CN105205833A - Moving object detection method and device based on space-time background model - Google Patents
Moving object detection method and device based on space-time background model Download PDFInfo
- Publication number
- CN105205833A CN105205833A CN201510586019.3A CN201510586019A CN105205833A CN 105205833 A CN105205833 A CN 105205833A CN 201510586019 A CN201510586019 A CN 201510586019A CN 105205833 A CN105205833 A CN 105205833A
- Authority
- CN
- China
- Prior art keywords
- background
- space
- time
- background model
- model
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
Abstract
The invention provides a moving object detection method based on a space-time background model. The method comprises the steps that 1, a time background model is obtained by collecting image modeling; 2, initial moving objects are obtained; 3, a background region of grown initial moving objects is obtained; 4, the space background model is obtained; 5, whether space-time model distribution is conformed or not is judged; 6, the time background model is updated. The invention further provides a moving object detection device based on the space-time background model, the moving object detection device comprises an image acquisition module, a time background modeling module, a moving object detection module, a space background modeling module, a moving object confirmation module and a background model updating module. The invention provides a space background modeling concept, it is considered that foreground objects and surrounding backgrounds have certain differences on features such as color, if the initial moving objects are in accordance with the surrounding backgrounds on the color feature, it is considered to be false detection, and the accuracy of moving object detection is improved.
Description
Technical field
The present invention relates to field of intelligent video surveillance, particularly relate to a kind of moving target detecting method based on time-and-space background model and device.
Background technology
The basic task of moving object detection is extracted from image background by moving target from sequence image, thus obtain the movable information of target, the work such as follow-up Images Classification, target following are had great importance, can Simplified analysis work in larger degree.Due in actual monitored scene, background is not often completely static, but the moment is in change, such as: the change of the rocking of the rule of the change of weather, the small of background, illumination, incorporate the target etc. in background gradually, make background modeling become a Focal point and difficult point problem of moving object detection.
Background subtraction, because speed is fast, accuracy is high, can extract the reasons such as complete objective contour, receives and pays close attention to widely and apply.Wherein, background modeling method is the core that background subtraction detects moving target.At present, typical background modeling method has: the method such as mixed Gaussian background modeling method, codebook method, nonparametric background modeling method, ViBe, PBAS.These traditional background modeling methods utilize the statistical nature of pixel in time series to carry out modeling, and underuse spatial information, therefore, in target detection process, often there is more flase drop, and in context update process, easily prospect is updated in background, and the speed of real context update is fast not.
Summary of the invention
Technical matters to be solved by this invention is to provide a kind of moving target detecting method based on time-and-space background model, can be applicable to the target detection under complex scene, effectively can adapt to the change of scene.
The present invention solves the problems of the technologies described above adopted technical scheme:
Based on a moving target detecting method for time-and-space background model, comprise the following steps:
Step 1: carry out background modeling according to the some two field pictures collected and obtain time background model,
Image acquisition is extracted from pending monitor video by image capture module;
Step 2: present frame and background model are asked poor, after binaryzation, obtains initial motion target, and initial motion target is the foreground picture after binaryzation;
Step 3: for each initial motion target, carry out region growing to background area, obtains the background area after growing;
Described growth refers to by the eight connectivity neighborhood of initial motion target to outgrowth, differ with the gray-scale value of moving target within the specific limits think homogeneous region, can expand, until cannot expand, thus obtain this background area belonging to initial motion target;
Step 4: use mixed Gauss model to describe above-mentioned background region, obtain space background model;
Step 5: judge whether initial motion target meets space background model profile, if met, then illustrates it is prospect; If do not met, after denoising, obtain final moving target;
Step 6: update time background model.
Time background model is used to obtain initial motion target, and update time, the effect of background model was: eliminate illumination gradual change to the impact of extracting foreground target.
While employing technique scheme, the present invention can also adopt or combine and adopt following further technical scheme:
The modeling process of described step 1 specifically comprises the following steps:
Step 1.1: gather some two field pictures, be converted to gray level image, supposes that the grey scale pixel value being positioned at point (x, y) is in the sequence: Y={y
1(x, y), y
2(x, y) ..., y
n(x, y) };
Step 1.2: the average and the standard deviation that calculate these sampled values (Y value that in step 1.1, each is put), setting threshold value t1, if standard deviation is less than threshold value t1, then being gathered is a class; If standard deviation is not less than threshold value t1, then by these sampled values with being divided into two classes, respectively computation of mean values and standard deviation.
Described step 2 specifically comprises the following steps:
Step 2.1: current frame image is converted to gray level image, deducts average corresponding to model respectively by grey scale pixel value, if corresponding difference is within 3 times of standard deviations, be then background, otherwise be prospect;
Step 2.2: by extracting connected region, sorts out foreground pixel and obtains initial motion target.
Growth in described step 3 refers to from the eight connectivity neighborhood of each initial motion object and background and grows to background area, obtains the background area after corresponding region growing.
Described step 6 specifically comprises the following steps:
Step 6.1: to motion target area, adopts less turnover rate to upgrade;
Step 6.2: for background area, adopts larger turnover rate to upgrade;
Above-mentioned comparatively large and less be step 6.1 and step 6.2 comparing each other.
Another technical matters to be solved by this invention is to provide a kind of moving object detection device based on time-and-space background model, described object detecting device comprises image capture module, time background MBM, moving object detection module, space background MBM, moving target confirms module and background model update module, described image capture module is used for acquisition monitoring image, described time background modeling MBM is used for background model Time Created in step 1, described moving object detection module is used for from time background model, extracting initial moving target in step 2, described space background MBM is used for calculating the corresponding background area of each initial target, and set up corresponding space background model, described moving target confirms that module is put into space background model for initial motion target step 2 extracted and verified, obtain final moving target, described background model update module is used for upgrading background and moving target,
Described image capture module connects watch-dog and gathers monitoring image wherein, described image capture module is connected to time background MBM and sends to it image information gathered, described time background MBM is connected to described moving object detection module and sends modeling information to it, described moving object detection model calling is to space background MBM and to its transmission initial motion target information, described space background MBM is connected to described moving target and confirms module and send space background modeling information and initial motion target information to it, described moving target to confirm that model calling sends final moving target information to background model update module to it, described background model update module upgrades moving target and background according to different turnover rates.
The invention has the beneficial effects as follows: a kind of moving target detecting method based on time-and-space background model of the present invention and device are integrated solutions, specifically there is following innovation: the concept 1, proposing space background modeling, think that the background of foreground object and surrounding has certain difference in the features such as color, if initial motion target is consistent with ambient background in the features such as color, then think flase drop, thus improve the accuracy rate of moving object detection.2, when background model update time, background area adopts very fast turnover rate, change of background is made to learn in model faster, and the motion target area after confirming adopts slower turnover rate, moving target static for a long time can be learnt in background gradually, turn improve the recall rate of the moving target stopped in short-term.
Accompanying drawing explanation
Fig. 1 is a kind of moving target detecting method process flow diagram based on time-and-space background model of the present invention.
Fig. 2 is a kind of moving object detection structure drawing of device based on time-and-space background model of the present invention.
Embodiment
Embodiment 1, a kind of moving target detecting method based on time-and-space background model, with reference to accompanying drawing 1.
Moving target detecting method of the present invention specifically comprises the following steps:
Step 1: carry out background modeling according to 3000 two field pictures collected and obtain time background model;
1. be converted to background image, suppose that the grey scale pixel value being positioned at point (x, y) is in the sequence: Y={y
1(x, y), y
2(x, y) ..., y
3000(x, y) };
2. average μ and the standard deviation δ of these sampled values and above-mentioned grey scale pixel value is calculated:
I in formula represents i-th sampled point,
If δ is <20, then using the time background model of this Gauss model as this pixel.If δ >=20, use K means Method, be divided into two classes, respectively computation of mean values and standard deviation, as time background model.
Step 2: present frame and background model are asked poor, after binaryzation, obtains initial motion target;
1. present frame is converted to gray level image, grey scale pixel value is deducted respectively average corresponding to model, if corresponding difference is within 3 times of standard deviations, be then background, otherwise be prospect;
R in formula represents binary image,
2. by extracting connected region, foreground pixel being sorted out and obtains initial motion target.
Step 3: for each initial motion target, carry out region growing to background area, obtains the background area after growing;
Concrete finger, grows from the eight connectivity neighborhood of each initial motion object and background, obtains the background area after corresponding region growing to background area;
Step 4: use mixed Gauss model to describe above-mentioned background region, obtain space background model;
Step 5: whether initial motion target meets space background model profile, the difference of the grey scale pixel value namely in initial motion target area and corresponding space background model average, being less than 3 times of standard deviations, is then background; Otherwise be prospect; Connected region in advance, removes the region that area is less than 20, obtains final moving target;
Step 6: use following formula background model update time:
μ
t＝(1-ρ)*μ
t-1+ρ*y
t
δ
t 2＝(1-ρ)*δ
t-1 2+ρ*(y
t-μ
t)
T(y
t-μ
t)，
In formula, ρ is turnover rate; T is the time, and yt represents the gray-scale value of current pixel, and μ t, δ t represent average and the standard deviation of this grey scale pixel value of t respectively, and μ t-1, δ t-1 represent average and the standard deviation of t-1 this grey scale pixel value of moment respectively, T representing matrix;
Motion target area, uses less turnover rate to upgrade, ρ=0.00001;
Background area, uses larger turnover rate to upgrade, ρ=0.0001.
Embodiment 2, a kind of moving object detection device based on time-and-space background model, with reference to accompanying drawing 2.
Moving object detection device of the present invention comprises image capture module 1, time background MBM 2, moving object detection module 3, space background MBM 4, moving target confirms module 5 and background model update module 6, described image capture module 1 is for acquisition monitoring image, described time background modeling MBM 2 is for background model Time Created in step 1, described moving object detection module 3 for extracting initial moving target in step 2 from time background model, described space background MBM 4 is for calculating the corresponding background area of each initial target, and set up corresponding space background model, described moving target confirms that module 5 is put into space background model for initial motion target step 2 extracted and verified, obtain final moving target, described background model update module 6 is for upgrading background and moving target,
Described image capture module 1 connects watch-dog and gathers monitoring image wherein, described image capture module 1 is connected to time background MBM 2 and sends to it image information gathered, described time background MBM 2 is connected to described moving object detection module 3 and sends modeling information to it, described moving object detection module 3 is connected to space background MBM 4 and sends initial motion target information to it, described space background MBM 4 is connected to described moving target and confirms module 5 and send space background modeling information and initial motion target information to it, described moving target confirms that module 5 is connected to background model update module 6 and sends final moving target information to it, described background model update module 6 upgrades moving target and background according to different turnover rates.
Claims (6)
1. based on a moving target detecting method for time-and-space background model, it is characterized in that: said method comprising the steps of:
Step 1: carry out background modeling according to the some two field pictures collected and obtain time background model;
Step 2: present frame and background model are asked poor, after binaryzation, obtains initial motion target;
Step 3: for each initial motion target, carry out region growing to background area, obtains the background area after growing;
Step 4: use mixed Gauss model to describe above-mentioned background region, obtain space background model;
Step 5: judge whether initial motion target meets space background model profile, if met, then illustrates it is background; If do not met, after denoising, obtain final moving target;
Step 6: update time background model.
2. a kind of moving target detecting method based on time-and-space background model as claimed in claim 1, is characterized in that: the modeling process of described step 1 specifically comprises the following steps:
Step 1.1: gather some two field pictures, be converted to gray level image, supposes that the grey scale pixel value being positioned at point (x, y) is in the sequence: Y={y
1(x, y), y
2(x, y) ..., y
n(x, y) };
Step 1.2: the average and the standard deviation that calculate these sampled values, setting threshold value t1, if standard deviation is less than threshold value t1, then being gathered is a class; If standard deviation is not less than threshold value t1, then by these sampled values with being divided into two classes, respectively computation of mean values and standard deviation.
3. a kind of moving target detecting method based on time-and-space background model as claimed in claim 1, is characterized in that: described step 2 specifically comprises the following steps:
Step 2.1: current frame image is converted to gray level image, deducts average corresponding to model respectively by grey scale pixel value, if corresponding difference is within 3 times of standard deviations, be then background, otherwise be prospect;
Step 2.2: by extracting connected region, sorts out foreground pixel and obtains initial motion target.
4. a kind of moving target detecting method based on time-and-space background model as claimed in claim 1, it is characterized in that: the growth in described step 3 refers to from the eight connectivity neighborhood of each initial motion object and background and grows to background area, obtains the background area after corresponding region growing.
5. a kind of moving target detecting method based on time-and-space background model as claimed in claim 1, is characterized in that: described step 6 specifically comprises the following steps:
Step 6.1: to motion target area, adopts less turnover rate to upgrade;
Step 6.2: for background area, adopts larger turnover rate to upgrade;
Above-mentioned comparatively large and less be step 6.1 and step 6.2 comparing each other.
6. the moving object detection device based on time-and-space background model, it is characterized in that: described object detecting device comprises image capture module, time background MBM, moving object detection module, space background MBM, moving target confirms module and background model update module, described image capture module is used for acquisition monitoring image, described time background modeling MBM is used for background model Time Created in step 1, described moving object detection module is used for from time background model, extracting initial moving target in step 2, described space background MBM is used for calculating the corresponding background area of each initial target, and set up corresponding space background model, described moving target confirms that module is put into space background model for initial motion target step 2 extracted and verified, obtain final moving target, described background model update module is used for upgrading background and moving target,
Described image capture module connects watch-dog and gathers monitoring image wherein, described image capture module is connected to time background MBM respectively and sends to it image information gathered, described time background MBM is connected to described moving object detection module and sends modeling information to it, described moving object detection model calling is to space background MBM and to its transmission initial motion target information, described space background MBM is connected to described moving target and confirms module and send space background modeling information and initial motion target information to it, described moving target to confirm that model calling sends final moving target information to background model update module to it, described background model update module upgrades moving target and background according to different turnover rates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510586019.3A CN105205833B (en) | 2015-09-15 | 2015-09-15 | A kind of moving target detecting method and device based on time-and-space background model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510586019.3A CN105205833B (en) | 2015-09-15 | 2015-09-15 | A kind of moving target detecting method and device based on time-and-space background model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105205833A true CN105205833A (en) | 2015-12-30 |
| CN105205833B CN105205833B (en) | 2018-03-16 |
Family
ID=54953493
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510586019.3A Active CN105205833B (en) | 2015-09-15 | 2015-09-15 | A kind of moving target detecting method and device based on time-and-space background model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105205833B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108174055A (en) * | 2017-12-29 | 2018-06-15 | 广东工业大学 | A kind of intelligent control method, system, equipment and storage medium |
| CN108921078A (en) * | 2018-06-27 | 2018-11-30 | 天津大学 | Time-and-space background based on video models sport foreground detection algorithm |
| CN109151376A (en) * | 2017-06-27 | 2019-01-04 | 南京东大智能化系统有限公司 | A kind of intelligent video analysis monitoring method |
| CN110099209A (en) * | 2018-01-30 | 2019-08-06 | 佳能株式会社 | Image processing apparatus, image processing method and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007323572A (en) * | 2006-06-05 | 2007-12-13 | Nec Corp | Object detector, object detection method, and object detection program |
| CN101325690A (en) * | 2007-06-12 | 2008-12-17 | 上海正电科技发展有限公司 | Method and system for detecting human flow analysis and crowd accumulation process of monitoring video flow |
| CN101635852A (en) * | 2009-08-26 | 2010-01-27 | 北京航空航天大学 | Method for detecting real-time moving object based on adaptive background modeling |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
| CN102568002A (en) * | 2011-12-20 | 2012-07-11 | 福建省华大数码科技有限公司 | Moving object detection algorithm based on fusion of texture pattern and movement pattern |
| KR20130063963A (en) * | 2011-12-07 | 2013-06-17 | 한국전자통신연구원 | Image processing method for detecting moving object |
-
2015
- 2015-09-15 CN CN201510586019.3A patent/CN105205833B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007323572A (en) * | 2006-06-05 | 2007-12-13 | Nec Corp | Object detector, object detection method, and object detection program |
| CN101325690A (en) * | 2007-06-12 | 2008-12-17 | 上海正电科技发展有限公司 | Method and system for detecting human flow analysis and crowd accumulation process of monitoring video flow |
| CN101635852A (en) * | 2009-08-26 | 2010-01-27 | 北京航空航天大学 | Method for detecting real-time moving object based on adaptive background modeling |
| KR20130063963A (en) * | 2011-12-07 | 2013-06-17 | 한국전자통신연구원 | Image processing method for detecting moving object |
| CN102568002A (en) * | 2011-12-20 | 2012-07-11 | 福建省华大数码科技有限公司 | Moving object detection algorithm based on fusion of texture pattern and movement pattern |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
Non-Patent Citations (2)
| Title |
|---|
| 文如泉 等: ""基于自适应混合高斯模型的运动目标检测"", 《萍乡高等专科学校校报》 * |
| 郭晓 等: ""基于时空的混合高斯背景建模的运动目标检测"", 《电视技术》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109151376A (en) * | 2017-06-27 | 2019-01-04 | 南京东大智能化系统有限公司 | A kind of intelligent video analysis monitoring method |
| CN109151376B (en) * | 2017-06-27 | 2021-09-28 | 南京东大智能化系统有限公司 | Intelligent video analysis monitoring method |
| CN108174055A (en) * | 2017-12-29 | 2018-06-15 | 广东工业大学 | A kind of intelligent control method, system, equipment and storage medium |
| CN108174055B (en) * | 2017-12-29 | 2020-11-13 | 广东工业大学 | Intelligent monitoring method, system, equipment and storage medium |
| CN110099209A (en) * | 2018-01-30 | 2019-08-06 | 佳能株式会社 | Image processing apparatus, image processing method and storage medium |
| CN110099209B (en) * | 2018-01-30 | 2021-03-23 | 佳能株式会社 | Image processing apparatus, image processing method, and storage medium |
| US11100655B2 (en) | 2018-01-30 | 2021-08-24 | Canon Kabushiki Kaisha | Image processing apparatus and image processing method for hiding a specific object in a captured image |
| CN108921078A (en) * | 2018-06-27 | 2018-11-30 | 天津大学 | Time-and-space background based on video models sport foreground detection algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105205833B (en) | 2018-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102568005B (en) | Moving object detection method based on Gaussian mixture model | |
| CN108664841B (en) | Dynamic and static target object identification method and device based on laser point cloud | |
| CN103258332B (en) | A kind of detection method of the moving target of resisting illumination variation | |
| CN107301376B (en) | Pedestrian detection method based on deep learning multi-layer stimulation | |
| CN105205833A (en) | Moving object detection method and device based on space-time background model | |
| CN111429418A (en) | Industrial part detection method based on YO L O v3 neural network | |
| CN102509085A (en) | Pig walking posture identification system and method based on outline invariant moment features | |
| CN110992365B (en) | Loss function based on image semantic segmentation and design method thereof | |
| CN102332167A (en) | Target detection method for vehicles and pedestrians in intelligent traffic monitoring | |
| CN103226712B (en) | A kind of remnant object detection method based on finite state machine | |
| CN104318588A (en) | Multi-video-camera target tracking method based on position perception and distinguish appearance model | |
| CN104252709B (en) | Multiple-target foreground detection method for look-down group-housed pigs in look-down state under complicated background | |
| Marcomini et al. | A comparison between background modelling methods for vehicle segmentation in highway traffic videos | |
| CN103473547A (en) | Vehicle target recognizing algorithm used for intelligent traffic detecting system | |
| CN110866943A (en) | Fish position tracking method for water quality monitoring | |
| CN101908214A (en) | Moving object detection method with background reconstruction based on neighborhood correlation | |
| CN103106796A (en) | Vehicle detection method and device of intelligent traffic surveillance and control system | |
| CN102054277B (en) | Method and system for detecting moving target, and video analysis system | |
| CN105469054A (en) | Model construction method of normal behaviors and detection method of abnormal behaviors | |
| CN108010050B (en) | Foreground detection method based on adaptive background updating and selective background updating | |
| CN101609552B (en) | Method for detecting characteristics of video object in finite complex background | |
| Wu et al. | Automatic gear sorting system based on monocular vision | |
| CN107871315B (en) | Video image motion detection method and device | |
| CN114049483A (en) | Target detection network self-supervision training method and device based on event camera | |
| CN102592125A (en) | Moving object detection method based on standard deviation characteristic |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |