CN110443834A - A kind of distributed object tracking system - Google Patents

Abstract

The invention discloses a kind of distributed object tracking systems, include: multiple image analysing computer devices, wherein each image analysing computer device analysis its object in a received at least real-time imaging crossfire to generate the analysis result of at least one object that can be tracked；An and cluster management service unit, it is connected to multiple image analysing computer device, wherein the cluster management service unit includes a tracing module to concatenate track of the analysis result of at least one object being tracked caused by the multiple image analysing computer device to track at least one object.

Description

A Distributed Object Tracking System

technical field

The invention relates to an object tracking system, in particular to a distributed object tracking system.

Background technique

Most of the current image analysis devices are for the image analysis of a single scene of a single camera. The analysis results of each image analysis device are independent, and it is impossible to connect the analysis results of different cameras to achieve cross-camera object tracking. Therefore, the industry needs a distributed object tracking system that can connect the analysis results of different cameras to achieve cross-camera object tracking.

Contents of the invention

An object of the present invention is to provide a distributed object tracking system, which can connect the object analysis results of multiple real-time image streams of multiple image analysis devices in series to track the trajectory of an object in real time, wherein the image analysis device Resource points can be utilized for object characterization to reduce system cost.

An object of the present invention is to provide a distributed object tracking system, which can connect the object analysis results of multiple real-time image streams of multiple image analysis devices across regions to track the trajectory of an object across regions in real time. To provide post-event search and abnormal behavior alarm services.

An embodiment of the present invention proposes a distributed object tracking system, including: a plurality of image analysis devices, wherein each image analysis device is connected to at least one camera device to analyze at least one real-time image transmitted by the at least one camera device object in the image stream to generate an analysis result of at least one object being tracked; and a cluster management service device connected to the plurality of image analysis devices, wherein the cluster management service device is serially connected to the plurality of image analysis devices The analysis result of the at least one tracked object generated by each image analysis device is used to generate the track of the at least one tracked object.

In one embodiment, each image analysis device receives a plurality of real-time image streams, wherein each image analysis device includes a tracking module to concatenate the feature analysis results of at least one object of the plurality of real-time image streams to generate The secondary trajectory of the at least one object, wherein the tracking module of the cluster management service device concatenates the multiple secondary trajectories generated by the plurality of image analysis devices to track the trajectory of the at least one object.

In one embodiment, the at least one object includes a person, and the analysis result includes facial features of the person.

In one embodiment, the at least one object includes a luggage, and the analysis result includes features of the luggage.

In one embodiment, the plurality of image analysis devices are distributed in a detection area, wherein at least one specific object in the detection area is predetermined not to be tracked.

In one embodiment, the detection area includes a station whose personnel are scheduled not to be tracked.

In one embodiment, if a certain specific object in the detection area is not detected for a predetermined period of time, it is determined that the specific object has left the detection area and is no longer detected by the plurality of images. The analysis device does real-time tracking.

In one embodiment, if a tracked object within the detection area continues to stay in a specific area within the detection area for more than a predetermined length of time, an abnormal stay event is issued.

An embodiment of the present invention proposes a distributed object tracking system, comprising: a first plurality of image analysis devices, wherein each image analysis device is connected to at least one first photographing device to analyze the at least one first photographing device an object in at least one first real-time image stream transmitted by the device to generate a first analysis result of the at least one object being tracked; a first cross-camera tracking device connected to the first plurality of image analysis devices, wherein The first cross-camera tracking device concatenates the first analysis results of the at least one tracked object generated by each image analysis device in the first plurality of image analysis devices to generate the at least one tracked object. A first track; a second plurality of image analysis devices, wherein each image analysis device is connected to at least one second camera device to analyze at least one second real-time image stream transmitted by the at least one second camera device object to generate a second analysis result of the tracked at least one object; a second cross-camera tracking device is connected to the second plurality of image analysis devices, wherein the second cross-camera tracking device is serially connected to the first A second analysis result of the at least one tracked object generated by each image analysis device in the two plurality of image analysis devices to generate a second trajectory of the at least one tracked object; and a cluster management service device, Connected to the first cross-camera tracking device and the second cross-camera tracking device to concatenate the first track and the second track of the at least one tracked object.

