CN105930807A - Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis - Google Patents
Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis Download PDFInfo
- Publication number
- CN105930807A CN105930807A CN201610261287.2A CN201610261287A CN105930807A CN 105930807 A CN105930807 A CN 105930807A CN 201610261287 A CN201610261287 A CN 201610261287A CN 105930807 A CN105930807 A CN 105930807A
- Authority
- CN
- China
- Prior art keywords
- cloud terrace
- pan
- template
- criterion
- tilt holder
- 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
- 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
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/44—Event detection
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The invention discloses a pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis. The pan-tilt holder movement and return detection algorithm mainly adopts the template matching rule, which means that template matching is inevitably invalid within a period of time in case that the pan-tilt holder moves or in case of zooming, and template matching succeeds again within a period of time if the pan-tilt holder returns. The algorithm comprises two main steps of: 1) searching and positioning an appropriate template from background images, selecting the matching rule and calculating the offset degree; 2) judging that the pan-tilt holder moves or returns. The algorithm of the invention realizes accurate detection of the condition of the pan-tilt holder movement, thereby improving the recognition rate and the stability of systems under the condition of moving a camera via a pan-tilt holder.
Description
Technical field
The present invention relates to the technical field of traffic incidents detection, refer in particular to a kind of special based on video sequence image
The The Cloud Terrace levying analysis moves and reset detection algorithm.
Background technology
Along with the social and development of technology and the demand of Operation and Management of Expressway, traffic incident detecting system
More and more has complexity.Whole nation highway is numerous, every highway from operation management to condition of road surface base
This has its particularity;Meanwhile, socio-economic development rapidly and highway Toll Free festivals or holidays prevailing,
Vehicle on highway gets more and more, block up, stop, the traffic events such as drive in the wrong direction of common occurrence, to society
The impact that economic benefit and resident trip cause is increasing, and this just proposes more to traffic incident detecting system
High requirement.On the one hand detecting system will be under normal circumstances to common event, such as parking, pedestrian, inverse
Go, block up etc. judges accurately, to be on the other hand also identified event in particular cases and to report
Alert.The highway of China great majority are provided with the camera gun of The Cloud Terrace, and this requires that system moves at The Cloud Terrace
Under conditions of can also be properly functioning, occur without the error detection of event.For having the digital camera of presetting bit
For machine, the signal that reliable method sends when being by obtaining camera gun The Cloud Terrace and moving judges The Cloud Terrace
Whether be moved, The Cloud Terrace is when mobile, it is meant that operator are being imaged by shake in the way of artificial
Machine carries out direct surveillance, and at this moment system stops carrying out event detection, automatically into personal monitoring's pattern.And one
Denier video camera stops mobile, and at this moment video camera is according to the camera preset bit data set, automatically to video camera
Resetting, after video camera returns to presetting bit, system proceeds by monitoring automatically.But this method needs
Know the interface of camera apparatus, do not possess versatility and actual operability.For there is no the mould of presetting bit
Intending video camera, this method cannot use.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, it is provided that a kind of based on video sequence image feature
The The Cloud Terrace analyzed moves and reset detection algorithm, possesses versatility and actual operability.
For achieving the above object, technical scheme provided by the present invention is: divide based on video sequence image feature
The The Cloud Terrace of analysis moves and reset detection algorithm, and main employing template matching criterion, if i.e. The Cloud Terrace is moved
Or zoom, then within a period of time, template matching will necessarily lose efficacy, in like manner, if The Cloud Terrace resets, then one
In the section time, template can the match is successful again;Specifically include following steps:
1) select matching criterior and calculate drift rate
The template matching criterion used has minimum mean square error criterion MSE, minimum average B configuration absolute value criterion
MAD, tri-kinds of modes of Normalized Cross Correlation Function criterion MCCF, select minimum mean square error criterion MSE
The amount of calculation of system can be reduced, answer for the pervasive of criterion, when detecting The Cloud Terrace and whether moving simultaneously
It is also adopted by side-play amount offset (i) to weigh, then re-defines threshold value M as the standard judged, threshold value M
Definition according to the degree of video jitter depending on;
2) judge that The Cloud Terrace moves and resets
The system detection mode when The Cloud Terrace changes is defaulted as not carrying out any detection, until The Cloud Terrace resets again
Carrying out flow and event detection after initializing again, its concrete determination step is as follows:
2.1) in background image, search positions suitable template, calculates and judge the side-play amount of each template
Whether offset (i) exceedes threshold value M;If it did not, directly return, if above, then carrying out below step
2.2);
2.2) judge whether unmatched time t meets or exceeds time threshold T, if it did not, directly return,
If met or exceeded, then judge that The Cloud Terrace has changed and carried out below step 2.3);
2.3) if all templates are mated again, and continue for some time J, then judge that The Cloud Terrace resets, return
To previous step 2.1).
