification ability into a strong classifier according to the decline of the training error index.
  • the HAAR feature provides a large number of weak classification features for the adaboost algorithm, which ensures that the adaboost algorithm generally finds weak classifications with excellent performance.
  • the use of integral histogram and cascade classifier greatly reduces the processing time while ensuring higher detection accuracy; human eye positioning is generally divided into two layers, wherein the first layer is coarse positioning. The positioning area selects most of the eye area including the eyebrows, the second layer is the precise positioning, and the positioning area only contains the eye area.
  • the coarse positioner Compared with the precision positioner, the coarse positioner has more regional information, so the stability of the positioning is higher, there is basically no large positional deviation, and the precise positioner can achieve precise positioning of the human eye, but is susceptible to The interference of the eyebrows and the corners of the eyes caused a positioning error.
  • the approximate position range of the human eye is determined by the geometric proportional relationship, and the precise positioning of the human eye is realized in the range.
  • Face standardization is a very important step in face recognition, and the standardization result directly affects the quality. The accuracy of face recognition. Face standardization mainly completes the geometric correction and brightness correction of the face image.
  • the eyes are corrected to the horizontal by the image rotation, and the face image is intercepted by the distance d between the eyes. As shown in Figure 9, the image is finally scaled to 80x80 pixels.
  • the plane parameters a, b, c can be obtained by linear regression.
  • the Gabor wavelet feature can be selected, and the Gabor transform has excellent performance in analyzing the texture of the local area of the image.
  • Two-dimensional Gabor filter ⁇ (can be expressed as: Where is the image coordinate, k is the center frequency of the filter, and K y represents the projection of k on the horizontal and vertical axes, respectively.
  • Shadow, % is the direction of the filter, "and v represents different values
  • the two-dimensional Gabor filter modulates the sine wave plane of a specific frequency and direction by a two-dimensional Gaussian function, and realizes the analysis of image textures of different scales and directions by changing the frequency and direction of the sine wave plane.
  • a face image of 80x80 size is obtained by face normalization.
  • five filter scales and eight filter directions are selected to obtain 40 Gabor filters of different directions and frequencies, and a face image is convoluted by a filter.
  • the final Gabor feature dimension is 163840.
  • Such a high-dimensional feature vector greatly reduces the speed of identifying the classification, so it is necessary to reduce the dimension of the feature vector.
  • 4x4 uniform downsampling is used to achieve feature dimensionality reduction.
  • the nearest neighbor classification method is used to match the face image of the query with the database face image, and the covariance distance between the vectors is used as the matching metric. At the same time, the reliability of the final matching can be measured by the covariance distance.
  • the signal processing module is connected with a storage module (Storage), and different types of storage media are used for different application situations, and SD and TF cards are used by SDIO.
  • Storage storage module
  • Control this type of storage medium is easy to replace; Nand is controlled by Nandflash, this type of integration is high, but it is not easy to replace the storage medium; SSD is controlled by PCI-E or SATA interface, this type of storage space can be very large;
  • An external power supply (DC IN) and a battery (Battery) are connected through a power management module; a memory (DDR) for controlling the operation of the CPU is connected;
  • Flash flash memory
  • wireless module 3G, WIFI
  • wireless module zigbee, Blue tooth or other wireless module
  • the linkage alarm module is connected with an external alarm light, a siren linkage device, a local sound and light alarm, or other related equipment.
  • FIG. 7 shows another preferred embodiment of a smart door and window anti-intrusion system according to the present invention.
  • the detection method of the video acquisition module is mainly divided into four parts: background modeling and motion detection, target confirmation, target tracking and behavior analysis.
  • the background modeling and motion detection are performed by multi-frame image averaging to obtain a background mean, and the background variance is initialized to a constant, by formula
  • the foreground detection separates the background pixels from the foreground pixels, where 1 represents the foreground pixel and 0 represents the background pixel.
  • the system thinks that the scene may have changed. At this time, the subsequent analysis is turned off. If a large area of foreground pixels is detected for a long time, the system thinks that the scene has indeed changed, and then the background is restarted. Mode.
  • the background update of the system uses a background update strategy based on a combination of target level and pixel level.
  • a background update strategy based on a combination of target level and pixel level.
  • ⁇ indicates f time, "is the update factor, generally a small value, such as 0.01.
  • the minimum value of the variance is ⁇ , which prevents the model from over-converging, resulting in weakening of the model's anti-noise ability, where ⁇ The value is 36.
  • the background update based on the pixel level is adopted, that is, if the pixel point is detected as a foreground for a long time in a fixed time.
  • the pixel point updates the pixel point to the background at one time; if it is determined by the target segmentation that the pixel point is the target pixel point, the background update method based on the target layer is adopted, that is, only when the target is not active for a long time , the pixel of the area where the target is located will be updated to the background at one time.
  • the target confirms that the foreground area that does not satisfy the size requirement is filtered according to the set size by the target size filter, and the highlighted foreground area is filtered according to the set threshold by the highlight target filter, and the time domain filter pair is adopted.
  • Motion continuity, motion saliency, and area change continuity are separately filtered.
  • the highlight target filter filters out the bright foreground areas, which may be interferences such as light sources or reflective media.
  • Judging method for highlighting the foreground area If the proportion of the highlighted pixel in the foreground area exceeds the set threshold, the foreground area is considered to be the highlighted foreground area;
  • Time domain filter Time domain filtering is based on the filtering of the target layer, Use target tracking technology.
  • Motion continuity filtering The occupied area of the two frames before and after the target is overlapped over a large area, that is, the position of the target does not change.
  • Motion Significant Filtering Assume that the motion of the target has a certain directionality, that is, the ratio of the sum of the vector of the target displacement and the scalar sum of the target displacement over a period of time should be greater than a certain threshold.
  • Area change continuity filtering Assume that the area of the target is relatively stable and no mutation occurs.
  • the tracking of the target adopts target tracking based on motion detection, and based on the motion detection, the target trajectory is formed by the target association between the frames, and provides a basis for subsequent target behavior analysis, according to the current frame and the previous target.
  • the position of the target is estimated at the position of the next frame, by the formula N - ⁇ and
  • N is the time window
  • - represents the abscissa of the center of the target outer rectangular frame at time tn, respectively.
  • the same reason - "- 1 respectively represent the abscissa and ordinate of the center of the target outer rectangular frame at time tn-1.
  • two features of the relatively stable and reliable target area S and the center of the target circumscribed rectangle are selected.
  • the target association matching can find the position of the previous frame target in the current frame and achieve the target positioning.
  • the search range of the target can be limited to a small distance range, and at the same time also greatly reduced The risk of mismatching.
  • ⁇ ( Q ⁇ C ⁇ ' Q ; — lie ⁇ ) ⁇ r formula the current frame target Q and the target 0 to be matched; - 1 is matched when the following formula is satisfied, where the search distance is wide, and needs to be based on the actual The scene is determined.
  • ⁇ is the weighting factor of the feature, here is taken as 0.5, where the upper limit of the error is set ⁇ to prevent mismatch, where 0.4 is taken.
  • the update of the target features is updated in a first-order smoothing manner, as shown in the following formula:
  • the behavior analysis determines the specific behavior by the user setting the alarm area and the corresponding rules.
  • the alarm area can be set as the area where the door is located. If the person entering the area exceeds the set value, it is considered to be a possible intrusion behavior and an alarm is triggered. It is also possible to set the larger area where the door is located as the detection area. If the person entering the area exceeds the set value, it is considered to be a possible stepping behavior, triggering an early warning.
  • FIG. 8 shows another preferred embodiment of a smart door and window anti-intrusion system of the present invention, a method for detecting an audio collection module
  • the sound-based intrusion detection process is as shown in the following figure.
  • the audio signal obtained by the audio collection device is first pre-processed, mainly including sub-frame emphasis, framing and windowing, to meet the needs of subsequent spectrum analysis and recognition, and then adopt hybrid
  • the Gaussian model establishes the background sound model. After the modeling is completed, the sound detection is started. When an abnormal sound different from the background sound is detected in the scene, the alarm intensity starts to accumulate. If the alarm intensity is greater than the set threshold, the intrusion behavior is triggered.
  • Alarm as follows, to obtain the power spectral density of the audio signal, set the sampling rate of the audio signal ⁇ ), (for example, take 48 kHz),
  • FFT fast Fourier transform
  • Power spectral density using formula Defined as the ratio of sub-band energy to total energy, where U (i and L W represent the upper and lower boundaries of the sub-band.
  • Hybrid Gaussian background modeling and sound detection based on sub-band energy ratio, mixed Gaussian background modeling assumes that the probability distribution of signal changes can be fitted with K Gaussian distributions, and the probability distribution of audio signals is expressed as ⁇
  • Prob(x) ⁇ ⁇ ⁇ ⁇ ( ⁇ , ⁇ ⁇ , ⁇ ; )
  • each Gaussian distribution HA' 0 ⁇ in the model is given a weight
  • ⁇ and respectively represent the mean and standard deviation of the Gaussian distribution.
  • the Gaussian distribution obtains the probability distribution of the signal through linear combination, which reflects the importance of the model to the ith Gaussian distribution.
  • ⁇ reflects the convergence of the Gaussian distribution.
  • the amplitude should be smaller around the mean, and ° is also smaller.
  • Such a Gaussian distribution is suitable for describing the change of the background.
  • the distribution is arranged in descending order of * ⁇ °", and the Gaussian distribution of the top ranks most representative of the distribution of the background, the former Gaussian distribution is used as the background model, and the remaining Gaussian distribution is used as the foreground model, where b
  • the method for preventing intrusion of the intelligent door and window anti-intrusion system is as follows:
  • the digital video signal is collected by the video acquisition module
  • the digital audio signal is collected by the audio collection module
  • the external heat temperature change information is collected by the heat release infrared detection alarm module and converted into a voltage signal
  • the external vibration is generated by the vibration detection module to generate deformation.
  • the information of the force transformation is converted into a voltage signal
  • the information of the glass breakage and the door magnet change information is closed by the glass breakage detection module and the door magnetic detection module and converted into a voltage signal;
  • the signal processing module receives the voltage signals of the heat release infrared detection alarm module, the vibration detection module, the glass break detection module and the door magnetic detection module, and combines the audio and video intelligent analysis, the owner identification verification, and comprehensively determines the intrusion behavior. Occurs, when there is an intrusion, an alarm signal is issued;
  • the alarm sending module receives the alarm signal
  • the receiving video collecting module collects the digital video signal and collects the digital audio signal through the audio collecting module, and transmits the data to the external management terminal or the handheld terminal through wired or wireless means.
  • the smart door and window intrusion prevention system is used as a method of visually locking as follows:
  • the digital video signal is collected by the video acquisition module
  • the digital audio signal is collected by the audio collection module
  • the external heat temperature change information is collected by the heat release infrared detection alarm module and converted into a voltage signal
  • the external vibration is generated by the vibration detection module to generate deformation.
  • the information of the force transformation is converted into a voltage signal
  • the information of the glass breakage and the door magnet change information is closed by the glass breakage detection module and the door magnetic detection module and converted into a voltage signal;
  • the signal processing module receives the voltage signals of the heat release infrared detection alarm module, the vibration detection module, the glass break detection module and the door magnetic detection module, and combines the audio and video intelligent analysis to determine whether the guest is visiting, if there is a guest to The occurrence of the interview behavior is to wirelessly send the video or photo information of the visitor outside the door to the mobile or fixed terminal such as the home owner's mobile phone, and activate the terminal to give a ringing, shaking or voice prompt.
  • the intelligent video analysis technology utilized by the video capture module is a technique for extracting useful information in a video from a video by using a computer or an embedded dual-core processor.
  • the technology utilizes the powerful data processing capabilities of various processors to perform high-speed analysis of massive data in video images, filtering out information that users do not care about, and providing useful key information only to users.
  • the pyroelectric infrared detection technology used by the infrared detection alarm module is a kind of passive infrared detection, which senses the difference between the temperature of the moving object and the background object.
  • the pyroelectric infrared sensor In the monitoring area, when there is no human body movement, the pyroelectric infrared sensor only senses the background temperature.
  • the infrared rays emitted by the human body are enhanced by the Fresnel lens and then collected on the pyroelectric infrared sensor.
  • the pyroelectric element in the discharge infrared senses the difference between the human body temperature and the background temperature, the temperature changes, the charge balance is lost, the charge is released outward, and an alarm signal is generated after the subsequent circuit detection and processing.
  • Vibration anti-theft alarm technology used by the vibration detection module The piezoelectric ceramic piece is used as a sensor to transform or react the external vibration into a voltage signal, and the alarm signal is sent and stopped by the subsequent auxiliary circuit.
  • the glass breakage detection technology used by the glass breakage detection module Using the piezoelectric effect of the piezoelectric ceramic sheet, a glass breakage intrusion detector can be fabricated.
  • the high-frequency glass breaking sound (1 Ok ⁇ 15kHZ) is effectively detected, and the sound signals below lOkHZ (such as speaking and walking sound) are strongly inhibited.
  • the two techniques of voice control and vibration detection are usually combined, and the alarm signal is output only when the high frequency sound signal emitted by the glass breakage and the vibration caused by the knocking glass are detected at the same time.
  • the door magnetic detection technology utilized by the magnetic door detection module consists of two parts: a permanent magnet and a normally open reed switch.
  • the permanent magnet is used to generate a constant magnetic field. When the permanent magnet and the reed switch are close together, the door The magnetic sensor is in the working state. When the permanent magnet leaves the reed switch for a certain distance, it is in the normally open state and triggers an alarm.
  • the workflow of this system is:
  • the user can turn the alarm function on or off as needed.
  • the detection method can select one or several of video detection, audio detection, pyroelectric infrared detection, vibration detection, glass breakage detection, and door magnetic detection, such as the use of heat in the window area. Discharge infrared detection, vibration detection, glass breakage detection, door magnetic detection, video detection, audio detection, pyroelectric infrared detection, vibration detection, door magnetic detection.
  • the alarm information can be one or more of video, audio or picture.
  • the signal processing module adds the operating system and application program from the FLASH, completes the initialization of the main processing chip and the configuration of the peripheral hardware, and then completes the initialization of each subsystem, and finally enters the normal state. Working status.
  • the main processor end of the main processing chip continuously collects the audio and video signals and the pyroelectric infrared detection signals of the monitoring site, and simultaneously receives the alarm signals of other distributed alarm sensors through the wireless receiving module, and sends the multi-source data into the processing.
  • the analyzer performs analysis and simultaneously performs audio and video pre-recording. If an alarm or warning signal is received from the processor side, the pre-recorded audio and video, together with the captured photo, is sent to the cell monitoring center or the mobile phone of the household owner through the alarm signal sending module.
  • the main processing chip runs the intelligent audio and video analysis algorithm and the intrusion behavior judgment from the processor end. Through audio analysis, vibration information, and infrared information to determine whether someone is about to enter the surveillance area, if it is, turn on the illumination, provide ambient illumination for the next video analysis, and facilitate the door opening. After the illumination is turned on, the motion detection is started. Since only the behavior of entering the door is concerned, it is necessary to judge the motion direction, and when it is judged to be the door entrance direction, the target is tracked. When the target existence time is greater than the set time, the description is suspicious. If the WIFI mode identity fails, the system will actively cooperate with the voice recognition to perform face recognition. If the identity verification fails, the intrusion behavior alarm is triggered and sent to the main processor. Alarm/warning signal.
  • the monitoring center and the head of the household can confirm the alarm information sent back and take countermeasures.
  • Installation of the system Suspended installation, ensuring that the monitoring range can cover the entire area of the door and some of the aisle area in front of the door, adjust the shooting angle to ensure that the front view can be obtained in the monitoring field of view, the lighting conditions are good, and the clear face can be obtained.
  • Image, face pixel area is greater than 80 * 80 pixels.
  • Suspicious personnel in the surveillance area The detection of suspicious persons is achieved by a combination of various alarm sensors.
  • Video face recognition mode After entering the area, the user actively cooperates with the camera to complete the face recognition, and compares with the registered face database to complete the identity verification.
  • WIFI wireless communication mode The camera and the user's mobile phone complete the authentication through WIFI communication.
  • the invention firstly performs pre-checking on the behavior of the entering area through audio analysis, vibration detection and infrared detection, and then performs regional entry detection by means of video motion detection, and in order to verify the actual performance of the product, 5 tests are performed, The subtest is 7*24 hours, and compared with the performance of a single sensor. For a single detection method, as long as the alarm is triggered, it is considered that the regional entry behavior has occurred.
  • the test results are as follows:
  • the system turns on authentication and sets the authentication timeout period to 10 seconds. If the user identity is not confirmed within 10 seconds, it is considered an illegal intrusion.
  • the two verification methods of face recognition and WIFI communication were tested respectively. The test results are as follows:

