CN102903123A - Self-adapting background subtracting method based on Gaussian mixture background reconstruction - Google Patents
Self-adapting background subtracting method based on Gaussian mixture background reconstruction Download PDFInfo
- Publication number
- CN102903123A CN102903123A CN2012103373298A CN201210337329A CN102903123A CN 102903123 A CN102903123 A CN 102903123A CN 2012103373298 A CN2012103373298 A CN 2012103373298A CN 201210337329 A CN201210337329 A CN 201210337329A CN 102903123 A CN102903123 A CN 102903123A
- Authority
- CN
- China
- Prior art keywords
- background
- frame
- utilize
- histogram
- video
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a self-adapting background subtracting method based on Gaussian mixture background reconstruction. The method comprises steps of collecting video frames, extracting an initial background frame, and initializing a background model; establishing Poisson distribution of a noise model by using R (red), G (green) and B (blue) component differences of the current frame and the background frame, counting a histogram of the Poisson distribution, and calculating relative variances for the obtained histogram; and ranking the obtained relative variances, finding a maximum value to serve as a segmentation threshold of R, G and B components of the current frame, conducting binaryzation and obtaining a foreground frame. The method is adapted to dynamic background perturbation and light change effect, moving objects in a video can be detected in real time, and the method is good in robustness.
Description
Technical field
The present invention relates to computer vision technique, particularly relate to a kind of adaptive background based on the mixed Gaussian background reconstruction and deduct method.
Background technology
Along with the development of science and technology and the people continuous enhancing to security precautions, have the video monitoring system of new generation of intellectual analysis function, beginning has begun to penetrate in the middle of our daily life in the very positive effect of security monitoring field performance.
Intelligent video monitoring refers in the situation that do not need human intervention, utilize the computer vision analysis method that video sequence is carried out automatic analysis, realize moving object detection, classification, identification, tracking etc., and on this basis, by predefined rule the behavior of target is analyzed, thereby provided for taking further measures with reference to (such as enter automatic alarm when setting up defences the district at object).Wherein, the purpose of motion detection is by the analysis to the monitor video image sequence, determines to have or not in the monitoring scene moving target, and then moving region (also claiming foreground area) is extracted from detected image.It is the basic premise that carries out the subsequent treatment such as motion target tracking, classification and identification that the moving region is cut apart accurately and effectively.
At present, using more extensive method for testing motion is the background subtraction method.The background subtraction method is at first set up background model for background image, then by comparing the difference of detected image and background model, judges whether there is moving target in the scene.Can background model correctly reflect real-time background effectively, can directly affect the accuracy of motion detection.But owing in the scene of complexity, usually can have the interference (such as illumination variation, video camera slight jitter, dynamic background element etc.) of various extraneous factors, have a strong impact on the precision of moving Object Segmentation.
In order to improve the accuracy of detection of moving target, existing adaptive background based on the mixed Gaussian background reconstruction deducts method provides a kind of adaptive background difference method that can adapt to complex background and illumination variation, the method is at first obtained frame of video, utilizes the mixed Gauss model method to carry out the background modeling initialization; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame.The method has very high real-time and robustness, and having solved effectively that the moving object detection precision is vulnerable to is the problem of impact of dynamic background disturbance and illumination variation, reaches purpose of design.
Prior art has a kind of moving target detecting method based on mixed Gauss model, just on the docket.At first gather frame of video; Extract the initial background frame, carry out the initialization of background model, set up the HSV component Model; Present frame gets the prospect frame with background frames phase difference; Resulting prospect frame is carried out binary conversion treatment; According to described prospect frame, introduce and upgrade weights, average and the variance that the factor is upgraded mixture Gaussian background model; Utilize the Jeffrey value to determine whether the moving target prospect; Utilize the mixed Gaussian shadow model to remove the shade of described moving target prospect.
But above technology has just proposed to carry out mixed Gauss model in the HSV color space carries out background modeling, present frame is not carried out self-adaption binaryzation.Since from the prospect frame that the mixed Gauss model modeling method obtains, relatively more responsive to the illumination variation under the complex background, so have a strong impact on the precision of moving Object Segmentation.
Summary of the invention
The invention provides a kind of adaptive background based on the mixed Gaussian background reconstruction and deduct method; the method can realize that the moving target under the complex scene accurately detects; solved in the scene of complexity; usually can there be the interference (such as illumination variation, video camera slight jitter, dynamic background element etc.) of various extraneous factors, has a strong impact on the problem of the precision of moving Object Segmentation.
To achieve these goals, the present invention includes following technical characterictic: comprise and at first obtain frame of video, utilize the mixed Gauss model method to carry out the background modeling initialization; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame; Described from obtaining frame of video, utilize mixed Gauss model to carry out the modeling initialization; Described noise model Poisson distribution is that R, the G that utilizes present frame and background frames, the difference value histogram of B component are set up, and adds up its relevant variance, utilizes maximum relevant variance that present frame is carried out binaryzation, obtains the prospect frame.
Compare with existing method, the present invention proposes and at first obtain frame of video, utilize the mixed Gauss model method to carry out the background modeling initialization; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame.The method has very high real-time and robustness, and having solved effectively that the moving object detection precision is vulnerable to is the problem of impact of dynamic background disturbance and illumination variation.
Description of drawings
Figure is overview flow chart of the present invention.
Embodiment
The present invention has designed a kind of adaptive background based on the mixed Gaussian background reconstruction and has deducted method; the method can realize that the moving target under the complex scene accurately detects; solved in the scene of complexity; usually can there be the interference (such as illumination variation, video camera slight jitter, dynamic background element etc.) of various extraneous factors, has a strong impact on the problem of the precision of moving Object Segmentation.
As shown in drawings, the method process flow diagram comprises the collection frame of video, utilizes mixed Gauss model to carry out background modeling; Then, add up respectively the difference value histogram of present frame and background frames, and set up the noise model Poisson distribution, it is calculated relevant variance, seek out its maximal value; At last, utilize its relevant variance maximal value respectively R, G, B component to be carried out binaryzation, obtain the prospect frame.
Specific implementation is: comprise the collection frame of video, and extract the initial background frame, carry out the initialization of background model; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame.
Described from obtaining frame of video, utilize mixed Gauss model to carry out the modeling initialization;
Described noise model Poisson distribution is that R, the G that utilizes present frame and background frames, the difference value histogram of B component are set up, and adds up its relevant variance, utilizes maximum relevant variance that present frame is carried out binaryzation, obtains the prospect frame.
As seen by above-mentioned, specific embodiment of the present invention is for to detect the moving target in the complex scene.Further, by the collection frame of video, and extract the initial background frame, carry out the initialization of background model; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame.Can realize that the moving target under the complex scene accurately detects; solved in the scene of complexity; usually can there be the interference (such as illumination variation, video camera slight jitter, dynamic background element etc.) of various extraneous factors, has a strong impact on the problem of the precision of moving Object Segmentation.
Therefore; easily understand, the above is preferred embodiment of the present invention only, is not be used to limiting spirit of the present invention and protection domain; the equivalent variations that any those of ordinary skill in the art make or replacement all should be considered as being encompassed within protection scope of the present invention.
Claims (3)
1. the adaptive background based on the mixed Gaussian background reconstruction deducts method, it is characterized in that: comprise and at first obtain frame of video, utilize the mixed Gauss model method to carry out the background modeling initialization; Then utilize R, G, the B component difference of present frame and background frames to set up the noise model Poisson distribution, and add up its histogram, to the relevant variance of histogram calculation of gained; At last, the relevant variance of gained is sorted, seek out maximal value, as R, the G of present frame, the segmentation threshold of B component, carry out binaryzation, obtain the prospect frame.
2. the adaptive background based on the mixed Gaussian background reconstruction according to claim 1 deducts method, it is characterized in that: described from obtaining frame of video, and utilize mixed Gauss model to carry out the modeling initialization.
3. the adaptive background based on the mixed Gaussian background reconstruction according to claim 1 deducts method, it is characterized in that: be that the difference value histogram of the R, the G that utilize present frame and background frames, B component is set up according to described noise model Poisson distribution, add up its relevant variance, utilize maximum relevant variance that present frame is carried out binaryzation, obtain the prospect frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103373298A CN102903123A (en) | 2012-09-08 | 2012-09-08 | Self-adapting background subtracting method based on Gaussian mixture background reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012103373298A CN102903123A (en) | 2012-09-08 | 2012-09-08 | Self-adapting background subtracting method based on Gaussian mixture background reconstruction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102903123A true CN102903123A (en) | 2013-01-30 |
Family
ID=47575336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012103373298A Pending CN102903123A (en) | 2012-09-08 | 2012-09-08 | Self-adapting background subtracting method based on Gaussian mixture background reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102903123A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113012137A (en) * | 2021-03-24 | 2021-06-22 | 滁州惠科光电科技有限公司 | Panel defect inspection method, system, terminal device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101241082A (en) * | 2008-03-14 | 2008-08-13 | 东华大学 | Nonwoven fibric fibre orientation distribution measuring systems and method |
| CN101489034A (en) * | 2008-12-19 | 2009-07-22 | 四川虹微技术有限公司 | Method for video image noise estimation and elimination |
| CN101763505A (en) * | 2009-12-29 | 2010-06-30 | 重庆大学 | Vehicle license character feature extracting and classifying method based on projection symmetry |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
-
2012
- 2012-09-08 CN CN2012103373298A patent/CN102903123A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101241082A (en) * | 2008-03-14 | 2008-08-13 | 东华大学 | Nonwoven fibric fibre orientation distribution measuring systems and method |
| CN101489034A (en) * | 2008-12-19 | 2009-07-22 | 四川虹微技术有限公司 | Method for video image noise estimation and elimination |
| CN101763505A (en) * | 2009-12-29 | 2010-06-30 | 重庆大学 | Vehicle license character feature extracting and classifying method based on projection symmetry |
| CN102568005A (en) * | 2011-12-28 | 2012-07-11 | 江苏大学 | Moving object detection method based on Gaussian mixture model |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113012137A (en) * | 2021-03-24 | 2021-06-22 | 滁州惠科光电科技有限公司 | Panel defect inspection method, system, terminal device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109117827B (en) | Video-based method for automatically identifying wearing state of work clothes and work cap and alarm system | |
| CN102307274B (en) | Motion detection method based on edge detection and frame difference | |
| CN109076198B (en) | Video-based object tracking occlusion detection system, method and equipment | |
| Jung | Efficient background subtraction and shadow removal for monochromatic video sequences | |
| CN109446942A (en) | Method for tracking target, device and system | |
| CN105940430B (en) | Personnel's method of counting and its device | |
| EP2801078B1 (en) | Context aware moving object detection | |
| US9792505B2 (en) | Video monitoring method, video monitoring system and computer program product | |
| CN103246896B (en) | A kind of real-time detection and tracking method of robustness vehicle | |
| CN104318263A (en) | Real-time high-precision people stream counting method | |
| CN105404847A (en) | Real-time detection method for object left behind | |
| CN103400117B (en) | A kind of personnel in the pit's positioning and tracing method based on computer vision | |
| CN105554462A (en) | Remnant detection method | |
| CN103150736A (en) | Camera motion detecting method based on video monitoring | |
| CN103593679A (en) | Visual human-hand tracking method based on online machine learning | |
| CN105513053A (en) | Background modeling method for video analysis | |
| CN103942812A (en) | Moving object detection method based on Gaussian mixture and edge detection | |
| CN104616006A (en) | Surveillance video oriented bearded face detection method | |
| CN105469038A (en) | Safety cap video detection method for electric power switching station | |
| CN104090658A (en) | Human body image detection method based on HOG features | |
| CN103810718A (en) | Method and device for detection of violently moving target | |
| CN104199009B (en) | A kind of radar image clutter suppression method based on time domain specification | |
| CN103400395A (en) | Light stream tracking method based on HAAR feature detection | |
| CN103903020A (en) | Method and device for fire disaster image identification based on Code Books | |
| CN107274382A (en) | A kind of state identification method, device and the electronic equipment of hard pressing plate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130130 |