2.4) if the time that The Cloud Terrace moves exceedes threshold k, then system re-initialization, can be at the current back of the body
Re-search on scape positioning new template, delimit new detection region;If now The Cloud Terrace is also in mobile shape
State, then return to previous step 2.1) continue to judge whether The Cloud Terrace is moved.
The present invention compared with prior art, has the advantage that and beneficial effect:
The signal that prior art sends when being by obtaining camera gun The Cloud Terrace and moving mostly judges whether The Cloud Terrace is sent out
Raw mobile, but this method is it is to be appreciated that the interface of capture apparatus, does not possess versatility and reality is operable
Property.Inventive algorithm accurate can verify the situation that The Cloud Terrace moves, thus raising is moved at The Cloud Terrace and taken the photograph
The discrimination of system and stability under the conditions of camera.
Accompanying drawing explanation
Fig. 1 is the flow chart that the present invention judges that The Cloud Terrace moves and resets.
Fig. 2 is that Shenzhen-Guangzhou superhighway band The Cloud Terrace moves the surface chart normally detected with the video resetted.
Fig. 3 is to move Shenzhen-Guangzhou superhighway band The Cloud Terrace to carry out The Cloud Terrace with the video that resets and move the interface of detection
Figure.
Fig. 4 is the interface that the video moving Shenzhen-Guangzhou superhighway band The Cloud Terrace and resetting carries out The Cloud Terrace reset detection
Figure.
Detailed description of the invention
Below in conjunction with specific embodiment, the invention will be further described.
General when carrying out event detection, a detection zone can be defined during system initialization on image, only
Vehicle and traffic events system in the detection just detect, to avoid the most extraneous interference.But
When the The Cloud Terrace of monitoring moves or during zoom, and the detection zone of definition can lose efficacy originally, and can affect and be
System is some parameters set by detection event when running.In this case it is necessary to system can be known in time
Do not go out The Cloud Terrace and have occurred and that movement, then redefine detection zone and some detection function.And it is of the present invention
The Cloud Terrace based on video sequence image feature analysis move and reset detection algorithm, mainly use template matching
Criterion, if i.e. The Cloud Terrace is moved or zoom, then within a period of time, template matching will necessarily lose efficacy, with
Reason, if The Cloud Terrace resets, then within a period of time, template can the match is successful again.Specifically include following steps:
1) select matching criterior and calculate drift rate
The template matching criterion used has minimum mean square error criterion MSE, minimum average B configuration absolute value criterion
MAD, tri-kinds of modes of Normalized Cross Correlation Function criterion MCCF, select minimum mean square error criterion MSE
The amount of calculation of system can be reduced, answer for the pervasive of criterion, when detecting The Cloud Terrace and whether moving simultaneously
It is also adopted by side-play amount offset (i) to weigh, then re-defines threshold value M as the standard judged, threshold value M
Definition according to the degree of video jitter depending on;
2) judge that The Cloud Terrace moves and resets
The mode that The Cloud Terrace changes is various, it may be possible to The Cloud Terrace translation scan, it may be possible to zoom,
It could also be possible that scanning etc. after zoom.It is true that be to be difficult to judge that The Cloud Terrace is by the way of image procossing
Doing which kind of action.Therefore, the system detection mode when The Cloud Terrace changes is defaulted as not carrying out any detection,
Until The Cloud Terrace resets after re-starting initialization carries out flow and event detection again.As shown in Figure 1, it is determined that
Step is as follows:
2.1) in background image, search positions suitable template, calculates and judge the side-play amount of each template
Whether offset (i) exceedes threshold value M;If it did not, directly return, if above, then carrying out below step
2.2);
2.2) judge whether unmatched time t meets or exceeds time threshold T, if it did not, directly return,
If met or exceeded, then judge that The Cloud Terrace has changed and carried out below step 2.3);
2.3) if all templates are mated again, and continue for some time J, then judge that The Cloud Terrace resets, return
To previous step 2.1).
