CN102568005A - Moving object detection method based on Gaussian mixture model - Google Patents

Moving object detection method based on Gaussian mixture model Download PDF

Info

Publication number
CN102568005A
CN102568005A CN201110447408XA CN201110447408A CN102568005A CN 102568005 A CN102568005 A CN 102568005A CN 201110447408X A CN201110447408X A CN 201110447408XA CN 201110447408 A CN201110447408 A CN 201110447408A CN 102568005 A CN102568005 A CN 102568005A
Authority
CN
China
Prior art keywords
model
frame
background
shade
moving target
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
Application number
CN201110447408XA
Other languages
Chinese (zh)
Other versions
CN102568005B (en
Inventor
宋雪桦
谢桂莹
王昌达
吴问云
顾金
闫振
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Priority to CN201110447408.XA priority Critical patent/CN102568005B/en
Publication of CN102568005A publication Critical patent/CN102568005A/en
Application granted granted Critical
Publication of CN102568005B publication Critical patent/CN102568005B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Image Analysis (AREA)

Abstract

The invention discloses a moving object detection method based on a Gaussian mixture model. The method comprises the steps of collecting a video frame first, extracting an initial background frame, carrying out initialization of a background model, and establishing an HSV (hue, saturation and value) component model; obtaining a foreground frame according to the difference between a current frame and the background frame; carrying out binarization treatment to the obtained foreground frame; according to the foreground frame, introducing an updating factor to update the weight, the average and the variance of the Gaussian mixture model; judging whether the foreground of the moving object is obtained by utilizing the Jeffrey value; and removing the shade of the foreground of the moving object by utilizing the Gaussian mixture model. The method can be adapted to the disturbance of a dynamic background and the influence of illumination change, can realize updating in real time and effectively remove the shade, and has good robustness.

Description

A kind of moving target detecting method based on mixed Gauss model
Technical field
The invention belongs to the digital image processing techniques field, be specifically related to a kind of improved moving target detecting method based on mixed Gauss model.
Background technology
Moving object detection is meant from image sequence region of variation is split from background.Traditional moving object detection mainly contains optical flow method, consecutive frame difference method, background subtraction etc.Wherein, background subtraction is to study maximum a kind of moving object detection and dividing method at present, also is the method that a kind of quilt extensively applies to the intelligent video monitoring technology.Background subtraction comprises mainly that background model is set up, background model is upgraded, background difference and post-treating and other steps, and it has, and method is simple, the characteristic of the moving target that is easy to realize, can provide more complete, have advantage such as quite good detecting effectiveness.But background subtraction is based on fixing, static background, yet actual environment is always complicated and changeable, and dynamic background (slight jitter of illumination variation, background disturbance and camera etc.) can influence the sensitivity and the accuracy of moving object detection.The key of background subtraction is the foundation of background model, and mixed Gaussian background model algorithm complex is low, step is simple, can satisfy the requirement of dynamic real-time background.Extracting the initial background frame is key one ring in the background modeling, and the quality that background frames extracts directly has influence on the accuracy of moving object detection, is related to the reliability of further analysis video image information.In addition; When carrying out moving object detection; Because the shade characteristic and the background of moving target are completely different; And shade has identical motion feature with moving target, usually shade is interpreted as moving target by error when causing moving object detection, and the existence of shade makes the shape of cutting apart of object and object and color all receive very big influence.
Summary of the invention
The purpose of this invention is to provide a kind of improved moving target detecting method based on mixed Gauss model; This method is the basis with the mixed Gaussian background model, adopts to combine improve one's methods (the MEAMO method) of median method and mode method that the deficiency that it exists when the initial background frame extracts is improved; Combine the mixed Gaussian shadow model to carry out shadow Detection and removal simultaneously; This method can adapt to dynamic background disturbance and illumination variation influence, can upgrade in real time, effectively removes shade, has good robustness simultaneously.
Technical scheme of the present invention is: a kind of moving target detecting method based on mixed Gauss model, and its step comprises: S1: gather frame of video; S2: extract the initial background frame, carry out the initialization of background model, set up HSV component model; S3: present frame gets the prospect frame with background frames phase difference; S4: resulting prospect frame is carried out binary conversion treatment; S5:, introduce and upgrade weights, average and the variance that the factor is upgraded the mixed Gaussian background model according to the prospect frame of said step S3; S6: utilize the Jeffrey value to judge whether to be the moving target prospect; S7: utilize the mixed Gaussian shadow model to remove the shade of said moving target prospect.
The concrete steps of said step S2 are:
S201: adopt the MEMAO method to extract the initial background frame;
S202: each pixel is set up the one section histogram of sampling in the time, and histogram is carried out Filtering Processing;
S203: obtain filtered histogrammic spike position;
S204: the parameter to the pairing Gaussian distribution of each spike among the step S203 is carried out initialization;
S205: set up the HSV model.
The concrete steps of said step S203 are:
Obtain the position of all Wave crest and wave troughs in the histogram through histogrammic first order difference; Crest to gained screens; if ; Then the position at this crest
Figure 154831DEST_PATH_IMAGE002
place is the position of spike, and wherein is constant.
The invention has the beneficial effects as follows: adopt to combine improve one's methods (the MEAMO method) of median method and mode method that the deficiency that it exists when the initial background frame extracts is improved; Combine the mixed Gaussian shadow model to carry out shadow Detection and removal simultaneously; Realized an algorithm of target detection that can adapt to the influence of the slight disturbance of background scene and ambient lighting, real-time update, shadow removal, robust.The weak point that exists when adopting this method to overcome to apply to the traffic intelligent video monitoring to mixed Gauss model has effectively improved the sensitivity and the accuracy of moving object detection.
Description of drawings
Fig. 1 is improved moving object detection algorithm flow chart based on mixed Gauss model;
Fig. 2 is the process flow diagram of video acquisition;
Fig. 3 is that improved mixed Gaussian background model is set up and initialized process flow diagram.
Embodiment
Improved moving object detection algorithm based on mixed Gauss model of the present invention is as shown in Figure 1, and its concrete steps are following:
Step S1 is for gathering frame of video, and the hardware device that this step adopts comprises ccd video camera, DSP digital signal processor and PC, and an end of DSP digital signal processor connects ccd video camera, and the other end connects PC.The flow process of video acquisition is as shown in Figure 2, and its detailed step is following:
Step S101 is through CCD camera collection frame of video;
Step S102 carries out analog to digital conversion with the sequence of frames of video data of gathering;
Step S103 incorporates the sequence of frames of video that is converted to into formation to DSP application cushion space;
Step S104 reads the sequence of frames of video in the buffer zone.
Step S2 carries out the initialization of background model for extracting the initial background frame, sets up HSV component model, and this flow process is as shown in Figure 3, and its detailed steps is following:
Step S201 adopts the MEMAO method to extract the initial background frame.
Suppose in very little a period of time interval (at interval) like 100 frame times; Pixel (x; Y) pixel value is along with change of time is
Figure 533039DEST_PATH_IMAGE004
,
Figure 81832DEST_PATH_IMAGE005
...
Figure 356825DEST_PATH_IMAGE006
.Existing weighted value (making even all at this) with intermediate value and mode value is as the current background pixel value of this pixel, and calculating formula is:
Figure 623858DEST_PATH_IMAGE007
In the formula;
Figure 557179DEST_PATH_IMAGE008
is the intermediate value in this section time interval, be the mode value in this section time interval.
Step S202 sets up the one section histogram of sampling in the time to each pixel, and histogram is carried out Filtering Processing.What grey level histogram was described is the number of pixels of this gray level in the image or the frequency that this gray-level pixels occurs, and the present invention confirms the distribution of this pixel through histogram medium wave peak and trough; The operation of Filtering Processing is to remove histogram medium-high frequency component and carry out smoothly with Gaussian function, and purpose is to be convenient to calculate the crest position.
Step S203 obtains filtered histogrammic spike position;
Be divided into for two steps: (1) obtains the position of all Wave crest and wave troughs in the histogram through histogrammic first order difference.
The pixel value of if
Figure 422684DEST_PATH_IMAGE010
represent pixel point; The pixel value of pixel equals the number of the pixel of
Figure 38659DEST_PATH_IMAGE012
in
Figure 911434DEST_PATH_IMAGE011
representative sample;
Figure 562044DEST_PATH_IMAGE013
, then this histogrammic difference form is:
Figure 726309DEST_PATH_IMAGE014
Therefore have:
Figure 178323DEST_PATH_IMAGE015
(2) crest to the following formula gained screens, and the crest position is expression with
Figure 719026DEST_PATH_IMAGE002
.if
Figure 96917DEST_PATH_IMAGE001
, then the position at this crest place is exactly the position of required spike.Wherein
Figure 635346DEST_PATH_IMAGE003
is a constant, and how many its sizes sets corresponding value according to the sampling number purpose.
Step S204 carries out initialization to the parameter of the pairing Gaussian distribution of each spike among the step S203.
If the spike position is
Figure 98688DEST_PATH_IMAGE016
; The position of adjacent two troughs with this spike is
Figure 630033DEST_PATH_IMAGE017
,
Figure 128010DEST_PATH_IMAGE018
( ); Then the sample areas that relied on of the corresponding Gaussian distribution of this spike is , i.e. pixel value all pixels in
Figure 857435DEST_PATH_IMAGE020
.Obtain average
Figure 209919DEST_PATH_IMAGE021
and variance that sample areas calculates each Gaussian distribution afterwards; And the sample number of each sample areas
Figure 262505DEST_PATH_IMAGE023
; Then the weights of each Gaussian distribution can be initialized as
Figure 948702DEST_PATH_IMAGE024
(
Figure 342643DEST_PATH_IMAGE025
is the number of Gaussian distribution; Generally get between 3 to 5, when the spike number that is obtained then get greater than 5 the time maximum preceding 5 of number of samples in the corresponding sample areas.）。
Step S205 sets up the HSV model.
With said method the lightness component V and the chromatic component H of image set up model; the individual lightness component V model that obtains respectively; Note is done: , the weights of each model are designated as
Figure 543314DEST_PATH_IMAGE027
; Chromatic component H model; Note is done:
Figure 870390DEST_PATH_IMAGE028
, the weights of each model be
Figure 341692DEST_PATH_IMAGE029
very.Wherein
Figure 488639DEST_PATH_IMAGE030
;
Figure 454321DEST_PATH_IMAGE031
;
Figure 635904DEST_PATH_IMAGE032
,
Figure 278107DEST_PATH_IMAGE033
is i Gaussian distribution obtaining in initialization procedure average and the variance about lightness component V and chromatic component H.
Step S3, present frame get the prospect frame with background frames phase difference.Background frames in image of current input (being present frame) and the background model carries out differential ratio, gets rid of identical part (being background information), thereby detects moving target (being the prospect frame).
Step S4 carries out binary conversion treatment to resulting prospect frame.When the value in the prospect frame that obtains during greater than preset threshold
Figure 177929DEST_PATH_IMAGE003
; This value is changed to 1; Be judged as the foreground point; Otherwise be 0, be judged as background dot.Threshold value of the present invention
Figure 744040DEST_PATH_IMAGE034
.
Step S5, the prospect that integrating step S3 obtains is introduced and is upgraded weights, average and the variance that the factor is upgraded the mixed Gaussian background model, so that reduce the influence of foreground target to background model.
Step S6 after the renewal of mixed Gaussian background model, has confirmed background model, utilizes the Jeffrey value to judge whether to be the moving target prospect.Jeffrey value (Jeffrey divergence measure) is obtained by following formula:
Figure 717812DEST_PATH_IMAGE035
Wherein,
Figure 343966DEST_PATH_IMAGE036
,
Figure 652456DEST_PATH_IMAGE037
they are
Figure 22258DEST_PATH_IMAGE038
individual Gaussian distribution average and variance,
Figure 116116DEST_PATH_IMAGE039
,
Figure 913170DEST_PATH_IMAGE040
be the average and the variance of
Figure DEST_PATH_IMAGE041
individual Gaussian distribution.
The Jeffrey value of each distribution in pixel value of the pixel that will mate
Figure 708957DEST_PATH_IMAGE042
and the mixed Gaussian background model compares, and whether belongs in the mixed Gaussian distribution that has existed through relatively judging this pixel pixel value
Figure 554553DEST_PATH_IMAGE042
.
Definition:
Figure 565234DEST_PATH_IMAGE043
Figure 720141DEST_PATH_IMAGE044
Wherein,
Figure 816273DEST_PATH_IMAGE045
is the minimum Jeffrey value of the t moment
Figure 465560DEST_PATH_IMAGE038
individual Gaussian distribution
Figure 596327DEST_PATH_IMAGE046
about the S component,
Figure 735185DEST_PATH_IMAGE047
be the Jeffrey value of the t moment
Figure 239984DEST_PATH_IMAGE038
individual Gaussian distribution
Figure 755279DEST_PATH_IMAGE046
about the S component;
Figure 678236DEST_PATH_IMAGE048
is the minimum Jeffrey value of t moment i Gaussian distribution
Figure 987994DEST_PATH_IMAGE046
about the V component,
Figure 793139DEST_PATH_IMAGE049
be the minimum Jeffrey value of the t moment
Figure 33497DEST_PATH_IMAGE038
individual Gaussian distribution
Figure 873277DEST_PATH_IMAGE046
about the V component.
Promptly get the minimum distribution of Jeffrey value in K the mixed Gaussian distribution that has existed; And setting threshold
Figure 291620DEST_PATH_IMAGE050
, it is following that then whether this pixel belongs to the Rule of judgment of background dot:
Figure 849640DEST_PATH_IMAGE051
Figure 441158DEST_PATH_IMAGE052
In the formula;
Figure 587975DEST_PATH_IMAGE053
is according to the selected coefficient of experience; Span (0 ~ 1); The ratio of value may command
Figure 956956DEST_PATH_IMAGE054
,
Figure 352165DEST_PATH_IMAGE055
component model is established
Figure 166538DEST_PATH_IMAGE053
=0.5 among the present invention.As
Figure 175951DEST_PATH_IMAGE056
during less than threshold value
Figure 442984DEST_PATH_IMAGE050
(the present invention gets 0.7); The individual distribution coupling of pixel value
Figure 641884DEST_PATH_IMAGE042
and also judges that this is a background dot; Otherwise be the foreground point, also just obtained the moving target prospect.
Step S7 utilizes the mixed Gaussian shadow model to remove shade to the moving target prospect that obtains among the step S6.Be divided into for two steps:
(1) doubtful shadow model:, judge whether pixel value possibly be shade according to following decision-making formula:
Figure 241810DEST_PATH_IMAGE058
,
Figure 185681DEST_PATH_IMAGE060
,
Figure 709066DEST_PATH_IMAGE061
and
Figure 811015DEST_PATH_IMAGE062
,
Figure 239591DEST_PATH_IMAGE063
,
Figure 780293DEST_PATH_IMAGE064
difference remarked pixel point (x in the formula; H, S, the V component of new input pixel value y) and background pixel value
Figure 696614DEST_PATH_IMAGE066
; (x y) is this mask to SP.
Figure 159956DEST_PATH_IMAGE067
,
Figure 691301DEST_PATH_IMAGE068
,
Figure 923699DEST_PATH_IMAGE069
,
Figure 695346DEST_PATH_IMAGE070
are parameter;
Figure 770618DEST_PATH_IMAGE071
;
Figure 653123DEST_PATH_IMAGE067
value will be considered the intensity of shade; When the shade that throws on the background was stronger,
Figure 5607DEST_PATH_IMAGE067
was more little;
Figure 885839DEST_PATH_IMAGE068
is used for strengthening the robustness to noise, and promptly the brightness H of present frame can not be too similar with background;
Figure 323773DEST_PATH_IMAGE072
,
Figure 196920DEST_PATH_IMAGE070
value is debugging by rule of thumb mainly.If
Figure 403911DEST_PATH_IMAGE073
is judged as doubtful shade; SP (x then; Y) be changed to 1, otherwise be changed to 0.
(2) mixed Gaussian shadow model: remove shade
If a certain Gaussian distribution satisfies in doubtful shade value of input
Figure 455043DEST_PATH_IMAGE074
and the mixed Gaussian shadow model:
Figure 380274DEST_PATH_IMAGE075
Wherein subscript S representes the mixed Gaussian shadow model; The standard deviation of the moment
Figure 75063DEST_PATH_IMAGE057
individual Gaussian distribution that
Figure 791533DEST_PATH_IMAGE076
is
Figure 118609DEST_PATH_IMAGE077
, the average and the standard deviation of the moment
Figure 884122DEST_PATH_IMAGE057
individual Gaussian distribution that , are respectively
Figure 702540DEST_PATH_IMAGE080
.
Then these distribution parameter weights, average and variance are pressed following Policy Updates:
Figure 277058DEST_PATH_IMAGE081
Wherein, Weights, average and the standard deviation of the moment
Figure 651407DEST_PATH_IMAGE057
individual Gaussian distribution that
Figure 911301DEST_PATH_IMAGE082
,
Figure 664362DEST_PATH_IMAGE083
,
Figure 966031DEST_PATH_IMAGE084
are respectively
Figure 529867DEST_PATH_IMAGE080
; Weights, average and the standard deviation of the moment
Figure 615821DEST_PATH_IMAGE057
individual Gaussian distribution that
Figure 208159DEST_PATH_IMAGE085
,
Figure 364334DEST_PATH_IMAGE086
,
Figure 161389DEST_PATH_IMAGE087
are respectively
Figure 707908DEST_PATH_IMAGE088
;
Figure 813453DEST_PATH_IMAGE089
,
Figure 781409DEST_PATH_IMAGE090
general value is 0.3.
If there is not Gaussian distribution and doubtful shadows pixels value coupling; Then the minimum Gaussian distribution of weights will be upgraded by new Gaussian distribution; The new average that distributes is ; Standard deviation that initialization is bigger
Figure 392016DEST_PATH_IMAGE092
and less weights
Figure 717824DEST_PATH_IMAGE093
; Remaining Gaussian distribution keeps identical average and variance; But their weights can be decayed, that is:
Figure 301252DEST_PATH_IMAGE094
At last; The weights normalization of all Gaussian distribution, and distribute each and arrange from big to small by .if
Figure 473924DEST_PATH_IMAGE096
;
Figure 970634DEST_PATH_IMAGE097
...
Figure 41358DEST_PATH_IMAGE098
is the order of each Gaussian distribution by
Figure 32448DEST_PATH_IMAGE099
descending arrangement; If following criterion is satisfied in the individual distribution of preceding N (
Figure 872228DEST_PATH_IMAGE100
); Then this N distribution is considered to the shade distribution, that is:
Figure 539838DEST_PATH_IMAGE101
Wherein, Subscript S representes the mixed Gaussian shadow model; is the t weights of K Gaussian distribution constantly;
Figure 627060DEST_PATH_IMAGE103
is threshold value; Span (0～1) is rule of thumb generally got between 0.5 to 0.8.
Judge at last if the absolute value of the difference of doubtful shade and each shade distribution average all greater than
Figure 238487DEST_PATH_IMAGE076
of this distribution standard deviation times, then
Figure 205175DEST_PATH_IMAGE091
is considered to moving target; Otherwise
Figure 600384DEST_PATH_IMAGE091
is judged to shade; From sport foreground, remove it, the sport foreground of having eliminated shade is moving target.

Claims (4)

1. the moving target detecting method based on mixed Gauss model is characterized in that may further comprise the steps: S1: gather frame of video; S2: extract the initial background frame, carry out the initialization of background model, set up HSV component model; S3: present frame gets the prospect frame with background frames phase difference; S4: resulting prospect frame is carried out binary conversion treatment; S5:, introduce and upgrade weights, average and the variance that the factor is upgraded the mixed Gaussian background model according to the prospect frame of said step S3; S6: utilize the Jeffrey value to judge whether to be the moving target prospect; S7: utilize the mixed Gaussian shadow model to remove the shade of said moving target prospect.
2. a kind of moving target detecting method based on mixed Gauss model according to claim 1 is characterized in that the concrete steps of said step S2 are:
S201: adopt the MEMAO method to extract the initial background frame;
S202: each pixel is set up the one section histogram of sampling in the time, and histogram is carried out Filtering Processing;
S203: obtain filtered histogrammic spike position;
S204: the parameter to the pairing Gaussian distribution of each spike among the step S203 is carried out initialization;
S205: set up the HSV model.
3. a kind of moving target detecting method based on mixed Gauss model according to claim 1 is characterized in that the concrete steps of said step S203 are:
Obtain the position of all Wave crest and wave troughs in the histogram through histogrammic first order difference; Crest to gained screens; if
Figure 201110447408X100001DEST_PATH_IMAGE001
; Then the position at this crest
Figure 824549DEST_PATH_IMAGE002
place is the position of spike, and wherein
Figure 201110447408X100001DEST_PATH_IMAGE003
is constant.
4. a kind of moving target detecting method based on mixed Gauss model according to claim 1 is characterized in that the concrete steps of said step S7 are:
S701: according to doubtful shadow model, judge whether pixel value possibly be shade, if then SP (x y) is changed to 1, otherwise is changed to 0; The formula of doubtful shadow model is:
Figure 901090DEST_PATH_IMAGE004
Figure 201110447408X100001DEST_PATH_IMAGE005
,
Figure 671468DEST_PATH_IMAGE006
,
Figure 201110447408X100001DEST_PATH_IMAGE007
and
Figure 858867DEST_PATH_IMAGE008
, ,
Figure 31091DEST_PATH_IMAGE010
difference remarked pixel point (x in the formula; H, S, the V component of new input pixel value y) and background pixel value ; SP (x; Y) be this mask;
Figure 201110447408X100001DEST_PATH_IMAGE013
,
Figure 349257DEST_PATH_IMAGE014
,
Figure 201110447408X100001DEST_PATH_IMAGE015
, are parameter;
Figure 201110447408X100001DEST_PATH_IMAGE017
,
Figure 367078DEST_PATH_IMAGE018
;
S702: utilize the mixed Gaussian shadow model to remove shade; Its concrete grammar is: if the absolute value of the difference of doubtful shade
Figure 201110447408X100001DEST_PATH_IMAGE019
and each shade distribution average is all greater than
Figure 847738DEST_PATH_IMAGE020
of this distribution standard deviation times, then
Figure 327129DEST_PATH_IMAGE019
is considered to moving target; Otherwise
Figure 184227DEST_PATH_IMAGE019
is judged to shade, from sport foreground, removes it.
CN201110447408.XA 2011-12-28 2011-12-28 Moving object detection method based on Gaussian mixture model Expired - Fee Related CN102568005B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201110447408.XA CN102568005B (en) 2011-12-28 2011-12-28 Moving object detection method based on Gaussian mixture model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201110447408.XA CN102568005B (en) 2011-12-28 2011-12-28 Moving object detection method based on Gaussian mixture model

Publications (2)

Publication Number Publication Date
CN102568005A true CN102568005A (en) 2012-07-11
CN102568005B CN102568005B (en) 2014-10-22

Family

ID=46413351

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201110447408.XA Expired - Fee Related CN102568005B (en) 2011-12-28 2011-12-28 Moving object detection method based on Gaussian mixture model

Country Status (1)

Country Link
CN (1) CN102568005B (en)

Cited By (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102903123A (en) * 2012-09-08 2013-01-30 佳都新太科技股份有限公司 Self-adapting background subtracting method based on Gaussian mixture background reconstruction
CN103049748A (en) * 2012-12-30 2013-04-17 信帧电子技术（北京）有限公司 Behavior-monitoring method and behavior-monitoring system
CN103208126A (en) * 2013-04-17 2013-07-17 同济大学 Method for monitoring moving object in natural environment
CN103218829A (en) * 2013-04-01 2013-07-24 上海交通大学 Foreground extracting method suitable for dynamic background
CN103440666A (en) * 2013-07-19 2013-12-11 杭州师范大学 Intelligent device for fast positioning moving regions under non-static background
CN103617632A (en) * 2013-11-19 2014-03-05 浙江工业大学 Moving target detection method with adjacent frame difference method and Gaussian mixture models combined
CN103632340A (en) * 2012-08-24 2014-03-12 原相科技股份有限公司 Object tracking apparatus and operating method thereof
CN103686074A (en) * 2013-11-20 2014-03-26 南京熊猫电子股份有限公司 Method for positioning mobile object in video monitoring
CN103700114A (en) * 2012-09-27 2014-04-02 中国航天科工集团第二研究院二O七所 Complex background modeling method based on variable Gaussian mixture number
CN103810703A (en) * 2014-01-22 2014-05-21 安徽科力信息产业有限责任公司 Picture processing based tunnel video moving object detection method
CN103905734A (en) * 2014-04-17 2014-07-02 苏州科达科技股份有限公司 Method and device for intelligent tracking and photographing
WO2015123967A1 (en) * 2014-02-24 2015-08-27 深圳市华宝电子科技有限公司 Method and apparatus for recognizing moving target
CN105184817A (en) * 2015-08-31 2015-12-23 清华大学深圳研究生院 Moving object detection method by overcoming static foreground
CN105205833A (en) * 2015-09-15 2015-12-30 杭州中威电子股份有限公司 Moving object detection method and device based on space-time background model
CN105205832A (en) * 2015-08-31 2015-12-30 清华大学深圳研究生院 Moving object detection method
CN105303191A (en) * 2014-07-25 2016-02-03 中兴通讯股份有限公司 Method and apparatus for counting pedestrians in foresight monitoring scene
CN106205217A (en) * 2016-06-24 2016-12-07 华中科技大学 Unmanned plane automatic testing method based on machine vision and unmanned plane method of control
CN106488180A (en) * 2015-08-31 2017-03-08 上海悠络客电子科技有限公司 Video shadow detection method
CN107346534A (en) * 2017-07-13 2017-11-14 河北中科恒运软件科技股份有限公司 VS shadow Detection and removing method and system in mediation reality
CN107871315A (en) * 2017-10-09 2018-04-03 中国电子科技集团公司第二十八研究所 A kind of video image motion detection method and device
CN109102526A (en) * 2018-07-27 2018-12-28 东莞职业技术学院 The foreground detection method and device of the monitoring system of unmanned plane
CN110488242A (en) * 2018-05-15 2019-11-22 宁波傲视智绘光电科技有限公司 Echo signal processing method and device, radar and storage device
CN111582070A (en) * 2020-04-22 2020-08-25 浙江大学 Foreground extraction method for detecting video sprinkles on expressway
CN112597806A (en) * 2020-11-30 2021-04-02 北京影谱科技股份有限公司 Vehicle counting method and device based on sample background subtraction and shadow detection
CN113554685A (en) * 2021-08-02 2021-10-26 中国人民解放军海军航空大学航空作战勤务学院 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium
CN114782880A (en) * 2022-06-22 2022-07-22 索日新能源科技(南通)有限公司 Monitoring system for off-grid photovoltaic power generation system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050276446A1 (en) * 2004-06-10 2005-12-15 Samsung Electronics Co. Ltd. Apparatus and method for extracting moving objects from video
CN101364304A (en) * 2008-09-25 2009-02-11 上海交通大学 Shadow detection method based on color invariance and Gauss model
CN101916447A (en) * 2010-07-29 2010-12-15 江苏大学 Robust motion target detecting and tracking image processing system
CN102184553A (en) * 2011-05-24 2011-09-14 杭州华三通信技术有限公司 Moving shadow detecting method and device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050276446A1 (en) * 2004-06-10 2005-12-15 Samsung Electronics Co. Ltd. Apparatus and method for extracting moving objects from video
CN101364304A (en) * 2008-09-25 2009-02-11 上海交通大学 Shadow detection method based on color invariance and Gauss model
CN101916447A (en) * 2010-07-29 2010-12-15 江苏大学 Robust motion target detecting and tracking image processing system
CN102184553A (en) * 2011-05-24 2011-09-14 杭州华三通信技术有限公司 Moving shadow detecting method and device

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
周西汉，刘勃，周荷琴，袁非牛: "一种基于奔腾SIMD指令的快速背景提取方法", 《计算机工程与应用》, no. 27, 21 September 2004 (2004-09-21) *
宋雪桦，陈瑜，耿剑锋，陈景柱: "基于改进的混合高斯背景模型的运动目标检测", 《计算机工程与设计》, vol. 31, no. 21, 16 November 2010 (2010-11-16) *

Cited By (43)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103632340A (en) * 2012-08-24 2014-03-12 原相科技股份有限公司 Object tracking apparatus and operating method thereof
CN102903123A (en) * 2012-09-08 2013-01-30 佳都新太科技股份有限公司 Self-adapting background subtracting method based on Gaussian mixture background reconstruction
CN103700114B (en) * 2012-09-27 2017-07-18 中国航天科工集团第二研究院二O七所 A kind of complex background modeling method based on variable Gaussian mixture number
CN103700114A (en) * 2012-09-27 2014-04-02 中国航天科工集团第二研究院二O七所 Complex background modeling method based on variable Gaussian mixture number
CN103049748A (en) * 2012-12-30 2013-04-17 信帧电子技术（北京）有限公司 Behavior-monitoring method and behavior-monitoring system
CN103049748B (en) * 2012-12-30 2015-12-23 贺江涛 Behavior monitoring method and device
CN103218829A (en) * 2013-04-01 2013-07-24 上海交通大学 Foreground extracting method suitable for dynamic background
CN103218829B (en) * 2013-04-01 2016-04-13 上海交通大学 A kind of foreground extracting method being adapted to dynamic background
CN103208126A (en) * 2013-04-17 2013-07-17 同济大学 Method for monitoring moving object in natural environment
CN103208126B (en) * 2013-04-17 2016-04-06 同济大学 Moving object monitoring method under a kind of physical environment
CN103440666A (en) * 2013-07-19 2013-12-11 杭州师范大学 Intelligent device for fast positioning moving regions under non-static background
CN103440666B (en) * 2013-07-19 2016-05-25 杭州师范大学 The moving region intelligent apparatus of location fast under a kind of non-static background
CN103617632B (en) * 2013-11-19 2017-06-13 浙江工业大学 A kind of moving target detecting method of combination neighbor frame difference method and mixed Gauss model
CN103617632A (en) * 2013-11-19 2014-03-05 浙江工业大学 Moving target detection method with adjacent frame difference method and Gaussian mixture models combined
CN103686074A (en) * 2013-11-20 2014-03-26 南京熊猫电子股份有限公司 Method for positioning mobile object in video monitoring
CN103810703A (en) * 2014-01-22 2014-05-21 安徽科力信息产业有限责任公司 Picture processing based tunnel video moving object detection method
CN103810703B (en) * 2014-01-22 2016-09-21 安徽科力信息产业有限责任公司 A kind of tunnel based on image procossing video moving object detection method
WO2015123967A1 (en) * 2014-02-24 2015-08-27 深圳市华宝电子科技有限公司 Method and apparatus for recognizing moving target
US10068343B2 (en) 2014-02-24 2018-09-04 Shenzhen Huabao Electronic Technology Co., Ltd. Method and apparatus for recognizing moving target
CN103905734A (en) * 2014-04-17 2014-07-02 苏州科达科技股份有限公司 Method and device for intelligent tracking and photographing
CN105303191A (en) * 2014-07-25 2016-02-03 中兴通讯股份有限公司 Method and apparatus for counting pedestrians in foresight monitoring scene
CN105184817A (en) * 2015-08-31 2015-12-23 清华大学深圳研究生院 Moving object detection method by overcoming static foreground
CN106488180A (en) * 2015-08-31 2017-03-08 上海悠络客电子科技有限公司 Video shadow detection method
CN105205832A (en) * 2015-08-31 2015-12-30 清华大学深圳研究生院 Moving object detection method
CN105205832B (en) * 2015-08-31 2017-08-25 清华大学深圳研究生院 A kind of method of moving object detection
CN105184817B (en) * 2015-08-31 2017-10-27 清华大学深圳研究生院 A kind of method for overcoming static foreground moving object to detect
CN105205833B (en) * 2015-09-15 2018-03-16 杭州中威电子股份有限公司 A kind of moving target detecting method and device based on time-and-space background model
CN105205833A (en) * 2015-09-15 2015-12-30 杭州中威电子股份有限公司 Moving object detection method and device based on space-time background model
CN106205217B (en) * 2016-06-24 2018-07-13 华中科技大学 Unmanned plane automatic testing method based on machine vision and unmanned plane method of control
CN106205217A (en) * 2016-06-24 2016-12-07 华中科技大学 Unmanned plane automatic testing method based on machine vision and unmanned plane method of control
CN107346534A (en) * 2017-07-13 2017-11-14 河北中科恒运软件科技股份有限公司 VS shadow Detection and removing method and system in mediation reality
CN107346534B (en) * 2017-07-13 2020-10-30 河北中科恒运软件科技股份有限公司 Method and system for detecting and eliminating shadow of video object in mediated reality
CN107871315B (en) * 2017-10-09 2020-08-14 中国电子科技集团公司第二十八研究所 Video image motion detection method and device
CN107871315A (en) * 2017-10-09 2018-04-03 中国电子科技集团公司第二十八研究所 A kind of video image motion detection method and device
CN110488242A (en) * 2018-05-15 2019-11-22 宁波傲视智绘光电科技有限公司 Echo signal processing method and device, radar and storage device
CN109102526B (en) * 2018-07-27 2022-07-05 东莞职业技术学院 Foreground detection method and device of monitoring system of unmanned aerial vehicle
CN109102526A (en) * 2018-07-27 2018-12-28 东莞职业技术学院 The foreground detection method and device of the monitoring system of unmanned plane
CN111582070A (en) * 2020-04-22 2020-08-25 浙江大学 Foreground extraction method for detecting video sprinkles on expressway
CN111582070B (en) * 2020-04-22 2022-06-21 浙江大学 Foreground extraction method for detecting video sprinkles on expressway
CN112597806A (en) * 2020-11-30 2021-04-02 北京影谱科技股份有限公司 Vehicle counting method and device based on sample background subtraction and shadow detection
CN113554685A (en) * 2021-08-02 2021-10-26 中国人民解放军海军航空大学航空作战勤务学院 Method and device for detecting moving target of remote sensing satellite, electronic equipment and storage medium
CN114782880A (en) * 2022-06-22 2022-07-22 索日新能源科技(南通)有限公司 Monitoring system for off-grid photovoltaic power generation system
CN114782880B (en) * 2022-06-22 2022-11-01 索日新能源科技(南通)有限公司 Monitoring system for off-grid photovoltaic power generation system

Also Published As

Publication number Publication date
CN102568005B (en) 2014-10-22

Similar Documents

Publication Publication Date Title
CN102568005B (en) Moving object detection method based on Gaussian mixture model
CN102184552B (en) Moving target detecting method based on differential fusion and image edge information
CN101859440A (en) Block-based motion region detection method
CN102307274A (en) Motion detection method based on edge detection and frame difference
CN107705254B (en) City environment assessment method based on street view
CN102332167A (en) Target detection method for vehicles and pedestrians in intelligent traffic monitoring
CN105163110A (en) Camera cleanliness detection method and system and shooting terminal
CN106204586B (en) A kind of moving target detecting method under complex scene based on tracking
CN103473550B (en) Based on the leaf image dividing method of Lab space and local dynamic threshold
CN102881160B (en) Outdoor traffic sign identification method under low-illumination scene
CN105424709A (en) Fruit surface defect detection method based on image marking
CN101216943B (en) A method for video moving object subdivision
CN102034240A (en) Method for detecting and tracking static foreground
CN103258332A (en) Moving object detection method resisting illumination variation
CN103473547A (en) Vehicle target recognizing algorithm used for intelligent traffic detecting system
CN103679677A (en) Dual-model image decision fusion tracking method based on mutual updating of models
CN107154044A (en) A kind of dividing method of Chinese meal food image
CN104700405A (en) Foreground detection method and system
CN102663362A (en) Moving target detection method t based on gray features
CN105427279B (en) A kind of grassland Drought Information Monitoring System and method based on computer vision and Internet of Things
CN104182976B (en) Field moving object fining extraction method
CN110399785B (en) Method for detecting leaf occlusion based on deep learning and traditional algorithm
CN103209321B (en) A kind of video background Rapid Updating
CN111062926B (en) Video data processing method, device and storage medium
CN103425958A (en) Method for detecting non-movable objects in video

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20141022

Termination date: 20151228

EXPY Termination of patent right or utility model