In one embodiment, the distributed object tracking system tracks a target according to a specific target image.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area, wherein the first detection area and the second detection area At least one specific object in the detection area is predetermined not to be tracked.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area. If a certain specific object in all the detection areas of the group management service device is not detected for more than a predetermined length of time, it is determined that the specific object has left all the detection areas under the jurisdiction of the cluster management service device, so that it is no longer detected by the first detection area. A plurality of image analysis devices perform real-time tracking with the second plurality of image analysis devices.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area, wherein, if a tracked object continues to stay in An abnormal stay event is issued if a specific area in all detection areas under the control of the cluster management service device exceeds a predetermined time length.

Description of drawings

FIG. 1 illustrates a schematic diagram of a distributed object tracking system in an embodiment of the present invention.

FIG. 2 illustrates a schematic diagram of a distributed object tracking system in an embodiment of the present invention.

FIG. 3 illustrates a schematic diagram of a distributed object tracking system in an embodiment of the present invention.

FIG. 4 illustrates a schematic diagram of a distributed object tracking system in an embodiment of the present invention.

Explanation of reference numerals: 101-photography device; 102-photography device; 105-image analysis device; 105a-image analysis module; 106-image analysis device; 106a-image analysis module; 110-swarm management service device; 110a-tracking module ;201-photographic device; 202-photographic device; 203-photographic device; 204-photographic device; 205-image analysis device; 205a-image analysis module; 205b-image analysis module; 206-image analysis device; 206a-image analysis module ; 206b-image analysis module; 210-swarm management service device; 210a-tracking module; 301-camera device; 302-photograph device; 303-photograph device; ; 310b-image analysis module; 310c-tracking module; 311-image analysis device; 311a-image analysis module; 311b-image analysis module; 311c-tracking module; 320-swarm management service device; 320a-tracking module; device; 402-camera device; 403-camera device; 404-camera device; 411-image analysis device; 412-image analysis device; 421-image analysis device; 422-image analysis device; 431-cross-camera tracking device; Tracking module; 432-cross-camera tracking device; 432a-tracking module; 440-swarm management service device; 440a-tracking module.

Detailed ways

The aforementioned and other technical contents, features and effects of the present invention will be clearly presented in the following detailed description of preferred embodiments with reference to the drawings. However, it should be noted that the following examples are not intended to limit the present invention.

An embodiment of the present invention proposes a distributed object tracking system. Please refer to FIG. 1, the distributed object tracking system includes: a plurality of image analysis devices 105, 106, wherein each image analysis device 105, 106 analyzes the real-time image series captured by its corresponding camera device 101, 102 Objects in the stream to generate analysis results of objects that can be tracked; and a cluster management service device 110, connected to the plurality of image analysis devices 105, 106, wherein the cluster management service device 110 includes a tracking module 110a to connect The analysis results of the tracked object generated by the plurality of image analysis devices 105 and 106 are used to track the trajectory of the object.

In one embodiment, the at least one object includes a person, and the analysis result includes facial features of the person.

In one embodiment, the at least one object includes a luggage, and the analysis result includes features of the luggage.

In one embodiment, the plurality of image analysis devices are distributed in a detection area, wherein at least one specific object in the detection area is predetermined not to be tracked.

In one embodiment, the detection area includes a station whose personnel are scheduled not to be tracked.

An embodiment of the present invention proposes a distributed object tracking system. Please refer to FIG. 2, the distributed object tracking system includes: a plurality of image analysis devices 205, 206, wherein each image analysis device 205, 206 has a corresponding plurality of image analysis modules 205a, 205b, 206a, 206b To analyze the objects in the real-time video streams captured by the corresponding multiple camera devices 201, 202, 203, 204 to generate analysis results of objects that can be tracked; and a cluster management service device 210, connected to the multiple Image analysis devices 205, 206, wherein the cluster management service device 210 includes a tracking module 210a for concatenating the analysis results of the tracked object generated by the plurality of image analysis devices 205, 206 to track the trajectory of the object.

An embodiment of the present invention proposes a distributed object tracking system. Please refer to FIG. 3, the distributed object tracking system includes: a plurality of image analysis devices 310, 311, wherein each image analysis device 310, 311 has a corresponding plurality of image analysis modules 310a, 310b, 311a, 311b to analyze the objects in the real-time image streams captured by the corresponding multiple photographing devices 201, 202, 203, 204 to generate analysis results of objects that can be tracked, wherein each image analysis device 310, 311 includes a The corresponding tracking modules 310c, 311c are used to concatenate the analysis results of the objects of the plurality of real-time video streams to generate a secondary trajectory of at least one object; and a cluster management service device 320 connected to the plurality of image analysis devices 310 , 311, wherein the cluster management service device 320 includes a tracking module 320a for concatenating the multiple sub-trajectories generated by the multiple image analysis devices 310, 311 to track the trajectory of the at least one object.

In one embodiment, the distributed object tracking system of the present invention can be divided into multiple detection areas, and each image analysis device can perform face detection on the real-time image stream captured by the camera in a single detection area or other non-human object detection to obtain facial features and other identifiable features to track said person or other non-human object in the single detection area. In one embodiment, each image analysis device can perform face detection or other non-human object detection on real-time image streams captured by multiple cameras from different locations in a single detection area to obtain Facial features and other identifiable features to track said person or other non-human object within the single detection area. In one embodiment, after performing preliminary face recognition or object recognition, objects that do not need to be tracked may be excluded. For example, at a station, the registered station personnel do not need to be tracked; or specify only special objects need to be tracked, such as luggage. If there is an object to be tracked, the analysis result of the object to be tracked is recorded for subsequent cross-detection area tracking, and the analysis result may include the characteristics of the object to be tracked. The tracking module of the cluster management service device can concatenate the analysis results of each detection area to track the trajectory of the at least one object in all detection areas.

In one embodiment, each image analysis device can perform face detection or other non-human object detection on real-time image streams captured by multiple cameras from different locations in a single detection area to obtain Face features and other identifiable features, wherein each image analysis device includes a tracking module to concatenate the analysis results of at least one object of the plurality of real-time image streams to generate a secondary trajectory of the at least one object, wherein The tracking module of the cluster management service device concatenates the plurality of sub-trajectories generated by the plurality of image analysis devices to track the trajectory of the at least one object.

In one embodiment, if a certain specific object in the detection area is not detected for a predetermined period of time, it is determined that the specific object has left the detection area and is no longer detected by the plurality of images. The analysis device does real-time tracking.

In one embodiment, comparing object features for a predetermined length of time, if an object feature reaches a certain score (this score can be dynamically adjusted according to scene learning), the object is marked as an object to be continuously tracked.

In one embodiment, if a tracked object within the detection area continues to stay in a specific area within the detection area for more than a predetermined length of time, an abnormal stay event is issued.

An embodiment of the present invention proposes a distributed object tracking system. Please refer to FIG. 4, the distributed object tracking system includes: a first plurality of image analysis devices 411, 412, wherein each image analysis device 411, 412 respectively analyzes the images of its corresponding plurality of photography devices 401, 402 objects in the captured real-time image stream to generate a first analysis result of at least one object that can be tracked; a first cross-camera tracking device 431 connected to the first plurality of image analysis devices 411, 412, wherein the The first cross-camera tracking device 431 includes a tracking module 431a for concatenating the first analysis results of the at least one tracked object generated by the first plurality of image analysis devices 411, 412 to track an object of the at least one object A first track; a second plurality of image analysis devices 421, 422, wherein each image analysis device 421, 422 analyzes the objects in the real-time image stream captured by its corresponding plurality of camera devices 403, 404 to generate possible A second analysis result of at least one object being tracked; a second cross-camera tracking device 432 connected to the second plurality of image analysis devices 421, 422, wherein the second cross-camera tracking device 432 includes a tracking module 432a tracking a second track of the at least one object by concatenating the second analysis results of the at least one tracked object generated by the second plurality of image analysis devices 421, 422; and a cluster management service device 440, Connected to the first cross-camera tracking device 431 and the second cross-camera tracking device 432, wherein the cluster management service device 440 includes a tracking module 440a for connecting the first track and the second track of the at least one object in series trajectory.

Please note that each image analysis device 411 , 412 , 421 , 422 in FIG. 4 can be the image analysis device in FIG. 1 , or the image analysis device in FIG. 2 , or the image analysis device in FIG. 3 .

In one embodiment, the distributed object tracking system tracks a target according to a specific target image.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area, wherein the first detection area and the second detection area At least one specific object in the detection area is predetermined not to be tracked.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area. If a certain specific object in all the detection areas of the group management service device is not detected for more than a predetermined length of time, it is determined that the specific object has left all the detection areas under the jurisdiction of the cluster management service device, so that it is no longer detected by the first detection area. A plurality of image analysis devices perform real-time tracking with the second plurality of image analysis devices.

In one embodiment, the first plurality of image analysis devices are distributed in a first detection area, and the second plurality of image analysis devices are distributed in a second detection area, wherein, if a tracked object continues to stay in An abnormal stay event is issued if a specific area in all detection areas under the control of the cluster management service device exceeds a predetermined time length.

The cluster management device of the present invention can communicate with the plurality of image analysis devices through the Internet, the cluster management device can have an Internet address, and the plurality of image analysis devices establish contact with the cluster management device through the fixed Internet address so that the The cluster management device brings the plurality of image analysis devices into its management scope. The cluster management device and the plurality of image analysis devices can use a communication protocol (Protocol) to communicate with each other. The cluster management device may have a management interface for the user to manage the plurality of image analysis devices, for example, the management interface may display the location distribution map of the plurality of image analysis devices, or input a target image through the management interface to Track that target.

The object tracking of the present invention can refer to the basis, such as facial features, or human shape features such as clothing, body shape, etc., or the location of the equipment that detects the object to exclude unreasonable action tracks.

The "cross-camera tracking device" in an embodiment of the present invention can increase layer superposition, so that computing resources can be distributed, and the entire detection area can be expanded to track objects.

The "real-time cross-camera object tracking" of the present invention can realize real-time event search, and the object can be tracked wherever it goes. The "real-time cross-camera object tracking" of the present invention can also provide abnormal behavior alarms to proactively notify relevant units of suspicious objects. The present invention can also collect a large number of events into objects in real time, which can simplify event presentation/storage and speed up post-event search.

In one embodiment, the image analysis device of the present invention can control the lens of the PTZ camera to follow the target object. PTZ means that the lens of the camera can perform different functions such as panning left and right (Pan), tilting up and down (Tile) and zooming in (Zoom). Function.

In one embodiment, the cluster management device is capable of image analysis, and the cluster management device can perform at least a part of the work.

In an embodiment, the cluster management device of the present invention can be combined with the image analysis device, that is to say, the cluster management device can also have the function of the image analysis device of the present invention, so as to reduce the cost of the system.

In one embodiment, the cluster management device of the present invention does not have the capability of image analysis.

In one embodiment, the tracking module of the image analysis device of the present invention includes at least one central processing unit (CPU).

In one embodiment, the tracking module of the image analysis device of the present invention includes at least one central processing unit (CPU) and at least one image processing unit (GPU) or an image processing hardware accelerator.

In one embodiment, the tracking module of the image analysis device of the present invention includes at least one central processing unit (CPU) and at least one image processing hardware accelerator.

Although the present invention is disclosed above with the aforementioned preferred embodiments, it is not intended to limit the present invention. Any person familiar with the similar art can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, The scope of patent protection of the present invention must be defined by the claims as the criterion.

Claims (12)

1. a kind of distributed object tracking system, characterized by comprising:

Multiple image analysing computer devices, wherein each image analysing computer device is connected to an at least camera, to analyze this, at least one is taken the photograph The object in an at least real-time imaging crossfire that image device is transmitted is to generate the analysis result of at least one object being tracked；With And

One cluster management service unit is connected to multiple image analysing computer device, and wherein the cluster management service unit concatenates institute State the analysis result of at least one object being tracked caused by each image analysing computer device in multiple image analysing computer devices To generate the track of at least one object being tracked.

2. distributed object tracking system as described in claim 1, which is characterized in that each image analysing computer device connects Multiple cameras are connected to analyze the object in multiple real-time imaging crossfires that multiple camera is transmitted to generate this The subslot for at least one object being tracked, wherein the cluster management service unit concatenates the multiple image analysing computer device The subslot caused by each image analysing computer device is to generate the track of at least one object being tracked.

3. distributed object tracking system as described in claim 1, which is characterized in that at least one object includes a people, The analysis result includes the face feature of the people.

4. distributed object tracking system as described in claim 1, which is characterized in that at least one object includes a luggage Case, the analysis result include the feature of the luggage case.

5. distributed object tracking system as described in claim 1, which is characterized in that multiple image analysing computer device distribution Region is detected in one, wherein this detects that at least special object in region is scheduled not to be tracked.

6. distributed object tracking system as claimed in claim 5, which is characterized in that it includes a station that this, which detects region, The staff at the station is scheduled to be tracked.

7. distributed object tracking system as claimed in claim 5, which is characterized in that if this detect it is a certain in region Special object is more than that a predetermined time length is not all detected, then assert that the special object has left this and detected region, from And real-time tracing is no longer done by multiple image analysing computer device.

8. distributed object tracking system as claimed in claim 5, which is characterized in that if this detects the quilt in region The object persistence of tracking hangs out in this and detects the specific region in region and be more than a predetermined time length, then issues exception Stay event.

9. a kind of distributed object tracking system, characterized by comprising:

A image analysing computer device more than one first, wherein each image analysing computer device is connected at least one first camera to analyze The object at least one first real-time imaging crossfire that at least one first camera is transmitted is tracked at least with generating One first analysis result of an object；

One first across video camera follow-up mechanism is connected to more than first a image analysing computer device, and wherein this first is chased after across video camera This is tracked at least caused by each image analysing computer device in more than first a image analysing computer devices described in track device serial connection The first of an object analyzes result to generate one first track of at least one object being tracked；

A image analysing computer device more than one second, wherein each image analysing computer device analysis be connected at least one second camera with The object analyzed in at least one second real-time imaging crossfire that at least one second camera is transmitted is tracked with generating this At least one object one second analysis result；

One second across video camera follow-up mechanism is connected to more than second a image analysing computer device, and wherein this second is chased after across video camera This is tracked at least caused by each image analysing computer device in more than second a image analysing computer devices described in track device serial connection The second of an object analyzes result to generate one second track of at least one object being tracked；And

One cluster management service unit, be connected to first across the video camera follow-up mechanism and second across the video camera follow-up mechanism with Concatenate first track and second track of at least one object being tracked.

10. distributed object tracking system as claimed in claim 9, which is characterized in that the distributed object tracing system System is according to a specific target image to track the target.

11. distributed object tracking system as claimed in claim 9, which is characterized in that more than first a image analysing computer dress It sets and is distributed in one first and detects region, which is distributed in one second and detects region, wherein if should The administrative all a certain special objects detected in region of cluster management service unit be more than a predetermined time length all not by Detect, then assert that the special object has left administrative all of the cluster management service unit and detected region, thus no longer by More than first a image analysing computer device and more than second a image analysing computer device do real-time tracing.

12. distributed object tracking system as claimed in claim 9, which is characterized in that more than first a image analysing computer dress It sets and is distributed in one first and detects region, which is distributed in one second and detects region, wherein a such as quilt The object persistence of tracking hangs out in administrative all of the cluster management service unit and detects in region a specific region and more than one Scheduled time span then issues exception and stays event.