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Burglar Alarm Systems (AREA)

Abstract

一种智能门窗防入侵装置，所述防入侵装置包括照明模块、视频采集模块、综合信息采集模块、报警发送模块、信号处理模块、照明控制模块。该照明模块为视频采集模块的视频采集提供照明光源；所述视频采集模块在照明模块开启时开始采集周围视频信息；所述综合信息采集模块实时采集周围信息，并实时提供给所述信号处理模块；所述照明控制模块决定照明模块的开启，其中所述照明控制模块根据所述信号处理模块对所述周围信息的处理结果来决定所述开启；所述信号处理模块根据所述视频信息和所述周围信息来分辨入侵行为，从而决定是否启动报警发送模块，发出报警信号。该防入侵装置对信息具有综合分析能力，对入侵行为具有智能识别能力，降低了误报。

Description

一种智能门窗防入侵装置以及系统、 智能门禁系统
技术领域 本发明涉及一种防非法入侵系统， 具体涉及一种智能门窗防入侵装置、 防入侵系统及 智能门禁系统。 背景技术 我们知道， 为了防止盗窃， 一般会在小区或者室外走廊等处安装摄像头之类的防盗装 置，例如一种基于智能视频的室内人员闯入的监控和报警系统，主要是通过对摄像头图像的 智能分析， 实现对室内场所移动目标的分析， 对人员的非法闯入达到及时发现和报警， 同时 忽略没有危险的运动， 降低虚警， 还有一种家居智能防盗系统， 其至少由防盗门、 机械式锁 具、 窗户、 露天阳台及阳台门构成。 通过安装在： （1 )防盗门上的门网传感器、 防盗门磁传 感器， （2 )安装在锁具上的锁舌传感器、 钥匙插入 /拔出传感器、 弹子跳动传感器， （3 )安 装在外侧把手组合件内的把手位置传感器，同时设置从露天阳台进入阳台门通道中的阳台路 径探测区，从窗户进入居室的通道中的窗户路径探测区， 实现自动识别居家中的主人与盗贼 的智能家居防盗系统。
现有的门窗防入侵系统存在如下的缺点：
1.单一的防入侵检测技术存在各自的缺陷， 在某些情况下无法正常工作， 一旦无法正 常工作， 整个检测系统处于瘫痪状态， 系统的稳定性差。 如： 智能视频分析检测可能会受到 环境光线变化、 镜面反射等无关运动的干扰， 造成大量的误报； 热释电红外检测易受温度、 强光、 环境复杂运动的干扰； 振动检测易受到风等外力引起门窗振动产生的干扰； 玻璃破碎 检测易受到外界环境杂音的干扰； 门磁检测需要门窗严格闭合，在室内需要通风的情况下将 无法使用， 如炎热的夏天， 用户往往需要开窗通风。
2.传统的防入侵系统无法对盗贼作案前的踩点行为进行预判的能力， 系统的预警能力 差；
3.需要在盗贼入室后才能触发报警， 而不能在实施作案的第一时间触发报警， 降低了 报警的及时性；
4.涉及大量的各式传感器， 安装调试非常复杂， 同时报警信息不直观， 无法得知现场 情况；
5.传统的报警系统在切断外接电源的情况下无法工作， 给非法入侵者以可乘之机。 发明内容 本发明的目的在于提供一种智能门窗防入侵系统， 解决了现有的防盗措施功能单一， 误报率高， 报警不及时以及容易停止工作的问题。
为解决上述的技术问题， 本发明采用以下技术方案：
一种智能门窗防入侵装置， 所述防入侵装置包括照明模块、 视频采集模块、 综合信息 采集模块、 报警发送模块、 信号处理模块、 照明控制模块， 其特征在于： 所述照明模块为视频采集模块的视频采集提供照明光源；
所述视频采集模块在照明模块开启时开始采集周围视频信息；
所述综合信息采集模块实时采集周围信息， 并实时提供给所述信号处理模块； 所述照明控制模块决定照明模块的开启， 其中， 所述照明控制模块根据所述信号处理 模块对所述周围信息的处理结果来决定所述开启；
所述信号处理模块根据所述视频信息和所述周围信息来分辨入侵行为， 从而决定是否 启动报警发送模块， 发出报警信号。
如上所述的防入侵装置， 其特征在于：
还包括电源管理模块， 对所述防入侵装置进行供电， 所述电源管理模块连接外接电源 和蓄电池；
当外接电源正常供电时， 整个产品使用外接电源；
当外接电源被切断时， 所述蓄电池自动启用为所述防入侵装置供电。
如上任一所述的防入侵装置， 其特征在于：
所述综合信息采集模块包括音频采集模块、 热释放红外线检测报警模块、 振动检测模 块、 玻璃破碎检测模块、 门磁检测模块中的一个或多个模块；
所述周围信息包括语音信息、 红外信息、振动信息以及玻璃破碎声音信息、 门磁信息。 如上所述的防入侵装置， 其特征在于：
所述玻璃破碎检测模块利包括玻璃破碎入侵探测器， 所述探测器对高频的玻璃破碎声 音进行有效检测， 而对低频的普通声音信号有较强的抑制作用。
如上任一所述的防入侵装置， 其特征在于：
所述照明模块中的照明灯呈环状， 所述热释放红外线检测报警模块、 振动检测模块、 视频采集模块、音频采集模块和报警发送模块分布安装在所述照明灯的外边缘 ,所述信号处 理模块、 电源管理模块以及开关控制模块安装在所述照明灯的内部。
如上任一所述的防入侵装置， 其特征在于：
所述玻璃破碎检测模块、 门磁检测模块分别通过有线或无线通信方式将检测到的信息 转变成电压信号向所述信号处理模块输出。
如上任一所述的防入侵装置， 其特征在于：
所述报警发送模块是采用有线或无线通信方式，将信号处理模块发出的报警信号发出。 如上所述的防入侵装置， 其特征在于：
所述无线通信方式包括 3G、 WIFK Zigbee. Bluetooth或者其它无线通信方式。
如上所述的防入侵装置， 其特征在于：
所述信号处理模块收集所述热释放红外线检测报警模块、 振动检测模块、 玻璃破碎检 测模块、 门磁检测模块、 视频采集模块和音频采集模块传输过来的信号， 进行综合分析， 从 而完成所述分辨入侵行为， 并通过所述报警发送模块发出报警信号。
如上任一所述的防入侵装置， 其特征在于：
所述信号处理模块包括主处理芯片， 所述主处理芯片采用主核处理器与从核处理器结 合的双核架构模式。
如上所述的防入侵装置， 其特征在于：
所述主核处理器接收视频采集模块和综合信息检测模块采集到的所述视频信息和所述 周围信息；
所述视频信息和所述周围信息构成多源信号传送给从核处理器；
所述从核处理器分辨入侵行为， 形成最终的决策信号。
如上所述的防入侵装置， 其特征在于：
所述分辨入侵行为主要包括运行音视频智能分析算法， 同时结合所述周围信息中的其 他信息进行分辨。
本发明还提出了一种智能防入侵系统， 所述系统包括存储模块和如上任一所述的防入 侵装置； 所述存储模块通过有线或无线通信的方式与所述防入侵装置建立通信连接，对所述 防入侵装置采集的信息进行存储；
其特征在于：
所述分辨入侵行为包括对接近防入侵装置的人进行身份识别， 所述身份识别是通过人 脸识别或其他方式识别来完成； 在进行所述身份识别的同时， 还对在人接近防入侵装置时所述综合信息采集模块采集 到的所述周围信息进行分析；
所述信号处理模块基于所述身份识别和对所述周围信息进行分析， 形成决策信号， 分 辨入侵行为是否成立。
如上所述的智能防入侵系统， 其特征在于：
所述其他方式识别包括通过随身携带的手机来进行身份识别， 所述防入侵装置与手机 进行通信来确定人的身份。
如上所述的防入侵系统， 其特征在于：
在进行所述人脸识别之前， 还包括人脸注册以形成人脸数据库；
所述人脸识别中， 采用 HAAR特征结合 adaboost算法进行人脸检测实现人脸定位； 采用两层人眼定位器， 通过 adaboost算法获得人眼定位， 利用人脸定位结果通过图像 旋转将双眼校正为水平实现人脸标准化；
通过二维 Gabor滤波器进行特征提取，利用向量间的协方差距离作为匹配度量方式，通 过最近邻分类法实现待识别的人脸图像与所述人脸数据库中的人脸图像进行匹配。
如上任一所述的系统， 其特征在于：
所述分辨入侵行为还包括对所述视频信息中的动作进行检测分析。
如上所述的系统， 其特征在于：
所述对动作进行检测分析包括背景建模及运动检测， 目标确认， 目标跟踪、行为分析； 所述背景建模及运动检测， 是采用多帧图像求平均的方式得到背景均值， 将背景的方 差初始化为一个常数， 通过公式
\ (χ, γ) - μ(χ, γ) \> Ν σ(χ, γ)
Figure imgf000005_0001
通过前景检测将背景象素与前景象素分离，其中 1表示前景象素， 0表示背景象素， f(x,y) 表示图像中 (x,y)处象素的灰度值， μ (x,y)表示对应的背景均值， σ (x,y)表示对应背景的方差； 所述目标确认，通过目标尺寸滤波器根据设定的尺寸滤除不满足尺寸要求的前景区域； 通过高亮目标滤波器根据设定阔值滤除高亮的前景区域；将前景区域同对应的背景做梯度差 分， 通过时域滤波器对运动连续性、 运动显著性和面积变化连续性分别进行滤波；
所述目标跟踪， 根据当前帧及之前目标的位置估计目标在下一帧的位置， 通过公式
∑(H —„— i )
N - l
对估计目标速度实现目标位置预测， 其中 和 v 分别表示 t时刻目标在 X方向和 y方向 的速度， N为时间窗口， ^。和 ^分别表示 t-n时刻目标外界矩形框中心的横坐标和纵坐标， 同理 ^^和 ^分别表示 t-n-1时刻目标外界矩形框中心的横坐标和纵坐标；
所述行为分析， 通过用户设定报警区域及相应的规则实现特定行为的判断。
如上任一所述的系统， 其特征在于：
对所述周围信息进行分析包括对所述音频采集模块所采集到的语音信息进行检测分 析，将对分辨入侵有帮助的语音信息作为前景信号从庞杂的作为背景信号的背景音频信息中 提取出来。
如上所述的系统， 其特征在于：
对所述语音信息进行检测分析包括：
取得所述语音信息中的音频信号的功率谱密度， 设音频信号 X(t)的采样率为 fs, 将 X(t) 依次经过子贞加重， 分帧和加窗处理， 在信号处理前首先去除均值， 采用经典 估计中的周期 图法， 使用快速傅里叶变换 FFT实现， 最终得到归一化的功率 X(t)(fn), 其中 n e [l,N] , N 为采样点数， t表示采样时刻； U (!·) 采用子带能量比率 Ai(t)描述功率傳密度， 利用公式 Α(ί) = ΖΧ(ί)(/„)计算出子带能 量比率 Ai(t)定义为子带能量与总能量的比值， 其中 U(i)和 L(i)表示子带 i的上边界和下边界这 里按照对数将频带划分为 M个子带；
基于子带能量比率的混合高斯背景建模及声音检测， 混合高斯背景建模假设信号变化 的概率分布可以用 K个高斯分布拟合， 音频信号的概率分布表示为
κ
ρΓθΗχ) = ^ ωίηί (χ, μί , σί )
i=l
模型中每个高斯分布1^ , 4 1, (7 1)都赋予一个权重0^,其中 和 cj i分别表示高斯分布 的均值和标准差， 多个高斯分布通过线性组合得到信号的概率分布， 将前 B个高斯分布作为 b
背景模型， 剩余的高斯分布作为前景模型， 其中 = & ηώι6(Ζ¾(; > Γ)
将当前信号幅值与前 Β个高斯分布进行匹配， 如果和其中的任何一个高斯分布匹配成 功则该信号为背景信号， 否则为前景信号。
如上任一所述的系统， 其特征在于：
所述系统进行所述身份识别、 对所述周围信息进行分析， 形成决策信号， 分辨入侵行 为是否成立包括如下步骤：
第一步， 通过所述视频采集模块收集数字视频信号， 通过所述音频采集模块收集数字 音频信号， 通过热释放红外线检测报警模块收集外界热量温度变化信息并转变为电压信号， 通过振动检测模块收集外界振动产生形变或受力转变的信息并转变为电压信号，通过玻璃破 碎检测模块和门磁检测模块收集玻璃破碎和门磁变化信息并转变为电压信号；
第二步， 所述信号处理模块接收到热释放红外线检测报警模块、 振动检测模块、 玻璃 破碎检测模块和门磁检测模块的电压信号， 并结合音视频智能分析、 完成身份识别， 基于识 别的身份来综合判定入侵行为的发生， 当存在入侵行为时， 发出报警信号；
第三步，报警发送模块收到报警信号，并通过有线或无线的方式向外部发出报警信息。 如上所述的系统， 其特征在于：
报警发送模块收到报警信号向外部发出报警信息包括：
报警发送模块接收所述数字视频信号和所述数字音频信号， 并将所述数字视频信号和 所述数字音频信号连同警报信号一起发送。
本发明还提出了一种智能门禁系统， 所述智能门禁系统包括如上任一所述的智能防入 侵系统，所述智能门禁系统利用所述智能防入侵系统来完成所述身份识别、对所述周围信息 进行分析， 识别是否为客人到访， 并自动将到访客人到达的信息发送到移动或固定终端。
如上所述的智能门禁系统， 其特征在于：
所述识别是否为客人到访包括如下步骤：
第一步， 通过所述视频采集模块收集数字视频信号， 通过所述音频采集模块收集数字 音频信号， 通过热释放红外线检测报警模块收集外界热量温度变化信息并转变为电压信号， 通过振动检测模块收集外界振动产生形变或受力转变的信息并转变为电压信号，通过玻璃破 碎检测模块和门磁检测模块收集玻璃破碎和门磁变化信息并转变为电压信号；
第二步， 信号处理模块接收到热释放红外线检测报警模块、 振动检测模块、 玻璃破碎 检测模块和门磁检测模块的电压信号， 并结合音视频智能分析判断是否为客人到访；
第三步， 如果存在客人到访行为的发生， 通过无线的方式将访客在门外的视频或照片 信息自动发送到在家室内主人手机等移动或固定终端，并激活终端给出振铃、震动或语音提 示。
与现有技术相比， 本发明的有益效果是： 本发明智能门窗防入侵系统采用热释电红外 检测、 振动检测、 音频检测、 玻璃破碎检测和门磁检测技术， 通过多维探测方式发现门窗区 域可能存在的非法入侵行为， 能准确的分辨出是非法入侵、 客人到访还是户主正常的出入， 将音视频或图片为栽体的实时、直观报警信号通过有线或无线通信手段， 第一时间送达诸如 户主手机、 小区监控计算机等报警接收及处理终端,也同时联动警灯， 警号实现事发现场声 光报警，或者联动其它关联设备； 另外本系统安装于室外， 能够最大程度将非法入侵消灭于 其实际发生之前； 相比于传统的报警系统， 本发明还自带备用电源， 在切断外接电源的情况 下可以为自身供电 2-24个小时， 在敏感时间段仍然能够完成防入侵检测。 附图说明 图 1为本发明智能门窗防入侵系统一个实施例的连接示意图。
图 2为本发明智能门窗防入侵系统的硬件连接示意图。
图 3为本发明明智能门窗防入侵系统的软件结构图。
图 4为本发明智能门窗防入侵系统的信号处理模块中主处理器端流程示意图。
图 5为本发明智能门窗防入侵系统的信号处理模块中从处理器端流程示意图。
图 6为本发明智能门窗防入侵系统的人脸注册和人脸识别流程示意图。
图 7为本发明智能门窗防入侵系统的基于视频的运动目标检测与跟踪流程示意图。 图 8为本发明智能门窗防入侵系统的音频检测流程示意图。
图 9为本发明智能门窗防入侵系统的人脸标准化时对接取人脸图像分析示意图。 具体实施方式 为了使本发明的目的、 技术方案及优点更加清楚明白， 以下结合附图及实施例， 对本 发明进行进一步详细说明。 应当理解， 此处所描述的具体实施例仅仅用以解释本发明， 并不 用于限定本发明。
图 1示出了本发明一种智能门窗防入侵系统的一个实施例：一种智能门窗防入侵系统， 包括
视频采集模块， 采用视频传感器采集视频流信号， 完成图像信号的数字化， 图像信号 的预处理，得到满足信号处理模块要求的数字视频信号并向信号处理模块输出其具体的检测 方式可以是区域方式或幕帘方式；
音频采集模块， 完成语音信号的数模转换， 语音信号的采样编码及滤波处理， 得到满 足信号处理模块要求的数字音频信号并向信号处理模块输出；
热释放红外线检测报警模块， 感应移动物体与背景物体的温度的差异， 当人体移动时 热释红外能够感应到人体温度与背景温度的差异信息，转换成电压信号后向信号处理模块输 出；
振动检测模块， 将外界振动产生形变或受力信息转变为电压信号， 向信号处理模块输 出；
玻璃破碎检测模块和门磁检测模块， 通过有线或无线通信方式将相应的信息转变成电 压信号向信号处理模块输出；
信号处理模块， 收集热释放红外线检测报警模块、 振动检测模块、 视频采集模块和音 频采集模块传输过来的信号， 进行综合分析， 分辨非法入侵或客人到访， 分别向报警发送模 块发出报警信号或向在家主人发送访客的照片等信息；
报警发送模块， 采用有线或无线通信方式， 主要用于接收信号处理模块发出的报警信 号及向户主手机或小区管理中心发送报警信息， 可采用有线、 3G、 WIFL Zigbee. Bluetooth 或者其它无线通信方式之一；
电源管理模块， 连接外接电源和蓄电池， 当外接电源正常供电时， 整个产品使用外接 电源， 当外界电源断电时， 内部备用蓄电池自动启用， 保证上述各模块在外接电源中断后继 续正常工作， 最好是保证摄像机在外接电源中断后继续工作 2-24小时。
在图 1中， 可以看出， 发明设置在环形灯具上,所述热释放红外线检测报警模块、振动 检测模块、视频采集模块、音频采集模块和报警发送模块安装在环形灯具的外边缘， 所述信 号处理模块、 电源管理模块以及简型灯具本身的开关灯控制模块安装在简型灯具的中部。所 述开关灯控制模块可以结合振动， 声音， 红外等信号判断是否有人进入， 如果有人进入， 开 启灯光照明， 为后续视频采集提供环境光照。 我们是将整个系统制作成类似照明灯的样子， 系统和灯的相互集合， 不仅能将本系统很好的隐藏起来， 防止不法分子对本智能门窗防入侵 系统进行破坏，还能对系统中视频采集模块的摄像头进行补光，让采集的视频信息更加清晰， 在环境照度不足时，给视频采集模块提供辅助光源，保证其在低照度的环境下仍然能够采集 到有效的图像数据。
图 2示出了本发明一种智能门窗防入侵系统的一个优选实施例， 信号处理模块， 主处 理芯片采用主核处理器与从核处理器结合的双核架构模式。主核处理器主要完成数字视频信 号和数字音频信号的收集，音视频编码，收集热释放红外线检测报警模块输出的信号和振动 检测模块输出的信号，接收玻璃破碎检测模块和门磁检测模块发送的信号，将多源信号连同 音视频数据传送给从核处理器；从核处理器进行入侵行为的判断和户主身份验证，其中入侵 行为判断主要包括运行音视频智能分析算法，通过视频和声音判断是否存在入侵行为， 同时 结合热释放红外线检测报警模块输出的信号、振动检测模块输出的信号、玻璃破碎检测模块 和门磁检测模块发送的信号， 形成最终的决策信号。
如图 3所示， 软件主要由信号采集， 信号分析， 报警发送和联动控制三部分组成。 信 号采集和报警发送运行在主核处理端，信号分析运行在从核处理端。信号采集接口包括了音 视频采集模块， 热释电红外报警信号采集模块， 振动报警信号采集模块、玻璃破碎检测模块 和门磁检测模块及无线报警信号采集模块。主核处理端完成开关量报警信号的采集及延时同 步， 连同音视频信号传送到从核处理端。从核处理端完成智能音视频分析， 多源报警信息融 合及决策， 通过人脸识别方式或无线 WIFI方式完成户主身份的确认， 在检测到入侵行为时 如果用户身份确认失败则向主核处理端发送报警 /预警信号， 主核处理端通过报警发送模块 向外部终端发送报警和联动控制信息。
如图 4所示， 主核处理端软件流程：
系统上电后， 首先完成主芯片主核处理端的初始化， 接下来完成外围采集设备， 通信 设备初始化， 启动主核处理程序。 创建音视频采集线程、 热释电红外检测线程、 无线接收线 程、 通信线程、 报警发送线程。
音视频采集线程： 当接收到音视频数据后组合一帧多源数据， 包括视频数据、 音频数 据、各种传感器报警强度及用户手机是否存在的标识，放入 FrameBuffer中供通信线程调用， 同时完成音视频的预录， 以便提供报警信息。
报警信号采集线程： 接收各种有线或无线接入的报警传感器的开关量报警信号 Ai, 其 中 Ai可以是热释电红外报警信号， 振动报警信号， 玻璃破碎报警信号， 门磁报警信号等。
通信线程： 接收从核处理器端发来的报警信号， 通知报警发送线程。如果 FrameBuffer 中存在一帧多源数据， 则将该数据发送到 DSP端。
报警发送线程： 接收到报警信号后， 组织相关的报警信息， 如预录的音视频， 图片等， 向外部终端发送。
如图 5所示， 从核处理器端软件流程：
首先主核处理器端启动后， 将多任务内核及应用程序加栽到从核处理器内存中， 从核 处理器完成一系列初始化， 自动创建通信线程及人脸识别线程。通信线程首先判断是否需要 发送报警 /预警信号， 如果是则向主处理端发送报警 /预警信号， 否则判断是否收到一帧多源 信息， 其中多源信息包括： 音视频数据， 各种传感器报警信号及用户手机是否存在的标识， 将多源信息放入帧緩存 FrameBuffer中， 同时向主处理端发送接收确认信号， 表明从处理端 工作正常。人脸识别线程从 FrameBuffer中取出视频数据完成人脸识别，如果人脸识别成功， 则置位身份确认的标识。 入侵检测线程从 FrameBuffer中读取一帧数据， 通过音频分析， 振 动信息， 红外信息判断是否有人将要进入监控区域， 如果是则开启照明， 为接下来的视频分 析提供环境光照，同时方便户主开门。开启照明后开始运动检测，由于只关心进门这一行为， 所以需要进行运动方向判断， 当判断为进门方向时， 对目标进行跟踪。 当目标存在时间大于 设定时间时， 说明为可疑人员， 如果 WIFI方式身份确认失败则系统会通过语音提示主动配 合进行人脸识别， 如果身份验证失败则触发入侵行为报警， 向 ARM端发送报警 /预警信号。
图 6示出了本发明一种智能门窗防入侵系统的另一个优选实施例， 户主身份判断包括 人脸识别或在用户手机安装 WIFI通信软件， 视频采集模块与用户手机通过 WIFI通信进行 身份确认， 其中所述人脸识别的方法主要包括以下部分：
人脸注册， 采集同一个人的不同角度的正面人脸照片， 通过人脸定位找到人脸在图像 中的位置， 并标准化人脸尺寸， 人脸角度和光照， 提取标准化人脸图像的特征， 录入注册用 户数据库；
人脸识别， 采用 HAAR特征结合 adaboost算法实现人脸检测实现人脸定位， 采用了 两层人眼定位器， 都是通过 adaboost算法获得人眼定位， 利用人脸定位结果通过图像旋转 将双眼校正为水平实现人脸标准化，通过二维 Gabor滤波器进行特征提取，利用向量间的协 方差距离作为匹配度量方式，通过最近邻分类法实现查询的人脸图像与数据库人脸图像的匹 配。 其中 Adaboost算法将大量分类能力一般的弱分类器按照训练误差指数下降的方式组合 为一个强分类器。而 HAAR特征为 adaboost算法提供了海量的弱分类特征，保证了 adaboost 算法总体找到性能优异的弱分类。在人脸检测实施过程中，积分直方图和级联分类器的使用 在保证较高检测精度的同时大大降低了处理时间；人眼定位时一般分为两层定位，其中第一 层为粗定位， 定位区域选择了包括了眼睛眉毛在内的大部分眼部区域， 第二层为精确定位， 定位区域只包含眼部区域。粗定位器相比于精确定位器由于包含了更多了区域信息， 因此定 位的稳定性更高， 基本不存在较大的位置偏差， 而精确定位器能够实现人眼的精确定位， 但 是容易受到眉毛、眼角的干扰造成定位错误。在粗定位的基础上通过几何比例关系确定人眼 的大致位置范围，在该范围内使用精确定位器实现人眼的精确定位。通过由粗到精的定位方 式， 减小了眉毛、 眼角等对定位的影响， 提高了定位的准确性； 而人脸标准化是人脸识别中 非常关键的一个步骤，标准化结果的好坏直接影响了人脸识别的精度。人脸标准化主要完成 人脸图像的几何校正及亮度校正。利用上一步人眼定位的结果很容易实现人脸图像的几何校 正， 首先通过图像旋转将双眼校正为水平， 通过双眼距离 d对人脸图像进行截取。 如图 9 所示， 其中最后将图像缩放到 80x80象素。
亮度校正主要是在一定程度上消除光照不均对后续识别的影响。 主要包括光照面拟 合， 扣除光照面， 直方图均衡及灰度值归一化到零均值， 单位方差。 这里假设光照面是一个 平面。 光照面上的点满足如下公式： /5 = + ^ + (写成矩阵形式即"¾ = , 其中 表示图像的象素点灰度值排成的列向量， N表示象素点对应的坐标， 第一列表示横坐标， 第二列表示纵坐标， 第三列填充 1 , p = [a b CF。 平面参数 a, b, c可以通过线性回归 的方式求得，
Figure imgf000009_0001
可以选择了 Gabor小波特征， Gabor变换在分析图像局部区域纹理方面具有优异的 性能。 二维 Gabor滤波器 ^ ( 可以表示为：
Figure imgf000009_0002
Figure imgf000010_0001
其中 为图像坐标， k为滤波器的中心频 ， 和 Ky分别表示 k在横轴和纵轴的投
影，％为滤波器的方向， "和 v代表不同的取值，
Figure imgf000010_0002
为复数值平面波。二维 Gabor滤波器通过二维高斯函数调制特定频率和方向的正弦波平面实 现， 通过改变正弦波平面的频率和方向实现不同尺度和不同方向图像纹理的分析。
通过人脸标准化得到了 80x80大小的人脸图像， 这里选择了 5个滤波器尺度， 8个滤 波方向， 得到 40个不同方向和频率的 Gabor滤波器， 对一张人脸图像通过滤波器卷积后得 到 40张 Gabor小波变换后的幅值图像， 最后得到的 Gabor特征维数为 163840。 这样一个高 维的特征向量中会大大降低识别分类的速度， 因此需要对特征向量进行降维。 这里采用 4x4 均匀向下采样实现特征降维。
采用最近邻分类法实现查询的人脸图像与数据库人脸图像的匹配， 采用向量间的协方 差距离作为匹配度量方式， 同时可以通过协方差距离衡量最终匹配的可信度。
根据本发明一种智能门窗防入侵系统的另一个优选实施例， 信号处理模块 连接有存储模块（ Storage ), 针对不同的应用情况， 会用到不同类型的存储介质， SD 和 TF卡通过 SDIO来控制， 此类型的存储介质方便更换； Nand通过 Nandflash控制， 此类 型集成度高， 但是不易于更换存储介质； SSD通过 PCI-E或者 SATA接口控制， 此类型存储 空间可以做到很大；
设置有以太网接口 （RJ45 );
通过电源管理模块 ( Power manager )连接有外接电源 （DC IN )和电池（Battery ); 连接有用于主控 CPU运行的内存（ DDR );
连接有用于存储系统启动程序、 配置参数、 日志信息的闪存（ Flash );
连接有无线模块 (3G、 WIFI), 用于音视频数据以及远程控制信号的传输； 连接有无线模块 (zigbee、 Blue tooth或其他无线模块)， 用于接收分布式报警器发送的 报警信号， 如玻璃破碎信号和门磁打开信号， 也可向其它设备发送控制信息。
连接有通过外接警灯， 警号联动设备， 实现本地声光报警,或控制其它关联设备的联动 报警模块。
图 7示出了本发明一种智能门窗防入侵系统的另一个优选实施例， 视频采集模块的检 测方法主要分为背景建模及运动检测， 目标确认， 目标跟踪和行为分析四部分， 其中
所述背景建模及运动检测， 是采用了多帧图像求平均的方式得到背景均值， 将背景的 方差初始化为一个常数， 通过公式
1 \ f (x, y) - μ(χ, γ) \> Ν σ(χ, γ)
0 else
将前景检测将背景象素与前景象素分离， 其中 1 表示前景象素， 0表示背景象素。 表示图像中（X， 处象素的灰度值， 表示对应的背景均值， σ( 应背景的方差，
如果检测到大面积前景象素则系统认为场景可能发生了变化， 这时关闭后续的分析， 如果长时间都检测到大面积前景象素则系统认为场景确实发生了变化，这时重新开始背景建 模。
本系统的背景的更新采用了基于目标层面和基于象素层面相结合的背景更新策略。 当 象素点被检测为背景点时， 采用基于象素层面的背景更新， 这里采用单高斯背景更新方法， 包括背景均值的更新和背景方差的更新， 如下公式所示：
M, (x, y) = d - ) x μ{_χ (χ, y) + ft (x, y) x a
。 x, y) = max(A， (l - a) x a _, (x, y) + (ft (x, y) - μ人 x, y)f x a)
其中下标^表示 f时刻， "为更新因子， 一般是一个很小的值， 如 0.01。 这里钳位了 方差的最小值为△, 防止模型过分收敛， 导致模型抗噪声能力减弱， 这里△的取值为 36。
当象素点被检测为前景点时， 如果通过目标分割判断象素点不是目标象素点则采用基 于象素层面的背景更新 ,即如果该象素点在固定的时间内长时间检测为前景象素点则将象素 点一次性更新到背景中；如果通过目标分割判断象素点是目标象素点则采用基于目标层面的 背景更新方法， 即只有判断目标在很长时间没有发生运动时， 才会将目标所在的区域象素点 一次性更新到背景中。
所述目标确认， 通过目标尺寸滤波器根据设定的尺寸滤除不满足尺寸要求的前景区 域，通过高亮目标滤波器根据设定阔值滤除高亮的前景区域，通过时域滤波器对运动连续性、 运动显著性和面积变化连续性分别进行滤波。 其中高亮目标滤波器： 滤除高亮的前景区域， 这些前景区域可能是光源或者反光介质等干扰。 高亮前景区域的判断方法： 如果前景区域中 高亮象素的比例超过设定阔值， 则认为该前景区域是高亮前景区域； 时域滤波器： 时域滤波 是基于目标层面的滤波，需要借助目标跟踪技术。 包括：运动连续性滤波，运动显著性滤波， 面积变化连续性滤波。运动连续性滤波: 支设目标在前后两帧的所占据区域是大面积重叠的， 即目标的位置不会发生突变。运动显著性滤波： 假设目标的运动是有一定的方向性的， 即在 一段时间内目标位移的矢量加和与目标位移的标量加和的比值应该大于一定的阔值。面积变 化连续性滤波： 假设目标的面积是相对稳定的， 不会发生突变。
所述目标的跟踪， 采用采用了基于运动检测的目标跟踪， 在运动检测的基础上通过帧 间的目标关联， 形成目标的运动轨迹， 为后续目标行为分析提供依据， 根据当前帧及之前目
Σ
V； = ^2
标的位置估计 目 标在下一帧 的位置 ， 通过公式 N - \ 和
∑(Λ—„ _ —„— 1 )
' N - 1 对估计目标速度实现目标位置预测， 其中 和 W分别表示 t时刻目 标在 X方向和 y方向的速度， N为时间窗口， 和 -"分别表示 t-n时刻目标外界矩形框 中心的横坐标和纵坐标， 同理 和 -"-1分别表示 t-n-1时刻目标外界矩形框中心的横坐 标和纵坐标。 特征选择方面这里选择了相对稳定可靠的目标面积 S 和目标外接矩形框的中 心 C 两个特征， 目标关联匹配就可以找到前一帧目标在当前帧的位置， 实现目标的定位。 根据相关跟踪原理，只要保证足够图像采样率的情况下， 同一个目标在相邻两帧之间的位置 变化不会太大， 因此可以将目标的搜索范围限定在一个较小的距离范围， 同时也大大降低了 误匹配的风险。利用 ^ (Q {C}'Q;— lie}) < r公式，当前帧目标 Q 与待匹配的目标0; - 1满 足下式时才进行特征匹配， 其中 为搜索距离阔值， 需要根据实际场景进行确定。
匹 arg < Τ
Figure imgf000012_0001
其中 ω为特征的权重因子， 这里取 0.5 , 这里设置了匹配的误差上限 Τ, 防止误匹配， 这里取 0.4。
目标特征的更新均采用一阶平滑的方式更新， 如下公式所示：
Ot{F} =axOt_1{F}+(l-a)xOt{F} F={S,Q
其中"为更新因子， 这里取 0.2。
所述行为分析， 通过用户设定报警区域及相应的规则实现特定行为的判断。 例如可以 将报警区域设置为门所在区域，进入该区域的人如果停留时间超过设定值则认为是可能的入 侵行为， 触发报警。 也可以将门所在的较大区域设置为徘徊检测区域， 进入该区域的人如果 停留时间超过设定值则认为是可能的踩点行为， 触发预警。
图 8示出了本发明一种智能门窗防入侵系统的另一个优选实施例， 音频采集模块的检 测方法，