2.4) if the time that The Cloud Terrace moves exceedes threshold k, then system re-initialization, can be at the current back of the body
Re-search on scape positioning new template, delimit new detection region;If now The Cloud Terrace is also in mobile shape
State, then return to previous step 2.1) continue to judge whether The Cloud Terrace is moved.This step exist reason be as
Really the The Cloud Terrace long period does not returns to initial position, and the template as coupling likely causes mould because of renewal
Plate lost efficacy.
Moving and the effect of reset detection algorithm for further illustrating the above-mentioned The Cloud Terrace of the present embodiment, this example is to extensively
Deep highway band The Cloud Terrace moves and carries out effect detection with the video resetted, the result obtained such as Fig. 2 to Fig. 4
Shown in.Wherein, in Fig. 2, warning message list display initialization completes, and starts detection, shows that system is
Generate background image, and search for and located suitable template, and carry out traffic incidents detection;Fig. 3
Middle warning message list display The Cloud Terrace moves, and as can be seen from the figure camera review has had the biggest change
Change, show successfully to be detected that by template matching The Cloud Terrace moves, do not carry out the detection of traffic events;Fig. 4
Middle warning message list display The Cloud Terrace moves and fixing beginning of The Cloud Terrace initializes, comparison diagram 4 and Fig. 2, permissible
Find out that The Cloud Terrace has resetted after being moved through, show by template matching successfully detect The Cloud Terrace move after answer
Position, and system re-initialization carries out traffic incidents detection.
Embodiment described above is only the preferred embodiments of the invention, not limits the enforcement model of the present invention with this
Enclose, therefore the change that all shapes according to the present invention, principle are made, all should contain within the scope of the present invention.
Claims (1)
1. The Cloud Terrace based on video sequence image feature analysis moves and reset detection algorithm, it is characterised in that:
Main employing template matching criterion, if i.e. The Cloud Terrace is moved or zoom, then template within a period of time
Joining and will necessarily lose efficacy, in like manner, if The Cloud Terrace resets, then within a period of time, template can the match is successful again;
Specifically include following steps:
1) select matching criterior and calculate drift rate
The template matching criterion used has minimum mean square error criterion MSE, minimum average B configuration absolute value criterion
MAD, tri-kinds of modes of Normalized Cross Correlation Function criterion MCCF, select minimum mean square error criterion MSE
The amount of calculation of system can be reduced, answer for the pervasive of criterion, when detecting The Cloud Terrace and whether moving simultaneously
It is also adopted by side-play amount offset (i) to weigh, then re-defines threshold value M as the standard judged, threshold value M
Definition according to the degree of video jitter depending on;
2) judge that The Cloud Terrace moves and resets
The system detection mode when The Cloud Terrace changes is defaulted as not carrying out any detection, until The Cloud Terrace resets again
Carrying out flow and event detection after initializing again, its concrete determination step is as follows:
2.1) template needed for search positions in background image, calculates and judges the side-play amount of each template
Whether offset (i) exceedes threshold value M;If it did not, directly return, if above, then carrying out below step
2.2);
2.2) whether judge templet unmatched time t meets or exceeds time threshold T, if it did not, directly
Returning, if met or exceeded, then judging that The Cloud Terrace has changed and carried out below step 2.3);
2.3) if all templates are mated again, and persistently set time J, then judge that The Cloud Terrace resets, returns
To previous step 2.1);
2.4) if the time that The Cloud Terrace moves exceedes threshold k, then system re-initialization, can be at the current back of the body
Re-search on scape positioning new template, delimit new detection region;If now The Cloud Terrace is also in mobile shape
State, then return to previous step 2.1) continue to judge whether The Cloud Terrace is moved.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610261287.2A CN105930807A (en) | 2016-04-25 | 2016-04-25 | Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610261287.2A CN105930807A (en) | 2016-04-25 | 2016-04-25 | Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105930807A true CN105930807A (en) | 2016-09-07 |
Family
ID=56836028
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610261287.2A Pending CN105930807A (en) | 2016-04-25 | 2016-04-25 | Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105930807A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108932732A (en) * | 2018-06-21 | 2018-12-04 | 浙江大华技术股份有限公司 | A kind of method and device obtaining monitoring object data information |
CN114882393A (en) * | 2022-03-29 | 2022-08-09 | 华南理工大学 | Road reverse running and traffic accident event detection method based on target detection |
-
2016
- 2016-04-25 CN CN201610261287.2A patent/CN105930807A/en active Pending
Non-Patent Citations (1)
Title |
---|
林浪桥: "基于视频图像的交通事件自动检测系统关键算法研究", 《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 |
---|---|---|---|---|
CN108932732A (en) * | 2018-06-21 | 2018-12-04 | 浙江大华技术股份有限公司 | A kind of method and device obtaining monitoring object data information |
CN114882393A (en) * | 2022-03-29 | 2022-08-09 | 华南理工大学 | Road reverse running and traffic accident event detection method based on target detection |
CN114882393B (en) * | 2022-03-29 | 2023-04-07 | 华南理工大学 | Road reverse running and traffic accident event detection method based on target detection |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Grassi et al. | Parkmaster: An in-vehicle, edge-based video analytics service for detecting open parking spaces in urban environments | |
US20140348390A1 (en) | Method and apparatus for detecting traffic monitoring video | |
JP2009510827A (en) | Motion detection device | |
KR101569919B1 (en) | Apparatus and method for estimating the location of the vehicle | |
JP2001216519A (en) | Traffic monitor device | |
EP2824416A1 (en) | Distance calculation device and distance calculation method | |
CN105629628A (en) | Automatic focusing method and device | |
KR101032495B1 (en) | Multi-function detecting system of illegally parked vehicles using digital ptz technique and method of detecting thereof | |
US11328441B2 (en) | Information processing device and information processing system | |
WO2014160027A1 (en) | Roadway sensing systems | |
KR101035761B1 (en) | Method of processing image for recognizing a lane and the system thereof | |
KR101974105B1 (en) | Photographing system and method for increasing recognition rate of vehicle number | |
CN110648539A (en) | In-vehicle device and control method | |
CN108922172B (en) | Road congestion monitoring system based on vehicle characteristic matrix sequence change analysis | |
CN111369824B (en) | Parking guiding method and system based on image recognition and positioning | |
KR102306789B1 (en) | License Plate Recognition Method and Apparatus for roads | |
JP2007316856A (en) | Traveling object detecting device, computer program, and traveling object detection method | |
CN105930807A (en) | Pan-tilt holder movement and return detection algorithm based on video sequence image feature analysis | |
CN112351190A (en) | Digital twin monitoring system and method | |
Chen et al. | Vision-based road bump detection using a front-mounted car camcorder | |
KR101568683B1 (en) | Method to estimate the distance between vehicles using image alalysis and apparatus therefor | |
KR101161557B1 (en) | The apparatus and method of moving object tracking with shadow removal moudule in camera position and time | |
KR102484691B1 (en) | Vehicle detection system and vehicle detection method using stereo camera and radar | |
Al-Kadi et al. | Road scene analysis for determination of road traffic density | |
JP4144300B2 (en) | Plane estimation method and object detection apparatus using stereo image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160907 |