AU2001248158A1 - An object monitoring system - Google Patents
An object monitoring systemInfo
- Publication number
- AU2001248158A1 AU2001248158A1 AU2001248158A AU4815801A AU2001248158A1 AU 2001248158 A1 AU2001248158 A1 AU 2001248158A1 AU 2001248158 A AU2001248158 A AU 2001248158A AU 4815801 A AU4815801 A AU 4815801A AU 2001248158 A1 AU2001248158 A1 AU 2001248158A1
- Authority
- AU
- Australia
- Prior art keywords
- monitoring system
- object monitoring
- sensors
- zone
- agents
- 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.)
- Abandoned
Links
Landscapes
- Burglar Alarm Systems (AREA)
Description
AN OBJECT MONITORING SYSTEM
The present invention relates to an object monitoring system and, in particular, to a security system and processes executed by the system to protect objects, and software for executing the processes. More specifically, the invention is directed to a distributed system architecture for monitoring a number of areas.
Security systems using closed circuit television (CCTV) or video cameras have been developed in various forms to monitor events in an area. When an event occurs which is considered to breach security in the monitored area an alarm condition is entered and an alarm signal can be generated. The level of sophistication of the monitoring based on the signals produced by the cameras varies considerably. For example, the video images produced by the cameras may simply be presented to display screens which are monitored by security personnel who will determine when an event gives rise to an alarm condition. Other systems may store the video frames generated by the cameras for subsequent processing and comparison to determine whether an event has occurred. To achieve this type of processing different electrical and electronic systems have been developed in the past to compare characteristics of adjacent video frames, which may involve image subtraction and differencing to determine whether differences have occurred in chronologically adjacent frames. These systems include dedicated frame processing circuitry and are limited in the forms of monitoring which can be executed by the system. For instance, frames of individual cameras are processed but there is no reference between frames of different cameras. In addition to being inflexible, the systems are subject to the detection of false events or alarms due to intensity changes or shadowing which may occur in a monitored area or even due to normal and allowed movement within an area.
It is desired to provide a system and architecture which alleviates the above difficulties or at least provides a useful alternative.
The present invention provides an object monitoring system, including: a slave system having slave processors and a plurality of cameras for monitoring one or more objects, images of said cameras being processed by respective ones of said slave processors, and said slave processors including sensors assigned to respective
cameras and at least parts of the fields of view of said cameras to detect predetermined characteristics of said at least parts and generate a trigger in response thereto; and a master system having a master processor including agents assigned to at least one of said objects and having one or more of said sensors allocated thereto, said agents each generating an event on receiving said trigger from the one or more allocated sensors.
The present invention also provides an object monitoring system, including: a plurality of sensors assigned to respective zones in a monitored area and which process pixels representing a zone to detect predetermined characteristics in said zone and generate a trigger in response thereto; and a plurality of agents respectively assigned to at least one object and having one or more of the sensors allocated thereto, said agents generating respective events on receiving triggers from the one or more allocated sensors.
Preferably, the sensors and agents are software objects and the triggers and events are messages generated by the objects.
The present invention also provides a method of monitoring an object, including: monitoring a zone in an area using a sensor which processes pixels representing said zone; detecting predetermined characteristics in said zone using said sensor and generating a trigger in response thereto; receiving said trigger at an agent assigned to said object and to which said sensor is assigned; and generating an event using said agent in response to receiving triggers from sensors assigned to the agent.
The present invention also provides computer software for executing the above steps.
Preferred embodiments of the present invention are hereinafter described, by way of example only, with reference to the accompanying drawings, wherein:
Figure 1 is a block diagram of a preferred embodiment of a security system;
Figure 2 is a block diagram of system components of a node of the security system;
Figure 3 is a block diagram of the software components of and associated with the nodes;
Figure 4 is a block diagram of an implementation of the software objects of the system;
Figure 5 is a floor plan of an area monitored by the security system;
Figure 6 is a diagram of an object monitored by the security system;
Figure 7 is a block diagram of a control platform connected to the security system;
Figures 8, 9, 10 and 11 are examples of screen displays generated by user interfaces of the platform and the system;
Figures 12 to 15 are entity relationship diagrams for software classes of the system; and
Figure 17 and 18 are process flow diagrams for the classes.
A security system 1, as shown in Figure 1, includes a number of nodes 2 which are connected by a node network 4 to a system server 6 and a graphic user interface server 8. The number of nodes 2 used in the system depends entirely upon the size of the area or environment to be secured and the number of objects in the area to be monitored and protected. The network 4 may be a TCP/IP network and the servers 6 and 8 standard personal computers (PCs).
The nodes 2, as shown in Figures 2 and 3, handle video feeds or signals generated by a plurality of video cameras 10. The video frames generated by the cameras are fed by a standard video switch 12 to slave processors 14. The slave processors 14 may each be PCs, and each have a number of dedicated software objects stored thereon. Each video camera 10 belongs to or is allocated to one slave processor 14. A tracking slave processor 16 is used to control and handle video frames generated by a pan tilt zoom (PTZ) camera 18. A node 2 also includes a master processor 20, which also may be a standard PC, that handles messages generated by the slave processors 14 and 16. All of the processors 14, 16, 20 are connected by a hub 22 of the network 4. In addition, the video signals from the cameras may also be fed to another video switch to be fed to a control room for manual monitoring. Messages generated by the master processor 20 may be sent by a paging transceiver 24 instead of or in addition to sending the messages on the network 4.
All of the processors 14, 16 and 20 and the servers 6 and 8 execute their own set of dedicated software objects and run only one instance of each object. The slave processors 14 and 16 of a node, as shown in Figures 3 and 4, host a set of sensors 30 and other objects for processing the received video frames, such as a frame grabber object 32. A sensor 30 is a software object which is responsible for detecting predetermined characteristics, such as changes, in a received video frame. The sensors 30 each analyse pixels of different areas or parts of the frames received, and the parts may overlap. For example, if a video camera is static then the field of view is fixed and the sensors 30 for the camera monitor respective parts of that field of view. A sensor 30 may also process all pixels of a frame. When a sensor 30 detects a predetermined characteristic or change then it generates and sends a trigger message. The sensors 30 may also be of different types, such as video motion sensors, tracking sensors and hardware input sensors, as described below.
The master processor 20 of each node 2 hosts a set of agents 34. The agents 34 are software objects which are each able to register interest in or be allocated to one or more respective sensors of the slave processors 14 and 16. An agent is able to generate and send event messages to the event server 6 when it determines an event has occurred on the basis of the trigger messages it receives from its registered sensors 30. The nature of an event depends on the environment or area being monitored but may include when a person enters a protected area or when a secure door is breached. For example, as shown, in Figure 4, agent 1 of the master processor 20 may only generate an event message when it receives a trigger message from sensor 2 of the second slave processor and sensor N of the third slave processor. Agent 3, on the other hand, generates an event message when it receives trigger messages from sensor 1 of the first slave processor, sensor 3 of the second slave processor and sensor 2 of the Nth slave processor. An agent may generate an event message in response to one trigger signal from one registered sensor or require a number of trigger messages to be generated from a single sensor or a number of sensors. The event messages are sent directly to the event server 6, but if the server 6 is unreachable for some reason then the images are placed on a persistent queue for subsequent retries.
The event server 6 receives and stores all event messages. It may forward all event messages to the user interface server 8 for display. Alternatively it may forward selected
event messages. The event server 6 is also able to determine when an alarm condition occurs. It may be configured so as to consider that an alarm condition occurs when any event message is sent from one of the nodes 2 or will require a particular combination of event messages to send a message to the user interface 8 indicating than an alarm condition has occurred. The particular combination of event messages may also need to be received within predetermined time constraints. Once an alarm condition has occurred, the event server 6 or the user interface server 8 can be used to send and generate appropriate alarm signals.
The user interface server 8 generates all displays for a graphic user interface for operators and allows an authorised operator to configure the security system, which includes the creation and establishment of required agents 34 and sensors 20 for the system. Configuration also involves defining the agents and servers and their relationship, and establishing the manner in which event messages are handled and how they are acted upon.
The security system 1 can monitor a number of areas within an environment simultaneously, and one of the areas may be a room 40, as shown in Figure 5. The room 40 may include a number of objects 42 that need to be monitored and protected. The cameras 10 of a node 2 can be placed in the room so as to provide effective monitoring of the objects 42. One or more cameras may be allocated to each object 42. The architecture of the system is particularly advantageous as it allows a sensor 30 to be assigned to a particular region within the field of view of the camera 10 to detect any changes or movement within that field of view. The sensors 30 for any camera are established using the user interface server 8 by drawing polygons around the boundary of the region in the field of view of a camera 10 to define a monitored zone which the sensor operates on. The sensor 30 will then only process pixels in that zone or region for the video frames received by its slave processor. For example, the video motion detection (VMD) sensors will only consider breaches or changes within that monitored zone. When a breach is detected in the monitored zone then the sensor generates a trigger. For a painting 50, as shown in Figure 6, a polygon can be drawn around the edge of the painting so as to create a sensor 30 for the boundaries of that painting. By using a number of different camera angles provided by different cameras 10 placed on the same object, sensors 30 can be created around the
object from different fields of view so as to eliminate nuisance alarms. For instance, in the situation illustrated in Figure 6, if only one camera was used to protect the object 50 then an alarm may sound when a person 52 interrupts one camera view of the object 50. This can be avoided by using at least two cameras 10 and two sensors 30 operating on different fields of view. Depending on the object 42 and the field of view it may also be appropriate to place a camera 10 vertically mounted above the object.
The VMD sensors each execute a detection process on the pixels of their monitored zone which involves comparing the pixels obtained for a video frame with a statistical model of normality for those pixels. When the security system is installed and the sensors have been defined, the sensors first establish the statistical model of normality by processing a number of pixels over successive video frames for the monitored zone when it is known that there are no breaches of the zone. The VMD sensors operate on the hue, saturation and brightness levels of the pixels with reference to the hue, saturation and intensity, colour space for video pixels. Operation on hue saturation, and brightness variations in the colour space with reference to the normality model gives rise to accurate generation of trigger messages when a sensor boundary is breached by another object other than the object being protected.
The tracking sensors 30 are configured to recognise and track regions in a zone that are recognised as belonging to a human. In stable illumination conditions, the colour of human skin forms a relative compact distribution in the hue, saturation, and intensity space of observable colours, and this is found to be irrespective of race. Although skin colour is inconsistent under drastic changes in illumination, once a face, for example, is detected and verified, its apparent colour can be robustly and consistently tracked by updating a scene specific, adaptive colour model for the tracking sensor. It has been determined that the colour model can be based on a skin pigment that falls within a predetermined hue and saturation range. The tracking sensors are accordingly set to track a cluster of pixels within this range.
The tracking sensors can be implemented based on a principle of integrated active multi-view face recognition. The security system 1 is unique in its ability to cope with a dynamic environment when human subjects are mostly in constant motion at varying
distances away from the cameras. The system takes an integrated approach to real-time face detection, tracking and recognition on the basis of the following:
(i) Face detection and tracking is executed automatically on any arbitrary view and varying distance in the given monitored zone of the sensor without any need for manual intervention. Faces appearing in the zone are automatically detected by the sensor and tracked in a distance away from the camera of the sensor with a very small target size, down to 5x5 pixels of a frame. The face tracking process executed by the sensor provides estimates of face position or displacement, scale and also a pose on a face that is sufficiently close to the camera.
(ii) The tracking sensor computes a 3D pose of a tracked face in real-time, and is able to generate control messages for a PTZ camera software object 35 of the tracking slave processor 16 to actively pan and tilt the PTZ camera 18 to lock on a moving face and sample images over time for recognition. The security system utilises the PTZ camera 18 as an active camera with closed loop feedback control from the tracking sensors 30 directly to the camera's pan, tilt and zone functions controlled by the camera object 35.
(iii) Recognition can be performed that models the view across a wide range of views of different cameras 10 using a number of tracking sensors rather than at a single restricted frontal view. Face identities can then be modelled through learning and 3D information about a face inferred implicitly, such as head pose and the correspondence between facial features observed from a profile and profile to profile views.
The security system 1 can be connected by a network connection 70, which may be
Ethernet or TCP/IP, to a control platform 72 which is able to receive data signals from a variety of sources 74, in addition to messages sent by and to the security system 1 over the connection 70. The platform 72 may be PC-based or comprise a set top box and have associated with it an infrared keyboard 74 or other peripheral devices, such as a telephone. The platform also has a display screen 76 for displaying user interfaces, as shown in Figures 8, 9, 10 and 11. The provision of the display screen 76 may remove the requirement for the security system 1 to have one or more display screens. For example, Figure 10 is an interface of the server 8 used for configuration of the system 1 and Figure
11 is an interface of the server 8 generated in response to events. The display screen 76 may be unique for the platform 72 or the user interface simply presented on existing display screens in a secured environment, such as TV screens. The various sources 74 may comprise other network services, such as ISDN, DSL or dial up Internet services, or TV services, such as cable and satellite pay TV digital feeds. The control platform allows control and direction of the various digital feeds from the sources 74 and information that can be obtained from the security system 1 and its cameras 10. All control is facilitated via the user interface displayed in Figures 8 and 9. This provides a number of controls to access selected TV services, Internet services and security services. Any of the video feeds from the cameras 10 or the pay TV services can be displayed on the screen and selected. For example, as shown in Figure 8, parties can be viewed at monitored areas in the environment and devices, such as doors, controlled using the interface.
A logical view report is provided on pages 10 to 19 which provides further details on the software objects and the messaging architecture used by the security system 1. Entity relationship diagrams for the objects are provided in Figures 12 to 16. Process examples are shown in Figure 17 and 18.
In the example of Figure 17 the signals from two cameras have been combined to form a single video signal. Objects have been instantiated to represent the two cameras, as well as the combined video signal. A single sensor object has been associated with each camera. Each sensor object utilises a video motion sensing algorithm object to detect motion. Finally, both sensor objects have been associated with the same agent object.
Video for Windows utilises its own thread for callbacks. When the frame grabber is called into by Video for Windows (not shown), the following sequence of events takes place on that thread. The grabber calls into callback, which in turn, calls back into the video signal.
The video signal calls newFrame() on the first of its cameras, which in turn, calls into its sensor. The sensor delegates its processing to its VMO algorithm which, in this example, returns true to indicate a trigger. As a result of this, the sensor sends a trigger message to its agent. Currently, only one of the agent's sensors has triggered, so it takes no action.
After the call into the first camera has returned, the video signal calls into the second camera. The net result of this is that the agent is sent another trigger message.
In the example of Figure 18 there are two slaves, a master and a server process. Each slave has a frame grabber card processing the video signal from a single camera. Software objects have been instantiated to represent the grabber and camera in each slave. They are linked via callback objects. Each camera has been associated with one sensor object. Each of the sensor objects utilises a video motion sensing algorithm object. Both sensor objects have been associated with the same agent object, which is in the master. Video for Windows utilises its own thread for callbacks. The processing is initiated by these callbacks. The system does not need to synchronise Video for Windows processing on the different slaves. In the above scenario, grabber 1 is slightly ahead of grabber 2 in time. The following is executed. Grabber 1 calls into callback 1, which in turn calls into camera 1. Camera 1 calls into sensor 1. Sensor 1 delegates its processing to alg 1 by calling newFrame(). At about this time, grabber 2 calls into callback 2. This causes a call sequence which results in newFrame () being called on alg 2. Some time later, the newFrame () call on alg 1 returns true. As a result of this, sensor 1 sends a trigger message to agent 1. Currently, only one of the agent's sensors has triggered, so it takes no action. Some time later yet again, the newFrame 0 call on alg 2 returns true. As a result of this, sensor 2 also sends a trigger message to agent 1. This time however, both of agent l's sensors are in a triggered state, so it creates an event object and fires it off to the event server object, which is in the server process.
Many modifications will be apparent to those skilled in the art without departing from the scope of the present invention as herein described with reference to the accompanying drawings.
LOGICAL VIEW REPORT
TABLE OF CONTENTS
TABLE OF CONTENTS 10 LOGICAL VIEW REPORT 11
LOGICAL VIEW 11
MASTER 11
ZORBA 12
SHMessageServer 12
SHMessageTarget 12
SHProxy 13
SHConnection 13 IMAGING 13
SHImageBase 13
SHImage 14
SHFrameSequence 14
SHVFWFrameGrabber 15 SHRGBColour 15
SHHLSColour 15
SHHSVColour 15
SHNewFrameCallback 75
SHNewFrameCallbackM 75 SHRect 16
SHSize 16
SHVMSemorAlg 16
SHPoint 16
SLAVE 16 SHVideoSignal 16
SHCamera. 17
SHVMSensor 17
SHSlave 17
TOTALS: 19 LOGICAL PACKAGE STRUCTURE 19
LOGICAL VIEW REPORT
\fY%
These diagrams summarise some of the main classes involved in video motion detection. Other classes are involved besides the ones specified. Furthermore, only some of the more 'interesting' class methods have been specified.
The Master library defines some of our higher level business objects. It is closely related to the Slave library. The two are designed to work in conjimction with each other. Along with the
Slave library, the Master library defines some of our main business objects, the most notable of which is the agent. SHAgent
An agent utilises sensors for the purpose of protecting or monitoring some kind of item. In its simplest form, an agent responds to remote trigger messages sent to it by its sensors (ie sensors remotely calling sensorTriggerO). If it detects that all of its sensors are concurrently triggered, it will create an event object (not shown) and send it to the event server (also not shown). Derived from SHMessageTarget
Pπvate Attributes.
RWCStrmg name
Public Operations: void _useSensor (SHProxy sensor) void _discardSensor (SHProxy sensor) size_t_numSensors 0 SHProxy _sensor (size_t i) void _sensorTrigger (SHProxy sensor, bool tπggerState) void arm 0 void disarm 0
SHMaster
This class represents the root object within the Master subsystem. When a master connects to a Slave, it obtains proxies to each of the Slave's cameras. These proxies are then cached by the Master. When the User Interface, for example, connects to a Master, it can directly access the cameras cached by the Master. Thus there is no need for the User Interface to have direct knowledge of Slaves.
Derived fiom SHMessageTarget
Private Attributes-
RWCStrmg name
Public Operations:
SHProxy _appendAgent 0 void_renameAgent (SHProxy agent) void _de!eteAgent (SHProxy agent) void _armAgents 0 void _disarmAgents 0 size_t_numCameras 0 SHProxy _camera ( size_t i)
Zorba implements our remote, inter-process messaging. SHMessageServer
This class is responsible for listening for connection requests made by other processes and for establishing connections to these processes in order to process messages that they send. When a connection to another process is established, a new message loop, operating on a seperate thread, is created. The message loop uses an SHConnection to continually read in the next message and dispatch it to the appropriate SHMessageTarget. Public Operations: void start 0 void stop 0 void join 0
Private Operations: void listenForConnections 0 void run essageLoop 0
SHMessageTarget
This class provides the ability to receive messages originating from other processes. A system of message maps is used to dispatch an incoming remote message to the correct handler method. Classes need to derive from this class and set up their message maps, in order to utilise this functionality. Inspired by MFC message maps! Public Operations: void processMessage (SHConnection con)
Processes message which is expected to be waiting for retreival from the specified SHConnection instance. The message is dispatched to the appropriate handler method. RWCString targetld 0 Returns a string which uniquely defines the instance of the message target within the process in which it resides. This string may be utilised by SHProxy instances in other processes.
SHProxy
This class represents an indirect link to an SHMessageTarget instance residing within another process. It allows messeges to be sent to that object. An SHProxy caches a counted reference to an SHConnection which it utilises to send messages to the remote SHMessageTarget. The SHProxy also caches the message target's unique target id which is used by the underlying messaging protocol.
Private Attributes:
SHConnection conn_ RWCString targetld_
Public Operations: void sendMessage ()
SHConnection
Basic wrapper around an RWSocketPortal. Represents a connection to another process and a means of sending and receiving data to and from that process. Implements templated streaming operators for sending and receiving data.
This class is provides reference counting services to other classes.
Private Attributes:
RWSocketPortal portaI_ size t refs
Public Operations: return addReference 0 size t removcReference 0 size_t references 0 void operator« 0
Templated streaming operator for data transmission. void operator» 0 Templated streaming operator for data reception.
Imaging is a library of image processing classes including points, rects, colours, images and video motion sensing algorithms. SHImageBase Abstract base class for an image. Derived classes need to override and implement pixelPtrAtQ. Offers various image processmg support methods, such as the ones specified.
Private Attributes: size_t idth_ size_t height_
Public Operations:
SHRGBColour colour At ( size_t x, size_t y) size_t hueAt ( size_t x, size_t y) size_t greyscaleAt ( size_t x, size_t y)
Protected Operations: const unsigned char* pixelPtr t ( size_t x, size_t y)
SHImage
Concrete image class implementation. Overrides and implements pixelPtrAt(). This class is platform specific. The current implementation is for WIN32 based compilers only. Derived from SHlmageBase
Private Attributes:
BITMAPENFO* bitmapInfoPtr_ unsigned char* pixelBuffcr_
Public Operations: bool draw ( HDC dc, const RECT& rect)
Protected Operations: const unsigned char* pixelPtrAt ( size_t x, size_t y)
SH rameSequence An abstract base class for something which is a changing sequence of image frames.
Provides the facililty for users to register a new-frame-callback which is called every time a derived concrete implementation calls newFrameO- This class is here in order to easily be able to implement other non-frame grabber types of frame sequence, such as MPEG or AVI frame sequences. Derived from SHImage
Public Operations: void regNewFrameCallback ( SHNewFrameCallback* callback)
Protected Operations: void newFrame ( unsigned char* framcBufferPtr) bool doStartO bool doStop 0
SHVFVVFrameGrabber
Concrete implementation of a frame sequence.
This particular class knows about Microsoft Video for Windows frame grabbing and hence is platform dependent. „
NOTE: Video for Windows frame grabbing owns its own thread of execution. It utilises a C-style callback to notify the outside world that a new frame is available for processing. We're stuck with doing it way. This class hides that paradigm and implements the more 00 paradigm of a callback object (via inheritance from SHFrameSequence). See
SHNewFrameCallback for more details. Derived from SHFrameSequence
Protected Operations: bool doStart O bool doStop 0
SHRGBColour SHHLSCoIour
SHHSVColour SHNewFrameCallback
Abstract base class for an OO-style callback. Derived classes need to override and implement newFrame(). When a derivative of this class is registered with an SHFrameSequence derivative, newFrame() is called polymorphically by the frame sequence.
Public Operations: void newFrame ( unsigned char* frameBufferPtr)
SHNewFrameCallbackM
A templated class whose template type is that of the object to be called back into. This class stores an instance pointer to that object, as well as a method pointer to a method of the correct signature for receiving new frame notifications. In the implementation of the virtual method newFrameO, the method is called on the object instance.
Derived from SHNewFrameCallback
Private Attributes: T* instancc_ void (T::*)(unsigned char* buf) memberFunc
Public Operations: void newFrame ( unsigned char* frameBufferPtr)
SHRect
SHSize
SHVMSensorAIg
This class does the work of video motion sensing. It caches a reference to an SHImageBase and uses the appropriate image processing support methods to do its stuff. One iteration of sensing is performed each time newFrame is called. This class stores a collection of SHPoints that define the algorithm's area of interest. On each newFrame the class returns whether motion is detected or not and, if so, where on the boundary and where internally that motion occurred. Public Operations: void newFrame : size_t 0
SHPoint
The Slave library defines classes used to develop a slave-like application. It is closely coupled to the Master library. The Slave library classes such as video signals, cameras and sensors. We consider these to be some of our business objects. They are higher level than the classes belonging to the Imaging and Zorba libraries.
SHVideoSignal
The video signal's sole existance is to allow for multiple (e.g. quaded) cams. The SHVideoSignal is an aggregate of one or more SHCamera instances. The video signal also owns a concrete SHFrameSequence derivative to which it registers a new frame callback. When newFrame() is called, via the callback, the video signal calls newFrameO for each of its cams. Private Attributes:
SHNewFrameCallbackM<SHVideoSignal> newFrameCaIlback size_t numXCams_ size t numYCams
Public Opeiations: void newFrame ( unsigned char* buf)
SHCamera
The SHCamera stores enough information to determine the region of the frame buffer passed to it in newFrameO, that belongs to itself. The SHCamera is also a concrete
SHImageBase derivation and uses this information in its override of pixelPtrAtO, to reference the correct pixel.
Derived from SHImageBase, SHMessageTarget
Private Attributes: const unsigned char* frameBuf SHSize frameBufSize SHRect camRect RWCString name_
Public Operations: void newFrame ( unsigned char* buf) const unsigned char* pixelPtrAt ( size_t x, size_t y) void load 0 void save 0 SHProxy _appendSensor 0 void _deleteSensor 0 void _renameSensor 0
SHProxy _sensor (size_t i) void _clear 0 RWCString _name 0
SHSize _resoiution 0
SHVMSensor The video motion sensor is basically implemented in terms of an SHVMSensorAlg.
However, the SHVMSensor is a SHMessageTarget while the SHVMSensorAlg is not. The separation is artificial, but it does allow us to develop sensor algorithm test rigs which are independent of Zorba and Rogue Wave. Derived from SHMessageTarget
Private Attributes:
RWCString name_
Public Operations: void _registerAgent (SHProxy agentProxy) void _discardAgent (SHProxy agentProxy) size_t_numAgents 0 SHProxy _agent (size_t i)
SHSlave
The SHSlave represents the root object within the Slave subsystem. All other SHMessageTarget derived objects within the Slave subsystem are accessed via this object. Derived from SHMessageTarget
Private Attributes:
RWCString name_
Public Operations: size_t_numCameras 0 SHProxy _camera ( size_t i)
TOTALS:
6 Logical Packages 23 Classes
LOGICALPACKAGE STRUCTURE
Logical View
Master Zorba Imaging Slave
Claims
1. An object monitoring system, including: a slave system having slave processors and a plurality of cameras for monitoring one or more objects, images of said cameras being processed by respective ones of said slave processors, and said slave processors including sensors assigned to respective cameras and at least parts of the fields of view of said cameras to detect predetermined characteristics of said at least parts and generate a trigger in response thereto; and a master system having a master processor including agents assigned to at least one of said objects and having one or more of said sensors allocated thereto, said agents each generating an event on receiving said trigger from the one or more allocated sensors.
2. An object monitoring system as claimed in claim 1, wherein said at least parts represent monitored zones and said predetermined characteristics include changes in said zones.
3. An object monitoring system as claimed in claim 1, wherein said agents are assigned to protect respective objects.
4. An object monitoring system as claimed in claim 1, wherein said agents are allocated said sensors of one of said slave processors.
5. An object monitoring system as claimed in claim 1, wherein said agents are allocated said sensors of a number of said slave processors.
6. An object monitoring system as claimed in claim 1 , wherein said event is an alarm.
7. An object monitoring system as claimed in claim 1, including an event system for receiving events from said agents and determining whether the received events represent an alarm condition.
8. An object monitoring system as claimed in claim 2, including a user interface for selectively displaying said zones in response to said events.
9. An object monitoring system as claimed in claim 2, including a user interface for selecting said zones assigned to said sensors and allocating said sensors to said agents.
10. An object monitoring system as claimed in claim 1, including a user interface for generating alarm signals in response to said events.
11. An object monitoring system as claimed in claim 7, including a user interface for generating displays in response to determination of said alarm condition.
12. An object monitoring system as claimed in claim 2, wherein said sensors include motion detection sensors that compare pixel values of the zone with a statistical model of normality for pixels of the zone to detect movement in said zone.
13. An object monitoring system as claimed in claim 12, wherein said motion detection sensors operate on the hue, saturation and brightness levels of said zones.
14. An object monitoring system as claimed in claim 1, wherein said predetermined characteristics include a cluster of pixels having a predetermined hue and saturation range representing human skin.
15. An object monitoring system as claimed in claim 2, wherein said sensors include a tracking sensor to track a cluster of pixels representing at least part of a human.
16. An object monitoring system as claimed in claim 15, wherein said tracking sensor generates control messages to control a moveable camera.
17. An object monitoring system as claimed in claim 15, wherein said cluster of pixels has a predetermined hue and saturation range corresponding to human skin.
18. An object monitoring system as claimed in claim 15, wherein said tracking system tracks a human face.
19. An object monitoring system as claimed in claim 18, wherein said sensor generates a 3D pose of a tracked face.
20. An object monitoring system as claimed in claim 19, wherein said system is adapted to execute recognition on the basis of said pose.
21. An object monitoring system as claimed in claim 1, including a control platform connected to the network and a display device, said control platform being adapted to receive and control the direction of a plurality of digital data feeds, such as video signals from said cameras and video signals from different services, and having a user interface to control display of any of the data feeds on the display device.
22. An object monitoring system, including: a plurality of sensors assigned to respective zones in a monitored area and which process pixels representing a zone to detect predetermined characteristics in said zone and generate a trigger in response thereto; and a plurality of agents respectively assigned to at least one object and having one or more of the sensors allocated thereto, said agents generating respective events on receiving triggers from the one or more allocated sensors.
23. An object monitoring system as claimed in claim 22, wherein the sensors and agents are software objects and the triggers and events are messages . generated, by the objects.
24. An object monitoring system as claimed in claim 22, including an event processor for receiving the events and generating an alarm signal when said events represent an alarm condition.
25. An object monitoring system as claimed in claim 22, wherein said sensors include motion detection sensors that compare pixel values of the zone with a statistical model of normality for pixels of the zone to detect movement in said zone.
26. An object monitoring system as claimed in claim 25, wherein said motion detection sensors operate on the hue, saturation and brightness levels of said zones.
27. An object monitoring system as claimed in claim 22, wherein said predetermined characteristics include a cluster of pixels having a predetermined hue and saturation range representing human skin.
28. An object monitoring system as claimed in claim 22, wherein said sensors include a tracking sensor to track a cluster of pixels representing at least part of a human.
29. An object monitoring system as claimed in claim 28, wherein said tracking sensor generates control messages to control a moveable camera.
30. An object monitoring system as claimed in claim 28, wherein said cluster of pixels has a predetermined hue and saturation range corresponding to human skin.
31. An object monitoring system as claimed in claim 28, wherein said tracking system tracks a human face.
32. An object monitoring system as claimed in claim 31, wherein said sensor generates a 3D pose of a tracked face.
33. An object monitoring system as claimed in claim 32, wherein said system is adapted to execute recognition on the basis of said pose.
34. A method of monitoring an obj ect, including: monitoring a zone in an area using a sensor which processes pixels representing said zone; detecting predetermined characteristics in said zone using said sensor and generating a trigger in response thereto; receiving said trigger at an agent assigned to said object and to which said sensor is assigned; and generating an event using said agent in response to receiving triggers from sensors assigned to the agent.
35. A method as claimed in claim 34, including detecting said event and determining whether an alarm condition exists.
I .
36. Computer software stored on computer readable storage media having code for executing the steps of the method of claim 34 or 35.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AU2001248158A AU2001248158A1 (en) | 2000-04-11 | 2001-04-11 | An object monitoring system |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AUPQ6846 | 2000-04-11 | ||
AUPQ6846A AUPQ684600A0 (en) | 2000-04-11 | 2000-04-11 | An object monitoring system |
PCT/AU2001/000421 WO2001078397A1 (en) | 2000-04-11 | 2001-04-11 | An object monitoring system |
AU2001248158A AU2001248158A1 (en) | 2000-04-11 | 2001-04-11 | An object monitoring system |
Publications (1)
Publication Number | Publication Date |
---|---|
AU2001248158A1 true AU2001248158A1 (en) | 2001-10-23 |
Family
ID=25628170
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
AU2001248158A Abandoned AU2001248158A1 (en) | 2000-04-11 | 2001-04-11 | An object monitoring system |
Country Status (1)
Country | Link |
---|---|
AU (1) | AU2001248158A1 (en) |
-
2001
- 2001-04-11 AU AU2001248158A patent/AU2001248158A1/en not_active Abandoned
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1279289A1 (en) | An object monitoring system | |
EP2283472B1 (en) | A system and method for electronic surveillance | |
JP7036493B2 (en) | Monitoring method and equipment | |
US7733369B2 (en) | View handling in video surveillance systems | |
US8854469B2 (en) | Method and apparatus for tracking persons and locations using multiple cameras | |
US9363487B2 (en) | Scanning camera-based video surveillance system | |
Boult et al. | Omni-directional visual surveillance | |
US20170277785A1 (en) | Method and system for modeling image of interest to users | |
CN102158689A (en) | Video monitoring system and method | |
JP2021132267A (en) | Video surveillance system and video surveillance method | |
JP2004519956A (en) | Target selection method for automatic video tracking system | |
CN110557603B (en) | Method and device for monitoring moving target and readable storage medium | |
US9183716B2 (en) | System and method for managing moving surveillance cameras | |
WO2023273029A1 (en) | Video processing method and apparatus, device, storage medium, and program product | |
CN119744408A (en) | Object Detection via Region of Interest | |
JP2007221693A (en) | Video surveillance system and video concentrator | |
JP2020119494A (en) | Distributed image analysis system | |
CN113660455B (en) | Method, system and terminal for fall detection based on DVS data | |
JP2003109001A (en) | System and program for detecting person | |
CN110263619A (en) | Image processing method and device, electronic equipment and computer storage medium | |
AU2001248158A1 (en) | An object monitoring system | |
JP2023020307A (en) | Information processing device, information processing method and program | |
GB2439184A (en) | Obstacle detection in a surveillance system | |
CN114218570B (en) | Method and system for preventing computer information from being stolen and stolen based on AI and RPA technologies | |
Sun et al. | Recording the region of interest from FLYCAM panoramic video |