基于声音的入侵检测流程如下图所示， 通过音频采集设备得到的音频信号， 首先进行 预处理， 主要包括子贞加重， 分帧和加窗， 满足后续频谱分析及识别的需要， 接下来采用混合 高斯模型建立背景声音模型，建模完成后开始声音检测， 当检测到场景中出现不同于背景声 音的异常声响时， 报警强度开始累加， 如果报警强度大于设定阔值， 说明存在入侵行为， 触 发报警， 具体如下， 取得音频信号的功率谱密度， 设音频信号 ^^)的采样率为 ， （例如取 48kHz ), 将
XW依次经过子贞加重， 分帧和加窗处理， （帧长为 48K个采样点， 帧移为 24K个采样点， 窗函数选择汉宁窗）， 在信号处理前首先去除均值， 避免直流分量对 = 0处附近的谱线产 生影响， 采用经典 估计中的周期图法， 使用快速傅里叶变换 FFT实现， 最终得到归一化 的功率谱 χ (ί)(Λ) , 其中 " e [l，N] , N为采样点数， t 表示采样时刻， （当采样率为 =48kHz, N的取值为 48000, 后续分析来说计算量较大），采用子带能量比率 A "描
述功率谱密度， 利用公式
Figure imgf000012_0002
定义为子带能量与总 能量的比值， 其中 U (i 和 LW表示子带 的上边界和下边界这里按照对数将频带划分为 M 个子带（当 M=10, =48kHz时对应的子带边界为 0.023kHz,0.046kHz,0.093kHz,0.187kHz, 0.375kHz,0.750kHz,1.5kHz,3kHz,6kHz, 12kHz和 24kHz ); 基于子带能量比率的混合高斯背景建模及声音检测， 混合高斯背景建模假设信号变化 的概率分布可以 用 K 个高斯分布拟合， 音频信号的概率分布表示为 κ
prob(x) = ωίηί (χ, μί , σ; )
ί=ι , 模型中每个高斯分布 H A' 0^都赋予一个权重 , 其中
Α和 分别表示高斯分布的均值和标准差， 多个高斯分布通过线性组合得到信号的概率分 布， 体现了模型对于第 i个高斯分布的重视程度， 当一段时间内信号的幅值落在该高斯 分布内时， 的值会逐渐增大， 反之 会逐渐减小。 σ反映了高斯分布的收敛性， 对于一 个趋于收敛的高斯分布来说， 幅值围绕均值的波动应该较小， °"也较小， 这样的高斯分布 才适合描述背景的变化。 Κ个高斯分布按照 *^°"的降序排列， 排列靠前的高斯分布最能够 代表背景的分布， 将前 Β个高斯分布作为背景模型， 剩余的高斯分布作为前景模型， 其中 b
5 = arg minfc ωη (χ) > Τ)
»=! , 将当前信号幅值与前 Β 个高斯分布进行匹配， 如果和其中 的任何一个高斯分布匹配成功则该信号为背景， 否则为前景信号， 利用
I I(x， t) - μί (χ, t - ί) \< c x σί (x, t - 1) 进行匹配。
根据本发明一种智能门窗防入侵系统的另一个实施例， 智能门窗防入侵系统的防入侵 方法， 步骤如下：
第一步，通过视频采集模块收集数字视频信号，通过音频采集模块收集数字音频信号， 通过热释放红外线检测报警模块收集外界热量温度变化信息并转变为电压信号，通过振动检 测模块收集外界振动产生形变或受力转变的信息并转变为电压信号，通过玻璃破碎检测模块 和门磁检测模块收集玻璃破碎和门磁变化信息闭并转变为电压信号；
第二步， 信号处理模块接收到热释放红外线检测报警模块、 振动检测模块、 玻璃破碎 检测模块和门磁检测模块的电压信号， 并结合音视频智能分析、主人身份识别验证， 综合判 定入侵行为的发生， 当存在入侵行为时， 发出报警信号；
第三步， 报警发送模块收到报警信号， 同时接收视频采集模块收集数字视频信号和通 过音频采集模块收集数字音频信号，并通过有线或无线的方式传输到外部管理终端或手持终 端。
根据本发明一种智能门窗防入侵系统的另一个实施例， 智能门窗防入侵系统其用作可 视门禁的方法如下：
第一步，通过视频采集模块收集数字视频信号，通过音频采集模块收集数字音频信号， 通过热释放红外线检测报警模块收集外界热量温度变化信息并转变为电压信号，通过振动检 测模块收集外界振动产生形变或受力转变的信息并转变为电压信号，通过玻璃破碎检测模块 和门磁检测模块收集玻璃破碎和门磁变化信息闭并转变为电压信号；
第二步， 信号处理模块接收到热释放红外线检测报警模块、 振动检测模块、 玻璃破碎 检测模块和门磁检测模块的电压信号，并结合音视频智能分析判断是否为客人到访，如果存 在客人到访行为的发生，则通过无线的方式将访客在门外的视频或照片信息自动发送到在家 室内主人手机等移动或固定终端， 并激活终端给出振铃、 震动或语音提示。
在本发明中： 视频采集模块利用的智能视频分析技术， 即采用计算机、 嵌入式双核处理器等从视频 中通过运算和分析，提取视频中有用信息的一项技术。该技术借助各种处理器强大的数据处 理能力， 对视频画面中的海量数据进行高速分析， 过滤掉用户不关心的信息， 仅仅为使用者 提供有用的关键信息。
红外线检测报警模块利用的热释电红外检测技术， 即被动红外探测的一种， 感应移动 物体与背景物体的温度的差异。监控区域内， 无人体移动时， 热释电红外感应器感应到的只 是背景温度， 当人体进入区域时，人体发射的红外线通过菲涅尔透镜增强后聚集到热释电红 外感应器上，热释电红外中的热释电元件感应到人体温度与背景温度的差异，温度发生变化， 失去电荷平衡， 向外释放电荷， 后续电路检测处理后产生报警信号。
振动检测模块利用的振动防盗报警技术： 采用压电式陶瓷片作为传感器， 将外界振动 产生形变或受力转变为电压信号， 通过后续辅助电路完成报警信号的发送和停止。
玻璃破碎检测检测模块所利用的玻璃破碎检测技术： 利用压电陶瓷片的压电效应， 可 以制成玻璃破碎入侵探测器。 对高频的玻璃破碎声音（ 1 Ok ~ 15kHZ )进行有效检测， 而对 lOkHZ 以下的声音信号 （如说话、 走路声）有较强的抑制作用。 通常将将声控与震动探测 两种技术组合在一起，只有同时探测到玻璃破碎时发出的高频声音信号和敲击玻璃引起的震 动， 才输出报警信号。
门磁检测模块利用的门磁检测技术： 由两部分组成： 永久磁铁和常开型的干簧管， 永 久磁铁用来产生恒定的磁场，当永磁体和干簧管靠得很近时，门磁传感器处于工作守候状态， 当永磁体离开干簧管一定距离后， 处于常开状态， 触发报警。
本系统工作流程是：
用户可以根据需要开启或关闭报警功能， 检测的方式可以选择视频检测、 音频检测、 热释电红外检测， 振动检测， 玻璃破碎检测， 门磁检测中的一种或几种， 如窗户区域采用热 释电红外检测、振动检测、玻璃破碎检测、 门磁检测， 门所在区域采用视频检测、音频检测、 热释电红外检测、振动检测、 门磁检测。报警信息可以是视频，音频或图片中的一种或几种。
( 1 )系统上电或复位后， 信号处理模块从 FLASH中加栽操作系统和应用程序， 完成 对主处理芯片的初始化和外围硬件的配置，接下来完成对各子系统的初始化，最后进入正常 工作状态。
( 2 )主处理芯片的主处理器端不断的采集监控现场的音视频信号及热释红外检测信 号， 同时通过无线接收模块接收其它分布式报警传感器的报警信号，将多源数据送入从处理 器端进行分析， 同时进行音视频预录。 如果收到从处理器端的报警或预警信号， 则将预录的 音视频， 连同抓拍的照片通过报警信号发送模块发送到小区监控中心或户主的手机上。
( 3 )主处理芯片的从处理器端运行智能音视频分析算法及入侵行为判断。 通过音频 分析， 振动信息， 红外信息判断是否有人将要进入监控区域， 如果是则开启照明， 为接下来 的视频分析提供环境光照， 同时方便户主开门。 开启照明后开始运动检测， 由于只关心进门 这一行为， 所以需要进行运动方向判断， 当判断为进门方向时， 对目标进行跟踪。 当目标存 在时间大于设定时间时， 说明为可疑人员， 如果 WIFI方式身份确认失败则系统会通过语音 提示主动配合进行人脸识别，如果身份验证失败则触发入侵行为报警， 向主处理器端发送报 警 /预警信号。
( 4 )监控中心和户主收到预警或报警后， 可以通过传回来的报警信息进行确认， 并 采取应对措施。
实验：
本系统的安装： 吸顶安装， 保证监控范围能够覆盖整个门所在区域及门前的部分过道 区域， 调整拍摄角度保证监控视野内能够得到近似正面的人像， 照明条件良好， 能够得到清 晰的人脸图像， 人脸象素面积大于 80*80象素。 本发明判断进入行为发生的两个依据是：
监控区域出现了可疑人员： 通过多种报警传感器相结合的方式实现可疑人员的 检测。
可疑人员身份验证失败： 户主可以通过两种方式完成身份验证： 视频人脸识 别方式和 WIFI无线通信方式。 视频人脸识别方式： 用户进入区域后主动配合摄像机 完成人脸识别， 通过与注册的人脸库进行比对， 完成身份的验证。 WIFI无线通信方式： 摄 像机与用户手机通过 WIFI通信完成身份验证。
因此将实验分为两部分： 区域进入检测和身份验证。
区域进入检测实验
本发明首先通过音频分析， 振动检测， 红外检测对进入区域的行为进行预检， 接下来 通过视频运动检测进行复核的方式实现区域进入检测， 为了验证产品的实际性能， 进行了 5 次测试， 每次测试为 7*24小时， 同时和单一传感器的性能进行了比较， 对于单一检测方式 来说只要触发了报警， 则认为是发生了区域进入行为， 测试结果如下：
Figure imgf000015_0001
通过实验证明了多种报警传感器的有效结合相比于单一的报警器无论在检测精度方 面， 还是在减小误报方面都有很大的提高。
身份验证实验
当检测到入侵行为时， 系统开启身份验证， 设定验证超时时间为 10秒， 在 10秒内如 果没有确认用户身份， 则认为是非法入侵。 分别对人脸识别和 WIFI通信这两种验证方式进 行了测试， 测试结果如下：
Figure imgf000015_0002
在实验中发现， 人脸识别对人脸姿态， 光照环境的要求都比较高， 如果环境侧光严重 或者人员姿态配合不默契会造成无法识别或错识别，而通过 WIFI通信的方式无需用户配合， 但是可能会受到其它 WIFI设备的干扰造成连接不成功， 通过实验说明， 采用 WIFI通信进 行身份确认优于人脸识别实现身份确认， 将两者结合可以进一步提高身份确认的准确率。
尽管这里参照本发明的多个解释性实施例对本发明进行了描述， 但是， 应该理解， 本 领域技术人员可以设计出很多其他的修改和实施方式，这些修改和实施方式将落在本申请公 开的原则范围和精神之内。 更具体地说， 在本申请公开、 附图和权利要求的范围内， 可以对 主题组合布局的组成部件和 /或布局进行多种变型和改进。 除了对组成部件和 /或布局进行的 变型和改进外， 对于本领域技术人员来说， 其他的用途也将是明显的。

Claims

权利要求
1、 一种智能门窗防入侵装置， 所述防入侵装置包括照明模块、 视频采集模块、 综合信 息采集模块、 报警发送模块、 信号处理模块、 照明控制模块， 其特征在于：
所述照明模块为视频采集模块的视频采集提供照明光源；
所述视频采集模块在照明模块开启时开始采集周围视频信息；
所述综合信息采集模块实时采集周围信息， 并实时提供给所述信号处理模块； 所述照明控制模块决定照明模块的开启， 其中， 所述照明控制模块根据所述信号处理 模块对所述周围信息的处理结果来决定所述开启；
所述信号处理模块根据所述视频信息和所述周围信息来分辨入侵行为， 从而决定是否 启动报警发送模块， 发出报警信号。
2、 如权利要求 1所述的防入侵装置， 其特征在于：
还包括电源管理模块， 对所述防入侵装置进行供电， 所述电源管理模块连接外接电源 和蓄电池；
当外接电源正常供电时， 整个产品使用外接电源；
当外接电源被切断时， 所述蓄电池自动启用为所述防入侵装置供电。
3、 如权利要求 1或 2任一所述的防入侵装置， 其特征在于：
所述综合信息采集模块包括音频采集模块、 热释放红外线检测报警模块、 振动检测模 块、 玻璃破碎检测模块、 门磁检测模块中的一个或多个模块；
所述周围信息包括语音信息、 红外信息、振动信息以及玻璃破碎声音信息、 门磁信息。
4、 如权利要求 1-3任一所述的防入侵装置， 其特征在于：
所述照明模块中的照明灯呈环状， 所述热释放红外线检测报警模块、 振动检测模块、 视频采集模块、音频采集模块和报警发送模块分布安装在所述照明灯的外边缘 ,所述信号处 理模块、 电源管理模块以及开关控制模块安装在所述照明灯的内部。
5、 如权利要求 1-4任一所述的防入侵装置， 其特征在于：
所述信号处理模块包括主处理芯片， 所述主处理芯片采用主核处理器与从核处理器结 合的双核架构模式。
6、 一种智能防入侵系统， 所述系统包括存储模块和如权利要求 1-5任一所述的防入侵 装置； 所述存储模块通过有线或无线通信的方式与所述防入侵装置建立通信连接，对所述防 入侵装置采集的信息进行存储；
其特征在于：
所述分辨入侵行为包括对接近防入侵装置的人进行身份识别， 所述身份识别是通过人 脸识别或其他方式识别来完成；
在进行所述身份识别的同时， 还对在人接近防入侵装置时所述综合信息采集模块采集 到的所述周围信息进行分析；
所述信号处理模块基于所述身份识别和对所述周围信息进行分析， 形成决策信号， 分 辨入侵行为是否成立。
7、 如权利要求 6所述的防入侵系统， 其特征在于：
在进行所述人脸识别之前， 还包括人脸注册以形成人脸数据库；
所述人脸识别中， 采用 HAAR特征结合 adaboost算法进行人脸检测实现人脸定位； 采用两层人眼定位器， 通过 adaboost算法获得人眼定位， 利用人脸定位结果通过图像 旋转将双眼校正为水平实现人脸标准化；
通过二维 Gabor滤波器进行特征提取，利用向量间的协方差距离作为匹配度量方式，通 过最近邻分类法实现待识别的人脸图像与所述人脸数据库中的人脸图像进行匹配。
8、 如权利要求 6-7任一所述的系统， 其特征在于：
所述分辨入侵行为还包括对所述视频信息中的动作进行检测分析。
9、 一种智能门禁系统， 所述智能门禁系统包括如权利要求 6-7任一所述的智能防入侵 系统，所述智能门禁系统利用所述智能防入侵系统来完成所述身份识别、对所述周围信息进 行分析， 识别是否为客人到访， 并自动将到访客人到达的信息发送到移动或固定终端。
10、 如权利要求 9所述的智能门禁系统， 其特征在于：
所述识别是否为客人到访包括如下步骤：
第一步， 通过所述视频采集模块收集数字视频信号， 通过所述音频采集模块收集数字 音频信号， 通过热释放红外线检测报警模块收集外界热量温度变化信息并转变为电压信号， 通过振动检测模块收集外界振动产生形变或受力转变的信息并转变为电压信号，通过玻璃破 碎检测模块和门磁检测模块收集玻璃破碎和门磁变化信息并转变为电压信号；
第二步， 信号处理模块接收到热释放红外线检测报警模块、 振动检测模块、 玻璃破碎 检测模块和门磁检测模块的电压信号， 并结合音视频智能分析判断是否为客人到访；
第三步， 如果存在客人到访行为的发生， 通过无线的方式将访客在门外的视频或照片 信息自动发送到在家室内主人手机等移动或固定终端，并激活终端给出振铃、震动或语音提 示。
PCT/CN2014/073258 2013-03-11 2014-03-11 一种智能门窗防入侵装置以及系统、智能门禁系统 WO2014139415A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN2013100759135A CN103198595A (zh) 2013-03-11 2013-03-11 智能门窗防入侵系统
CN201310075913.5 2013-03-11

Publications (1)

Publication Number Publication Date
WO2014139415A1 true WO2014139415A1 (zh) 2014-09-18

Family

ID=48721085

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/073258 WO2014139415A1 (zh) 2013-03-11 2014-03-11 一种智能门窗防入侵装置以及系统、智能门禁系统

Country Status (2)

Country Link
CN (2) CN103198595A (zh)
WO (1) WO2014139415A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105578134A (zh) * 2015-12-16 2016-05-11 武汉烽火众智数字技术有限责任公司 一种视频采集系统及方法
CN106205378A (zh) * 2016-08-26 2016-12-07 吕跟来 智能多功能电子门牌系统
CN108062132A (zh) * 2017-12-25 2018-05-22 广东星鼎通信科技有限公司 一种基于自动化控制技术的通信控制室
CN109101370A (zh) * 2018-08-30 2018-12-28 重庆富民银行股份有限公司 主从双向同步的数据存储方法
CN109271280A (zh) * 2018-08-30 2019-01-25 重庆富民银行股份有限公司 存储故障快速切换处理方法
US11735018B2 (en) 2018-03-11 2023-08-22 Intellivision Technologies Corp. Security system with face recognition

Families Citing this family (83)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105245822A (zh) * 2014-06-24 2016-01-13 陈凯柏 用影像辨识系统控制的照明装置
CN103198595A (zh) * 2013-03-11 2013-07-10 成都百威讯科技有限责任公司 智能门窗防入侵系统
CN103455147B (zh) * 2013-09-10 2016-08-31 惠州学院 一种光标控制方法
CN103487622B (zh) * 2013-09-25 2016-08-17 国网河南镇平县供电公司 一种防止盗电及具有监控功能的电表箱装置
CN104658007A (zh) * 2013-11-25 2015-05-27 华为技术有限公司 一种实际运动目标的识别方法及装置
CN103593927A (zh) * 2013-11-29 2014-02-19 赵智朋 一种家用防盗报警器
CN104754290A (zh) * 2013-12-30 2015-07-01 南京理工大学常熟研究院有限公司 一种网络视频监控装置
WO2015145299A1 (en) * 2014-03-27 2015-10-01 Koninklijke Philips N.V. Detection and notification of pressure waves by lighting units
CN104122863A (zh) * 2014-07-17 2014-10-29 江苏惠居乐信息科技有限公司 面向家居装修的远程智能监控系统及其控制方法
CN104166411A (zh) * 2014-07-21 2014-11-26 苏州昊枫环保科技有限公司 多室级联控制的感应对比监控系统
CN104268963A (zh) * 2014-08-06 2015-01-07 成都百威讯科技有限责任公司 一种智能门锁系统及智能门锁、智能报警门
KR20170045340A (ko) * 2014-09-02 2017-04-26 루돌프 킹 도어 및 홈 보안 시스템 및 방법
CN104269005A (zh) * 2014-09-20 2015-01-07 常州市武进南夏墅静电喷塑有限公司 智能家庭防盗系统
CN104240354A (zh) * 2014-09-26 2014-12-24 南昌大学 一种智能门户锁系统
CN104331684A (zh) * 2014-10-20 2015-02-04 西北工业大学 基于双机通信的高光谱图像目标检测系统
CN104463174A (zh) * 2014-12-16 2015-03-25 广州南方电力集团科技发展有限公司 一种多角度动态人物识别及行为预测系统
CN105788119B (zh) * 2014-12-23 2018-02-02 张宇成 一种远程智能报警方法、报警器和系统
CN104574727A (zh) * 2014-12-25 2015-04-29 安徽泓光网络工程有限公司 一种家庭智能防盗系统
CN104537811B (zh) * 2014-12-31 2017-06-23 上海青橙实业有限公司 一种智能报警方法及系统
CN105989683A (zh) * 2015-02-12 2016-10-05 贝斯科技国际有限公司 增强型住宅安全系统
CN104766395A (zh) * 2015-03-11 2015-07-08 张瞳 一种智能锁的安全控制方法
CN104766428A (zh) * 2015-03-11 2015-07-08 苏州中德宏泰电子科技有限公司 一种家庭安防监控系统
CN104852977B (zh) * 2015-04-30 2018-09-21 海尔优家智能科技（北京）有限公司 一种动态告警推送方法、系统和通信设备
CN106297120A (zh) * 2015-06-03 2017-01-04 天津维斯顿科技有限公司 一种智能型人体红外感测联动报警系统
CN104978810A (zh) * 2015-06-23 2015-10-14 浙江生辉照明有限公司 玻璃破碎检测装置、方法及led照明装置
CN104966369A (zh) * 2015-06-26 2015-10-07 苏州昊枫环保科技有限公司 基于霍尔传感检测与实时信息传输的监控系统
CN105139569A (zh) * 2015-08-20 2015-12-09 冯鹏程 一种智能门控设备与智能灯控设备的联动控制系统
CN105155940A (zh) * 2015-08-27 2015-12-16 成都众山科技有限公司 智能门锁系统
CN105178721A (zh) * 2015-08-27 2015-12-23 成都众山科技有限公司 节能安全门锁控制装置
CN106530445B (zh) * 2015-09-09 2019-11-29 伟林电子股份有限公司 远端门禁管控系统
CN105185066B (zh) * 2015-10-15 2017-08-25 上海斐讯数据通信技术有限公司 一种家庭智能云防盗系统及其使用方法
CN105303738B (zh) * 2015-10-22 2018-01-23 中国石油天然气股份有限公司 油井防盗油报警器
CN105279872A (zh) * 2015-10-30 2016-01-27 沈阳智汇万通科技发展有限公司 用于入户门的防盗监控装置及监控系统
CN105336076A (zh) * 2015-11-02 2016-02-17 苏州美达瑞电子有限公司 基于门磁感应和图像分析的智能监控系统
CN105336087A (zh) * 2015-12-01 2016-02-17 嘉兴职业技术学院 一种基于信息技术的安全监控系统
CN105653037A (zh) * 2015-12-31 2016-06-08 张小花 一种基于行为分析的交互系统及方法
CN105551211A (zh) * 2016-02-02 2016-05-04 苏州中恒普瑞能源互联网科技有限公司 一种基于wifi入侵报警器及入侵报警系统
CN105809878A (zh) * 2016-04-25 2016-07-27 成都五方传奇物联网技术有限公司 带无线通信功能的环境温度及振动检测电路
CN105805894A (zh) * 2016-04-25 2016-07-27 成都五方传奇物联网技术有限公司 一种基于wifi联网设计的具有环境温度检测功能的报警器
CN105743931B (zh) * 2016-05-11 2019-10-29 百度在线网络技术（北京）有限公司 应用于智能家居平台的控制方法和装置
CN105957284A (zh) * 2016-06-21 2016-09-21 太仓市美斯门窗有限公司 一种智能防盗报警门窗
CN106101631A (zh) * 2016-07-04 2016-11-09 安霸公司 云摄像机
CN106204837A (zh) * 2016-07-06 2016-12-07 苏州华安普电力科技股份有限公司 一种无人值守变电所安全门控系统
CN106056820A (zh) * 2016-07-15 2016-10-26 珠海格力电器股份有限公司 家庭安保装置及其控制方法、家庭安保系统、空调器
CN106127985A (zh) * 2016-08-16 2016-11-16 汪景安 多功能感应监控报警系统
CN106127986A (zh) * 2016-08-29 2016-11-16 龙元 一种app控制的窗户防盗远程及本地报警系统
CN106251532A (zh) * 2016-10-12 2016-12-21 无锡派乐科技有限公司 一种带拍照拍摄传输的室内安全监控系统
CN106447869A (zh) * 2016-10-26 2017-02-22 江苏金米智能科技有限责任公司 基于亮度调节的楼宇门禁自动化控制装置
FR3061392B1 (fr) * 2016-12-27 2019-08-30 Somfy Sas Procede de controle de presence et systeme de surveillance
CN106982325A (zh) * 2017-03-23 2017-07-25 移康智能科技（上海）股份有限公司 一种录像报警方法与装置
CN106934971A (zh) * 2017-03-30 2017-07-07 安徽森度科技有限公司 一种电网异常入侵预警方法
CN107067524A (zh) * 2017-06-03 2017-08-18 合肥佳洋电子科技有限公司 一种智能小区安防系统
CN107168294B (zh) * 2017-06-26 2024-01-19 西安热工研究院有限公司 一种火电水系统设备无人巡检监控方法
CN107204106B (zh) * 2017-07-11 2020-02-21 河北金锁安防工程股份有限公司 一种安防系统中滤除虚假警情的报警方法及系统
CN107564214A (zh) * 2017-07-19 2018-01-09 天盯科技（深圳）有限公司 门窗防盗方法、装置以及系统
CN107393073A (zh) * 2017-07-20 2017-11-24 克立司帝控制系统（上海）有限公司 基于图像识别技术的门禁监控控制系统及方法
CN107481463A (zh) * 2017-08-31 2017-12-15 亚月（武汉）智能科技有限公司 一种具有识别功能的婴儿车
CN108573584A (zh) * 2018-04-23 2018-09-25 深圳市沃特沃德股份有限公司 智能音响和基于智能音响的监控系统及方法
CN108447208B (zh) * 2018-05-17 2019-10-29 普联技术有限公司 灯光报警方法、设备、网络摄像机及存储介质
CN108564746A (zh) * 2018-05-21 2018-09-21 王连队 一种基于智能家居的门禁报警设备
CN108961634A (zh) * 2018-05-28 2018-12-07 马鞍山光标网络科技有限公司 一种智能城市用小区安防系统
CN108696726A (zh) * 2018-07-02 2018-10-23 天津煋鸟科技有限公司 一种无盲区监控系统
CN108765832B (zh) * 2018-07-06 2020-06-19 厦门信通慧安科技有限公司 一种入侵检测方法及装置
CN109118688A (zh) * 2018-08-24 2019-01-01 郑素娟 防盗栅栏安全防护系统
CN109087471A (zh) * 2018-09-25 2018-12-25 广州小全银电子科技有限公司 一种智能红外报警安防短信装置
CN109785474A (zh) * 2018-11-13 2019-05-21 北京码牛科技有限公司 一种安全、节能的多功能智能锁系统
CN109649335B (zh) * 2018-12-19 2021-08-27 台州科技职业学院 一种汽车玻璃窗打破提醒方法
CN109920182B (zh) * 2018-12-29 2022-03-04 国网北京市电力公司 防护处理方法、装置、存储介质及电子装置
CN110107188A (zh) * 2019-06-12 2019-08-09 广东工业大学 一种智能窗系统
CN110751771A (zh) * 2019-09-27 2020-02-04 深圳市大拿科技有限公司 门体异常告警方法及相关产品
CN111103807A (zh) * 2019-12-17 2020-05-05 青岛海信智慧家居系统股份有限公司 一种家用终端设备的控制方法及装置
CN111405390B (zh) * 2020-03-02 2022-10-18 浙江依圆医疗科技有限公司 一种检测门、窗开关状态的无线无源发射装置
CN111369469B (zh) * 2020-03-10 2024-01-12 北京爱笔科技有限公司 图像处理方法、装置及电子设备
CN111553386B (zh) * 2020-04-07 2022-05-20 哈尔滨工程大学 一种基于AdaBoost和CNN的入侵检测方法
CN114078301A (zh) * 2020-08-17 2022-02-22 漳州立达信光电子科技有限公司 一种安防系统、安全防范方法、装置及电子设备
CN112183337A (zh) * 2020-09-28 2021-01-05 华北电力大学（保定） 一种输电线路防机械入侵的检测方法及装置
CN112598862A (zh) * 2020-12-14 2021-04-02 合肥诺必达信息技术有限责任公司 一种智能家居用安防监控预警系统
CN112274143B (zh) * 2020-12-29 2021-06-01 四川写正智能科技有限公司 一种人体存在状态的检测方法及检测装置
CN113362546B (zh) * 2021-06-23 2023-03-17 深圳市凯莱特科技股份有限公司 一种住宅用安全防御系统及防御方法
CN113627570B (zh) * 2021-10-12 2021-12-24 北京创米智汇物联科技有限公司 一种区域入侵检测方法、装置、电子设备及存储介质
CN114697370A (zh) * 2022-03-29 2022-07-01 安徽景诺铝业有限公司 一种防盗用智能门窗系统
CN114758305A (zh) * 2022-06-15 2022-07-15 成都西物信安智能系统有限公司 一种构建入侵预警监控数据库的方法
CN118130074A (zh) * 2024-05-10 2024-06-04 四川名人居门窗有限公司 基于振动监测的门窗寿命预测及报警系统和方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN200990672Y (zh) * 2006-12-26 2007-12-12 黄德胜 远距监控装置
WO2008019467A1 (en) * 2006-08-16 2008-02-21 Tyco Safety Products Canada Ltd. Intruder detection using video and infrared data
JP2010176177A (ja) * 2009-01-27 2010-08-12 Panasonic Electric Works Co Ltd 負荷制御システム
CN101933058A (zh) * 2008-01-31 2010-12-29 传感电子有限责任公司 具有目标和事件分类的视频传感器和警报系统以及方法
CN201965714U (zh) * 2010-12-29 2011-09-07 羊恺 基于人脸识别的家用智能预警安防系统
CN202261630U (zh) * 2011-09-16 2012-05-30 沈易宽 无线可移动红外线侦测视频监控装置
CN103198595A (zh) * 2013-03-11 2013-07-10 成都百威讯科技有限责任公司 智能门窗防入侵系统

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101587630A (zh) * 2008-05-21 2009-11-25 上海新联纬讯科技发展有限公司 可过滤误报警的智能视频监控系统
CN101848371A (zh) * 2010-03-17 2010-09-29 上海大亚科技有限公司 基于3g技术的智能视频、音频无线固话安防终端系统

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008019467A1 (en) * 2006-08-16 2008-02-21 Tyco Safety Products Canada Ltd. Intruder detection using video and infrared data
CN200990672Y (zh) * 2006-12-26 2007-12-12 黄德胜 远距监控装置
CN101933058A (zh) * 2008-01-31 2010-12-29 传感电子有限责任公司 具有目标和事件分类的视频传感器和警报系统以及方法
JP2010176177A (ja) * 2009-01-27 2010-08-12 Panasonic Electric Works Co Ltd 負荷制御システム
CN201965714U (zh) * 2010-12-29 2011-09-07 羊恺 基于人脸识别的家用智能预警安防系统
CN202261630U (zh) * 2011-09-16 2012-05-30 沈易宽 无线可移动红外线侦测视频监控装置
CN103198595A (zh) * 2013-03-11 2013-07-10 成都百威讯科技有限责任公司 智能门窗防入侵系统

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105578134A (zh) * 2015-12-16 2016-05-11 武汉烽火众智数字技术有限责任公司 一种视频采集系统及方法
CN106205378A (zh) * 2016-08-26 2016-12-07 吕跟来 智能多功能电子门牌系统
CN108062132A (zh) * 2017-12-25 2018-05-22 广东星鼎通信科技有限公司 一种基于自动化控制技术的通信控制室
US11735018B2 (en) 2018-03-11 2023-08-22 Intellivision Technologies Corp. Security system with face recognition
CN109101370A (zh) * 2018-08-30 2018-12-28 重庆富民银行股份有限公司 主从双向同步的数据存储方法
CN109271280A (zh) * 2018-08-30 2019-01-25 重庆富民银行股份有限公司 存储故障快速切换处理方法

Also Published As

Publication number Publication date
CN103839346B (zh) 2016-08-24
CN103198595A (zh) 2013-07-10
CN103839346A (zh) 2014-06-04

Similar Documents

Publication Publication Date Title
WO2014139415A1 (zh) 一种智能门窗防入侵装置以及系统、智能门禁系统
CN103839373B (zh) 一种突发异常事件智能识别报警装置及报警系统
US11184583B2 (en) Audio/video device with viewer
US20200066122A1 (en) Dynamic Assessment Using an Audio/Video Recording and Communication Device
US10593174B1 (en) Automatic setup mode after disconnect from a network
WO2015172445A1 (zh) 一种家用多功能智能机器人
US11252378B1 (en) Batteryless doorbell with rectified power delivery
US11647165B1 (en) Audio/video recording and communication doorbell devices including transistor assemblies, and associated systems and methods
US11217076B1 (en) Camera tampering detection based on audio and video
US11240474B1 (en) Reporting connectivity problems for electronic devices
US10839204B2 (en) Sharing identification data with audio/video recording and communication devices and local processing of the shared data
CN104268963A (zh) 一种智能门锁系统及智能门锁、智能报警门
US10803719B1 (en) Batteryless doorbell with energy harvesters
CN105427421A (zh) 基于人脸识别的门禁控制方法
US11064167B2 (en) Input functionality for audio/video recording and communication doorbells
US11341825B1 (en) Implementing deterrent protocols in response to detected security events
US10713928B1 (en) Arming security systems based on communications among a network of security systems
US11501618B1 (en) Security device with user-configurable motion detection settings
US10943442B1 (en) Customized notifications based on device characteristics
US11349707B1 (en) Implementing security system devices as network nodes
US10733857B1 (en) Automatic alteration of the storage duration of a video
US11164435B1 (en) Audio/video recording and communication doorbell devices with supercapacitors
US11032128B2 (en) Using a local hub device as a substitute for an unavailable backend device
US11412189B1 (en) Batteryless doorbell with multi-load power delivery
US11032762B1 (en) Saving power by spoofing a device

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14765805

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14765805

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC