CN108388845A - Method for checking object and system - Google Patents
Method for checking object and system Download PDFInfo
- Publication number
- CN108388845A CN108388845A CN201810111667.7A CN201810111667A CN108388845A CN 108388845 A CN108388845 A CN 108388845A CN 201810111667 A CN201810111667 A CN 201810111667A CN 108388845 A CN108388845 A CN 108388845A
- Authority
- CN
- China
- Prior art keywords
- zone
- foreground
- hot
- monitoring hot
- monitoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
This disclosure relates to which a kind of method for checking object for the detection of video surveillance hot-zone, includes the following steps:The background that input video frame is rejected using background removal algorithm, to obtain the foreground image of video frame;The foreground image is divided into several foreground subregions and draws its contour line, and detect the foreground subregion all or part of whether be located at monitoring hot-zone in;If all or part of of foreground subregion is located in monitoring hot-zone, object detection is carried out to current video frame;And if there is object type to be detected is present within the scope of monitoring hot-zone, then trigger alarm.It realizes that hot-zone is detected by using the mode that Analysis on Prospect and object detection are combined, testing result can be quickly obtained.
Description
Technical field
This disclosure relates to which a kind of method for checking object and device, are especially for the object detection of video hot-zone detection
Method and apparatus.
Background technology
It is supervised in the hot-zone (ROI, Region of Interest) of such as particular place of safety and protection monitoring, traffic monitoring etc
In control, many method for checking object are had existed.Some method for checking object are by the special pedestrian's identification model of training, whole
Object detection is carried out in a video flowing.These methods are usually whole to realize object detection using deep learning algorithm, time-consuming.
For example, Chinese patent application Publication No. CN107301380A discloses a kind of side identified for pedestrian in video monitoring scene
Method can just obtain being no more than 5 frame results in one second with this method.
In addition, some method for checking object according to the difference of front and back frame find foreground object, then using foreground object as
Target carries out object detection.Then, such method does not use object detection algorithm that object is identified, therefore can not root
Selectively alarm is realized according to object type.For example, Chinese patent application, which discloses CN102236902A, discloses a kind of target inspection
Method and apparatus are surveyed, it uses the methods for being compared front and back frame to be changed analysis, to search intrusion object position.
But the method does not use object detection algorithm that object is identified, therefore can not be realized according to object type selectable
Alarm.
In addition, also some method for checking object find mobile object according to the difference of front and back frame, then according to monitoring hot-zone
The ratio of interior discrepancy judges intrusion object.It discloses a kind of movement for example, Chinese patent application discloses CN102340619A and detects
Method and apparatus are surveyed, this method and device are exactly to find mobile object according to the difference of front and back frame, then according in monitoring hot-zone
The ratio of discrepancy judges intrusion object.However, this method do not account for caused by due to camera angles problem " project into
Invade ", therefore object image projecting can not be eliminated in monitoring hot-zone, and the pseudo- announcement that physical location is generated when monitoring outside hot-zone
It is alert.This in order to overcome the problems, such as, common mode is rearranged to camera position so that camera face monitored space
Domain (i.e. common hot-zone), to eliminate pseudo- warning situation caused by projection invasion.But this side for rearranging camera
Formula quantities is big, and track remodelling is complicated, and huge improvement cost is brought for user.Moreover, once transformation is completed, monitoring hot-zone is not
It changes.Once change monitoring hot-zone changes or monitoring direction changes, and needs arranged direction or cloth that camera is transformed again
Seated position, it is also possible to lead to track remodelling, this can lead to the repetition cost payout of user.
Therefore, it is necessary to one kind can quickly, selectively identify object, can be adjusted flexibly monitoring hot-zone can avoid again
The method for checking object of puppet warning.
Invention content
For this purpose, the purpose of this patent is Utilization prospects separation and the technology that object detection is combined, ensureing detecting system
On the basis of processing speed, more accurate Object identifying is realized.Meanwhile this patent judges that algorithm is effective by the invasion of optimization
Reduce due to " projection invasion " caused by video camera setting angle pseudo- alarm, to for security protection application provide it is a kind of it is efficient, can
It leans on, the method and apparatus of intelligence.
According to one aspect of the disclosure, a kind of method for checking object is provided, is included the following steps:A) background is utilized
The background of algorithm rejecting input video frame is removed to obtain the foreground image of video frame, which is divided into several foregrounds
Subregion simultaneously draws its contour line, and detect the foreground subregion all or part of whether be located at monitoring hot-zone in;With
And if b) foreground subregion all or part of be located at monitoring hot-zone in, to current video frame carry out object detection, and
If there is object type to be detected is present within the scope of monitoring hot-zone, then signal (the triggering announcement for indicating to detect object is sent out
It is alert).
Preferably, step a) includes:A1 foreground/background) is carried out to received video frame using background subtraction
Separation, to export foreground image;A2 morphological transformation) is carried out to foreground image, to eliminate the gap between foreground pixel, and is eliminated
Noise point;A3 the foreground image after optimization) is divided into several subregions and calculates and draw out its contour line;And it a4) examines
Survey the foreground subregion all or part of whether be located at monitoring hot-zone in.
Preferably, step b) includes:B1 object detection) is carried out in raw video image;B2) judgement detects
Whether object is located in monitoring hot-zone, and if the object detected is located in monitoring hot-zone, sends out expression and detect pair
The signal of elephant.
At least one of preferably, the step b2) include the following steps:Judge the entire object outline that detects whether with
Monitor hot-zone overlapping;Judge whether the object outline bottom edge detected is Chong Die with monitoring hot-zone;Judge the object outline detected
Whether the specific fragment on bottom edge is Chong Die with monitoring hot-zone;With the specific region that judges in the object outline that detects whether with monitoring
Hot-zone is overlapped.
Preferably, the step b2) further include:The object outline detected is plotted on raw video image;And it will
The video frame of rendered object profile achieves, and sends out the signal for indicating to detect object.
Preferably, the method for checking object further includes:C) if there is object type to be detected is present in monitoring hot-zone model
In enclosing, then the object is continuously tracked in video.
Preferably, the monitoring hot-zone refers to the specified region in monitor video, and if not specifying region, it is described
It is exactly entire monitored picture region to monitor hot-zone.
A kind of object test equipment for the detection of video surveillance hot-zone another aspect of the present disclosure provides,
Including:Foreground detection device, using the background of background removal algorithm rejecting input video frame to obtain the foreground image of video frame,
The foreground image is divided into several subregions and draws its contour line, all or part of for detecting the foreground subregion is
It is no to be located in monitoring hot-zone;And object identifiers, foreground subregion all or part of be located at monitoring hot-zone in when,
Object detection is carried out with current video frame, and sends out the signal for indicating to detect object when detecting object.
Preferably, which includes:Foreground separator, for utilizing background subtraction to received video
Frame carries out foreground/background separation, to export foreground image;Modality conversion device, for being optimized to foreground area, before elimination
Gap between scene element and foreground noise;Profile renderer, for the foreground image after optimization to be divided into several subregions
And draw its contour line;And the first monitoring hot-zone comparator, all or part of for detecting the foreground subregion are
It is no to be located in monitoring hot-zone.
Preferably, which includes:Object detector, for carrying out object detection in the video frame of image;
And the second monitoring hot-zone comparator, subject area for judging to detect whether with monitor hot-zone there are Chong Die, if deposited
It is being overlapped, is then exporting the subject area of intrusion monitoring hot-zone to send out the signal for indicating to detect object.
Preferably, which is configured as realizing at least one of following action:Judgement detects
Entire object outline whether with monitoring hot-zone it is Chong Die;Judge whether the object outline bottom edge detected is Chong Die with monitoring hot-zone;
Judge whether the specific fragment on the object outline bottom edge detected is Chong Die with monitoring hot-zone;In the object outline detected with judgement
Specific region whether with monitoring hot-zone it is Chong Die.
Preferably, which is also plotted in the object outline detected on raw video image,
The video frame of rendered object profile is achieved, and sends out the signal for indicating to detect object.
Preferably, which further includes object tracing device, for being present in if there is object type to be detected
It monitors within the scope of hot-zone, then the object is continuously tracked in video.
The disclosure realizes that hot-zone is detected in such a way that Analysis on Prospect and object detection are combined, and can be quickly obtained more
Detailed testing result information, such as invade the classification of object.Also, algorithm is invaded in the monitoring hot-zone due to the use of optimization,
Therefore it can reduce to the maximum extent due to " projection invasion " caused by camera setting angle and send out expression and detect object
False signal, the expression sent out detects the signal quality higher, lower to equipment installation requirement of object, can be to greatest extent
Ground is monitored using existing device, to substantially reduce implementation cost.
Description of the drawings
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is the schematic block diagram according to the object detection systems of one embodiment of the disclosure;
Fig. 2 is the detailed schematic block diagram according to the object detection part of one embodiment of the disclosure;
Fig. 3 is the block diagram according to the foreground detection device of one embodiment of the disclosure;
Fig. 4 is the block diagram according to the object identifiers of one embodiment of the disclosure;
Fig. 5 is the flow chart according to the method for checking object of one embodiment of the disclosure;
Fig. 6 is the foreground image according to the video frame of an example;
Fig. 7 is the foreground image carried out to foreground image shown in fig. 6 after morphological transformation;
Fig. 8 is the profile diagram of the foreground subregion of foreground image shown in Fig. 7;
Fig. 9 be in foreground image shown in Fig. 8 only retain with monitor hot-zone there are the subregion of Chong Die foreground point after
View;
Figure 10 A be take whole object outline regions carry out the object detection result after invasion judgement output preserve part
The view of upper display;
Figure 10 B are to carry out being preserved on part in output for the object detection result after invasion judgement using object outline bottom edge
The view of display;
Figure 10 C are to carry out being protected in output for the object detection result after invasion judgement using the part on object outline bottom edge
The view shown on nonresident portion;And
Figure 11 is the final detection result view based on Analysis on Prospect.
Specific implementation mode
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
It is the purpose only merely for description specific embodiment in the term that the disclosure uses, is not intended to be limiting and originally opens.It removes
Non- defined otherwise, every other scientific and technical terms used herein have and those skilled in the art
Normally understood identical meaning.The "an" of singulative used in disclosure and the accompanying claims book, " institute
State " and "the" be also intended to including most forms, unless context clearly shows that other meanings.It is also understood that making herein
Term "and/or" refer to and include one or more associated list items purposes any or all may combine.
It will be appreciated that though various information, but this may be described using term first, second, third, etc. in the disclosure
A little information should not necessarily be limited by these terms.These terms are only used for same type of information being distinguished from each other out.For example, not departing from
In the case of disclosure range, first can also be referred to as second, and vice versa.Depending on context, word as used in this
Language " if " can be construed to " ... when " or " when ... " or " in response to determination ".
In order to make those skilled in the art more fully understand the disclosure, with reference to the accompanying drawings and detailed description to this public affairs
It opens and is described in further detail.
Fig. 1 is the schematic block diagram according to the object detection systems of one embodiment of the disclosure.As shown in Fig. 1, this is right
As detecting system includes that video acquisition part 110, object detection part 120 and output preserve part 130.According to the one of the disclosure
A embodiment, video acquisition part 110 are used to acquire video data, including by such as monitoring IP camera (not shown) etc
Picture pick-up device shoot video flowing, video stream data is received by network (not shown) etc., and (can not be shown from storage device
Go out) in read video stream data etc..Video flowing that is produced or receiving is output to object detection portion by video acquisition part 110
Divide 120.
After receiving video stream data in video acquisition part 110, processing first is received object detection part 120
To data with obtain for object detection video frame and according to user require or default setting determine monitoring hot-zone.According to this
Open, user can be according to the demand of oneself, appointing in the field range that video frame or video acquisition part 110 can be acquired
A meaning part is set as monitoring hot-zone.Then object detection part 120 according to identified monitoring hot-zone come after detection process
Supervision object in video requency frame data.In addition, object detection part 120 exports testing result when detecting supervision object
Part 130 is preserved to output.The structure and function of object detection part 120 will be described in further detail below.
Output preserves part 130 and can include but is not limited to video display (not shown), storage device (not shown), prison
All or part of in wall (not shown), loud speaker (not shown) etc. is controlled, for visually audibly showing object detection result, example
It such as sends out the signal (for example, triggering audio alarm, light warning etc.) for indicating to detect object and/or preserves testing result and retain
Put on record.
Fig. 2 is the detailed schematic frame according to the object detection part shown in FIG. 1 120 of one embodiment of the disclosure
Figure.As shown in Fig. 2, the object detection part 120 includes hot-zone determining device 210, video frame converter 220, foreground detection device
230, object identifiers 240 and object tracing device 250.
According to one embodiment of the disclosure, the video frame converter 220 receive produced by video acquisition part 110 or
The video flowing received, and it is converted into video frame.
According to one embodiment of the disclosure, hot-zone determining device 210 can be can screening device, be used to be wanted according to user
It asks or default, specifies monitoring section domain as monitoring hot-zone in monitoring video flow.If without hot-zone determining device 210
Or hot-zone determining device 210 without specified monitoring thermal region, then the monitoring hot-zone acquiescence is exactly entire monitored picture region.
Next, foreground detection device 230 receives the video frame from video frame converter 220, and utilize background subtraction
Deng carrying out foreground separation to received video frame, foreground image is generated to reject the image background in video image frame.
Then, foreground detection device 230 carries out morphological transformation to foreground image, to eliminate the gap between foreground pixel, and eliminates foreground
Noise, and be converted into several including foreground by foreground point using the border following Analysis of Topological Structure algorithm of such as black white image
The subregion of point, and calculate the contour line of each foreground subregion.Foreground detection device 230 also detects the whole or one of foreground subregion
Whether part is located in monitoring hot-zone (whether there is the foreground subregion Chong Die with monitored hot-zone), until before separation
It is found with the presence of the foreground subregion Chong Die with monitored hot-zone in scape image.There is the case where foreground subregion of overlapping
Under, foreground detection device 230 just will be seen that the video frame with the presence of the foreground subregion Chong Die with monitored hot-zone is sent to pair
As identifier 240.
Then, object identifiers 240 have been found that with the presence of foreground Chong Die with monitored hot-zone received
Specified object in the video frame in region is identified, and the output intrusion subject area after confirming specified object invasion.Specifically
Ground says that object identifiers 240 carry out object detection in raw video image, judges whether the object detected is located at monitoring heat
In area, and when the object detected is located in monitoring hot-zone, pair in the monitoring hot-zone that object identifiers 240 detect
As profile is plotted in video frame, it is output to output shown in FIG. 1 and preserves part 130, to be shown, played to send out table
Show the signal for detecting object and/or preserve to retain on a storage device and puts on record.
According to one embodiment of the disclosure, intrusion subject area can also be sent to object and chased after by object identifiers 240
Track device 250, and by object tracing device 250 according to received intrusion subject area after video frame converter 220 exports
It carries out independent trails in continuous video frame, and the band of position of continuous updating object and result is output to output preserves part
130。
Here, foreground detection device 230 in the foreground image in separation until finding with the presence of, ability Chong Die with monitored hot-zone
It will be seen that and be sent to object identifiers 240 with the presence of the video frame Chong Die with monitored hot-zone.So those are not present
The video frame that foreground image has hot-zone to be overlapped can be ignored, and disregard, and therefore, reduce mass data processing task, therefore
Progress image recognition can be used with high efficiency, low cost.
Fig. 3 is the block diagram according to the foreground detection device 230 as shown in Figure 2 of one embodiment of the disclosure.Such as Fig. 3 institutes
Show, which includes foreground separator 310, modality conversion device 320, the monitoring of profile renderer 330 and first hot-zone
Comparator 340.
First, foreground separator 310 carries out foreground/background separation using background subtraction to received video frame,
White represents foreground area in the black white image of output.
Then modality conversion device 320 optimizes foreground area, to eliminate the gap between foreground pixel.Generally use
Closing (CLOSING) algorithm repairs foreground, i.e., (Dilation) integration region is first expanded to foreground, then corrode
(Erosion) noise is eliminated, but it is (such as open (OPENING- first corrodes further expansion) that other optimization algorithms can also be used herein
Algorithm or exclusive use expansion or erosion arithmetic).
Foreground point is converted by profile renderer 330 using the border following Analysis of Topological Structure algorithm of such as black white image
It is several to include the subregion of foreground point, and the contour line of each foreground subregion is calculated, then sketch the contours of each foreground subregion
Profile.
Then, the first monitoring hot-zone comparator 340 detects each foreground subregion profile whether there are Chong Die for hot-zone with monitoring.
If there is overlapping, then the video frame is sent to object identifiers 240 as shown in Figure 2.
Fig. 4 is the block diagram according to the object identifiers 240 as shown in Figure 2 of one embodiment of the disclosure.Such as Fig. 4 institutes
Show, which includes that object detector 410 and second monitors hot-zone comparator 420.
First, object detector 410 carries out object detection in the video frame of image.Then rectangular object profile is sent out
The second monitoring hot-zone comparator 420 is given, judging subject area, whether there are Chong Die with monitoring hot-zone.If there is overlapping, then
The subject area of output intrusion monitoring hot-zone.
According to one embodiment of the disclosure, in order to reduce such as due to the setting angle of video camera caused by " project into
Invade " puppet alarm signal, the object outline for being sent to the second monitoring hot-zone comparator 420 can be the bottom edge of primary object profile.
According to one embodiment of the disclosure, object actual boundary is more than to rectangular object profile in order to be further reduced
Caused by " projection invasion " pseudo- alarm signal, the object outline for being sent to the second monitoring hot-zone comparator 420 can be original right
As the bottom edge of profile a part (such as centered on original bottom edge midpoint, width be original base width 50% portion
Point).
According to one embodiment of the disclosure, caused by order to be further reduced specific shape object, " projection invasion " is pseudo- accuses
Alert signal can choose rectangle as the arbitrary subregion in profile as comparison domain is sent to the second monitoring hot-zone comparator
420。
According to one embodiment of the disclosure, the second monitoring hot-zone comparator 420 can paint the object outline detected
System is output to output preservation part 130 as shown in Figure 1 on raw video image, by the video frame of rendered object profile and achieves
Card is stayed, and sends out the signal for indicating to detect object.
Fig. 5 is the flow chart according to the method for checking object of one embodiment of the disclosure.As shown in figure 5, in step
In S510, is received using video acquisition part 110 shown in FIG. 1 or generate video flowing, then by generated video stream
To object detection part 120 as shown in Figure 1.
Next, in step S520, video flowing is converted by video by video frame converter 220 (as shown in Figure 2)
Frame, and determine monitoring thermal region by the way that hot-zone determining device 210 is (as shown in Figure 2), in order to from the monitoring of the video frame
Object is detected in hot-zone.
Next, in step S530, Utilization prospects separator 310 (as shown in Figure 3) from video frame converter 220 (such as
Shown in Fig. 2) video frame is received, and the image background in video image frame is rejected to generate foreground image using background subtraction.
Then, in step S540, morphological transformation is carried out to foreground image using modality conversion device 320 (as shown in Figure 3),
To eliminate the gap between foreground pixel.
Then, in step S550, such as profile renderer 330 as shown in Figure 3 utilizes the boundary of such as black white image
It follows Analysis of Topological Structure algorithm that foreground point is converted into several subregions for including foreground point, and calculates and draw each foreground
The contour line of subregion.
In step S560, the whole of foreground subregion is detected using the first monitoring hot-zone comparator 340 as shown in Figure 3
Or whether a part is located in monitoring hot-zone (whether there is the foreground subregion Chong Die with monitored hot-zone), until detaching
Foreground image in find with the presence of the foreground subregion Chong Die with monitored hot-zone.
If detecting the presence of coincidence in step S560, the first monitoring hot-zone comparator 340 will be seen that and supervised
The video frame that calorimetric area has overlapping is sent to object identifiers 240.If not detecting the presence of weight in step S560
It closes, then continues to monitor video flowing.
Next, in step S570, the detection of object identifiers 240 carries out specified object to received video frame and knows
Not, specify whether object is located in monitoring hot-zone to confirm, and the output intrusion subject area after object invasion is specified in confirmation, with
Trigger out the signal that expression detects object.Specifically, using object detector 410 as shown in Figure 4 in original video figure
Carry out object detection as in, by the second object for detecting of the monitoring judgement of hot-zone comparator 420 such as Fig. 4 shown in whether position
In monitoring hot-zone in, and the object detected be located at monitoring hot-zone in when, output testing result indicate to detect to send out
The signal of object.
Then, in step S580, the object outline in the monitoring hot-zone that object identifiers 240 detect, which is plotted in, to be regarded
On frequency frame, and it is output to output and preserves part 130.
According to one embodiment of the disclosure, in step S590, monitoring that object identifiers 240 can also will detect
Object outline in hot-zone sends object tracing device 250 as shown in Figure 2 to, so that object tracing device 250 passes through subsequent video
Stream is persistently tracked the object, and the video frame position with newer object outline position is output to output storage unit
Divide 130.
Example
The purpose of the example is that the video flowing generated using the shooting of IP video cameras realizes the intrusion detection of monitoring hot-zone.It detected
Journey is as follows.
1) foreground image is generated from video frame
Fig. 6 is the foreground image according to the video frame of the example.For the stream video file using the shooting of IP video cameras, lead to
Cross background/foreground partitioning algorithm (the Gaussian Mixture-based Background/ based on Gaussian Mixture
Foreground Segmentation Algorithm), calculate the mixing K Gaussian Profiles (mixture of each pixel in video
Of K Gaussian distribution), the pixel of relative quiescent in video frame is designated as background dot (in figure 6 with black
Color table shows), and the higher point of relative change rate in video frame is designated as foreground point (being indicated in figure 6 with white).In this way, just
Original video frame can be converted into the foreground image (as shown in Figure 6) of black and white.It should be appreciated that other classes can also be used in the present invention
As algorithm generate foreground image.
2) morphological transformation is carried out to foreground image.
Fig. 7 is the foreground image carried out to foreground image shown in fig. 6 after morphological transformation.As seen from Figure 6, due to original
The influence of the factors such as color, light, foreground image high granular in beginning video frame need to carry out form change to foreground image
It changes, to eliminate the gap between foreground pixel.It realizes region fusion, while removing discrete noise point.Therefore, in the present embodiment
Morphological transformation is carried out to foreground image shown in fig. 6 using closing algorithm, obtains the foreground image after morphological transformation (such as Fig. 7 institutes
Show).It should be appreciated that in other embodiments, it can be with specific features according to the embodiment, selection is open, and (OPENING- first corrodes
Further expansion) algorithm or expansion is used alone or erosion arithmetic carries out foreground transformation.
3) foreground subregion profile frame is generated:
Fig. 8 is the profile diagram of the foreground subregion of foreground image shown in Fig. 7.In fig. 8, it is handled to improve video
Foreground point in Fig. 7 is converted into several including foreground by efficiency using the border following Analysis of Topological Structure algorithm of black white image
The subregion of point, is used in combination rectangular shaped rim to identify.
4) judge whether foreground subregion overlaps with monitoring hot-zone
Fig. 9 be in foreground image shown in Fig. 8 only retain with monitor hot-zone there are the subregion of Chong Die foreground point after
View.As shown in figure 9, the monitoring for monitoring hot-zone, can filter out a large amount of foreground interference regions, analyze speed is promoted.It answers
Work as understanding, the monitoring hot-zone individually drawn may be not present in other embodiments of the invention, then in this case may be used
Whole image region is analyzed as monitoring hot-zone.
5) Object identifying is carried out to the image that there is the intrusion of monitoring hot-zone using Identifying Technique of Object
Figure 10 A-10C are that the object detection result of the example preserves the view shown on part in output, and wherein Figure 10 A are
Whole object outline regions are taken to carry out the view of the object detection result after invasion judgement shown on output preservation part, figure
10B be using object outline bottom edge carry out the object detection result after invasion judgement output preserve part on show regard
Figure, and Figure 10 C are to carry out being preserved in output for the object detection result after invasion judgement using the part on object outline bottom edge
The view shown on part.
The view of Figure 10 A-10C in order to obtain detects that (white side is trapezoidal in figure for monitoring hot-zone by Analysis on Prospect method
Frame) intrusion original video frame be sent to object identifiers 240 carry out Object identifying.Equally, to Object identifying and classification
On the basis of, object identifiers 240 match the contour area of the object of identification with monitoring hot-zone, check for weight
It is folded.If detecting overlapping, exports object outline and preserve part 130, such as monitor wall to output.It is also possible to will
Video frame storage is put on record to memory 1300.
It is to take whole object outline regions to carry out invasion to judge that is obtained regards comprising " projection invasion " in the view of Figure 10 A
Figure." projection invasion " pseudo- alarm signal, is monitoring caused by the problem of in order to eliminate in Figure 10 A due to camera setting angle
Object outline bottom edge may be used carry out invasion judgement when the intrusion detection of hot-zone, that is, will determine that object outline bottom edge whether with monitoring
There is the foundation for overlapping and judging as invasion in hot-zone.Its result is shown in fig. 1 ob.
Sentence in addition, the part on object outline bottom edge can also be used to carry out invasion when doing the intrusion detection of monitoring hot-zone
It is disconnected, that is, will determine that the part (such as 50%) on object outline bottom edge whether exist with monitoring hot-zone overlaps as invade judgement according to
According to.Its result is shown in fig 1 oc.As illustrated in figure 10 c, " projection invasion " pseudo- alarm signal is effectively eliminated in figure.
It should be appreciated that in other embodiments of the invention, the arbitrary part that object outline may be used optimizes
Invasion judge.For example, when monitoring hot-zone perpendicular to the door area on ground or fence, then it may be used and be included in object outline two
The line or part of it at longitudinal edge midpoint carry out invasion judgement.
Figure 11 is the final detection result view based on Analysis on Prospect.As shown in figure 11, although Utilization prospects isolation technics
It can judge in original video frame with monitoring hot-zone detection technique and mark the intrusion of monitoring hot-zone, but this judgement can not be right
It invades object and carries out Classification and Identification, often also need to artificial secondary judgement, the automation for greatly limiting detection technique is realized.And
Based entirely on the solution of Object identifying, there are computationally intensive, slow-footed defects, it is difficult to be applied in engineering.Therefore this is special
Profit in such a way that Object identifying is combined, is being ensured system running speed and is realizing the base of object type identification using Analysis on Prospect
On plinth, calculating task is minimized, implementation cost has been saved.
6) image that there is the intrusion of monitoring hot-zone is kept track using object tracing technology
When detecting the object invasion of monitoring hot-zone, Object identifying profile can be utilized to generate object tracing device, rear
The object is constantly tracked in continuous video frame.
Being described above with reference to attached drawing makes according to the method for checking object and device of the disclosure in one embodiment
Realize that hot-zone is detected with the mode that Analysis on Prospect and object detection are combined, so it can be obtained than traditional Analysis on Prospect method
To more detailed testing result information, such as the classification of intrusion object.
In another embodiment of the disclosure, prospect of the application analysis and the mode that object detection is combined realize that hot-zone is examined
It surveys, so it is faster than traditional method for checking object speed.
In the further embodiment of the disclosure, the monitoring hot-zone invasion algorithm of optimization is used, therefore can be with maximum limit
Degree ground is reduced since " projection invasion " pseudo- alarm signal, alarm quality higher install equipment caused by camera setting angle
It is required that it is lower, existing device can be maximally utilised and be monitored, to substantially reduce implementation cost.
The present invention is not limited to the range of specific embodiments described herein, these embodiments are intended as exemplary implementation
Example.Functionally identical product and method is obviously contained in the range of invention described herein.
The basic principle of the disclosure is described above in association with specific embodiment, however, it is desirable to, it is noted that this field
For those of ordinary skill, it is to be understood that the whole either any steps or component of disclosed method and device, Ke Yi
Any computing device (including processor, storage medium etc.) either in the network of computing device with hardware, firmware, software or
Combination thereof is realized that this is that those of ordinary skill in the art use them in the case where having read the explanation of the disclosure
Basic programming skill can be achieved with.
Therefore, the purpose of the disclosure can also by run on any computing device a program or batch processing come
It realizes.The computing device can be well known fexible unit.Therefore, the purpose of the disclosure can also include only by offer
The program product of the program code of the method or device is realized to realize.That is, such program product is also constituted
The disclosure, and the storage medium for being stored with such program product also constitutes the disclosure.Obviously, the storage medium can be
Any well known storage medium or any storage medium that developed in the future.
It may also be noted that in the device and method of the disclosure, it is clear that each component or each step are can to decompose
And/or reconfigure.These decompose and/or reconfigure the equivalent scheme that should be regarded as the disclosure.Also, execute above-mentioned series
The step of processing, can execute according to the sequence of explanation in chronological order naturally, but not need to centainly sequentially in time
It executes.Certain steps can execute parallel or independently of one another.
Above-mentioned specific implementation mode does not constitute the limitation to disclosure protection domain.Those skilled in the art should be bright
It is white, design requirement and other factors are depended on, various modifications, combination, sub-portfolio and replacement can occur.It is any
Modifications, equivalent substitutions and improvements etc. made by within the spirit and principle of the disclosure, should be included in disclosure protection domain
Within.
Claims (13)
1. a kind of method for checking object, includes the following steps:
A) foreground image is divided by the background of removal input video frame with obtaining the foreground image of the input video frame
Several foreground subregions simultaneously draw its contour line, and detect the foreground subregion all or part of whether be located at monitoring heat
Qu Zhong;And
If b) foreground subregion all or part of be located at monitoring hot-zone in, to current video frame carry out object detection, and
And if there is object type to be detected is present within the scope of monitoring hot-zone, then send out the signal for indicating to detect object.
2. method for checking object according to claim 1, wherein step a) include:
A1 it) carries out foreground/background to received video frame using background subtraction to detach, to export foreground image;
A2 morphological transformation) is carried out to foreground image, to eliminate the gap between foreground pixel;
A3 the foreground image after optimization) is divided into several subregions and calculates and draw out its contour line;And
Whether all or part of for a4) detecting the foreground subregion is located in monitoring hot-zone.
3. method for checking object according to claim 1, wherein step b) include:
B1 object detection) is carried out in raw video image;
B2) judge whether the object detected is located in monitoring hot-zone, and if the object detected is located in monitoring hot-zone,
Then send out the signal for indicating to detect object.
At least one of 4. method for checking object according to claim 3, wherein the step b2) include the following steps:
Judge whether the entire object outline detected is Chong Die with monitoring hot-zone;
Judge whether the object outline bottom edge detected is Chong Die with monitoring hot-zone;
Judge whether the specific fragment on the object outline bottom edge detected is Chong Die with monitoring hot-zone;With
Judge whether the specific region in the object outline detected is Chong Die with monitoring hot-zone.
5. method for checking object according to claim 3, wherein the step b2) further include:
The object outline detected is plotted on raw video image;And
The video frame of rendered object profile is achieved, and sends out the signal for indicating to detect object.
6. method for checking object according to claim 1, further includes:
C) if there is object type to be detected is present within the scope of monitoring hot-zone, then the object is continuously chased after in video
Track.
7. method for checking object according to claim 1, wherein the monitoring hot-zone refers to the specified area in monitor video
Domain, and if not specifying region, the monitoring hot-zone is exactly entire monitored picture region.
8. a kind of object test equipment for the detection of video surveillance hot-zone, including:
Foreground detection device, using the background of background removal algorithm rejecting input video frame to obtain the foreground image of video frame,
The foreground image is divided into several subregions and draws its contour line, all or part of for detecting the foreground subregion is
It is no to be located in monitoring hot-zone;And
Object identifiers, foreground subregion all or part of be located at monitoring hot-zone in when, for current video frame into
Row object detection, and the signal for indicating to detect object is sent out when detecting object.
9. object test equipment according to claim 8, wherein the foreground detection device include:
Foreground separator is detached for carrying out foreground/background to received video frame using background subtraction, before output
Scape image;
Modality conversion device, for being optimized to foreground area, to eliminate the gap between foreground pixel;
Profile renderer, for the foreground image after optimization to be divided into several subregions and draws its contour line;And
First monitoring hot-zone comparator, for detect the foreground subregion all or part of whether be located at monitoring hot-zone
In.
10. object test equipment according to claim 8, the wherein object identifiers include:
Object detector, for carrying out object detection in the video frame of image;And
Second monitoring hot-zone comparator, subject area for judging to detect whether with monitor hot-zone there are Chong Die, if deposited
It is being overlapped, is then exporting the subject area of intrusion monitoring hot-zone to send out the signal for indicating to detect object.
11. object test equipment according to claim 10, wherein the second monitoring hot-zone comparator are configured as realizing
At least one of following action:
Judge whether the entire object outline detected is Chong Die with monitoring hot-zone;
Judge whether the object outline bottom edge detected is Chong Die with monitoring hot-zone;
Judge whether the specific fragment on the object outline bottom edge detected is Chong Die with monitoring hot-zone;With
Judge whether the specific region in the object outline detected is Chong Die with monitoring hot-zone.
12. object test equipment according to claim 10, wherein the second monitoring hot-zone comparator will also detect
Object outline be plotted on raw video image, the video frame of rendered object profile is achieved, and send out expression and detect
The signal of object.
13. object test equipment according to claim 8 further includes object tracing device, for if there is object to be detected
Type is present within the scope of monitoring hot-zone, then is continuously tracked in video the object.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810111667.7A CN108388845A (en) | 2018-02-05 | 2018-02-05 | Method for checking object and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810111667.7A CN108388845A (en) | 2018-02-05 | 2018-02-05 | Method for checking object and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108388845A true CN108388845A (en) | 2018-08-10 |
Family
ID=63075220
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810111667.7A Pending CN108388845A (en) | 2018-02-05 | 2018-02-05 | Method for checking object and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108388845A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111126317A (en) * | 2019-12-26 | 2020-05-08 | 腾讯科技(深圳)有限公司 | Image processing method, device, server and storage medium |
CN112084813A (en) * | 2019-06-12 | 2020-12-15 | 杭州萤石软件有限公司 | Abnormal target detection method and device and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101483001A (en) * | 2009-01-04 | 2009-07-15 | 北京中星微电子有限公司 | Video-based intrusion detection method, detection apparatus and chip |
CN102842130A (en) * | 2012-07-04 | 2012-12-26 | 贵州师范大学 | Method for detecting buildings and extracting number information from synthetic aperture radar image |
CN103021186A (en) * | 2012-12-28 | 2013-04-03 | 中国科学技术大学 | Vehicle monitoring method and vehicle monitoring system |
CN103067692A (en) * | 2012-12-30 | 2013-04-24 | 同方(深圳)云计算技术股份有限公司 | Post-treatment detection method and system based on net-harddisk video recorder dangerous invasion |
CN103824070A (en) * | 2014-03-24 | 2014-05-28 | 重庆邮电大学 | Rapid pedestrian detection method based on computer vision |
CN104392464A (en) * | 2014-09-30 | 2015-03-04 | 天津艾思科尔科技有限公司 | Human intrusion detection method based on color video image |
CN105678811A (en) * | 2016-02-25 | 2016-06-15 | 上海大学 | Motion-detection-based human body abnormal behavior detection method |
CN106951899A (en) * | 2017-02-24 | 2017-07-14 | 李刚毅 | Method for detecting abnormality based on image recognition |
-
2018
- 2018-02-05 CN CN201810111667.7A patent/CN108388845A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101483001A (en) * | 2009-01-04 | 2009-07-15 | 北京中星微电子有限公司 | Video-based intrusion detection method, detection apparatus and chip |
CN102842130A (en) * | 2012-07-04 | 2012-12-26 | 贵州师范大学 | Method for detecting buildings and extracting number information from synthetic aperture radar image |
CN103021186A (en) * | 2012-12-28 | 2013-04-03 | 中国科学技术大学 | Vehicle monitoring method and vehicle monitoring system |
CN103067692A (en) * | 2012-12-30 | 2013-04-24 | 同方(深圳)云计算技术股份有限公司 | Post-treatment detection method and system based on net-harddisk video recorder dangerous invasion |
CN103824070A (en) * | 2014-03-24 | 2014-05-28 | 重庆邮电大学 | Rapid pedestrian detection method based on computer vision |
CN104392464A (en) * | 2014-09-30 | 2015-03-04 | 天津艾思科尔科技有限公司 | Human intrusion detection method based on color video image |
CN105678811A (en) * | 2016-02-25 | 2016-06-15 | 上海大学 | Motion-detection-based human body abnormal behavior detection method |
CN106951899A (en) * | 2017-02-24 | 2017-07-14 | 李刚毅 | Method for detecting abnormality based on image recognition |
Non-Patent Citations (2)
Title |
---|
方芳: "基于智能监控系统的视频图像质量检测的研究与实现", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
杨琦等: "基于opencv的绊线监控系统设计", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112084813A (en) * | 2019-06-12 | 2020-12-15 | 杭州萤石软件有限公司 | Abnormal target detection method and device and storage medium |
WO2020248543A1 (en) * | 2019-06-12 | 2020-12-17 | 杭州萤石软件有限公司 | Abnormal target detection method and device, and storage medium |
CN111126317A (en) * | 2019-12-26 | 2020-05-08 | 腾讯科技(深圳)有限公司 | Image processing method, device, server and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105678811B (en) | A kind of human body anomaly detection method based on motion detection | |
US8682036B2 (en) | System and method for street-parking-vehicle identification through license plate capturing | |
KR101935399B1 (en) | Wide Area Multi-Object Monitoring System Based on Deep Neural Network Algorithm | |
RU2484531C2 (en) | Apparatus for processing video information of security alarm system | |
US7995843B2 (en) | Monitoring device which monitors moving objects | |
CN104378582B (en) | A kind of intelligent video analysis system and method cruised based on Pan/Tilt/Zoom camera | |
US7778445B2 (en) | Method and system for the detection of removed objects in video images | |
US8320613B2 (en) | Detecting and tracking targets in images based on estimated target geometry | |
CN101751744B (en) | Detection and early warning method of smoke | |
CN100545867C (en) | Aerial shooting traffic video frequency vehicle rapid checking method | |
CN110276284A (en) | Flame identification method, device, equipment and storage medium based on video quality assessment | |
EP0878965A2 (en) | Method for tracking entering object and apparatus for tracking and monitoring entering object | |
CN110390229B (en) | Face picture screening method and device, electronic equipment and storage medium | |
CN109711322A (en) | A kind of people's vehicle separation method based on RFCN | |
CN104966304A (en) | Kalman filtering and nonparametric background model-based multi-target detection tracking method | |
CN108596157B (en) | Crowd disturbance scene detection method and system based on motion detection | |
KR101030257B1 (en) | Method and System for Vision-Based People Counting in CCTV | |
KR20110035662A (en) | Intelligent image search method and system using surveillance camera | |
CN108388845A (en) | Method for checking object and system | |
JP2001357402A (en) | Method and device for detecting vehicle | |
KR20160044757A (en) | Method and Apparatus for Analyzing Traffic Situation | |
CN110175553B (en) | Method and device for establishing feature library based on gait recognition and face recognition | |
KR101840042B1 (en) | Multi-Imaginary Fence Line Setting Method and Trespassing Sensing System | |
Hardas et al. | Moving object detection using background subtraction shadow removal and post processing | |
CN108376407A (en) | Hot-zone object aggregation detection method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |