CN102034240A - Method for detecting and tracking static foreground - Google Patents
Method for detecting and tracking static foreground Download PDFInfo
- Publication number
- CN102034240A CN102034240A CN 201010601984 CN201010601984A CN102034240A CN 102034240 A CN102034240 A CN 102034240A CN 201010601984 CN201010601984 CN 201010601984 CN 201010601984 A CN201010601984 A CN 201010601984A CN 102034240 A CN102034240 A CN 102034240A
- Authority
- CN
- China
- Prior art keywords
- static
- foreground
- tracking
- image
- prospect
- 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 relates to a method for detecting and tracking a static foreground. The method comprises the following steps of: (1) receiving a real-time data stream; (2) establishing a background model; (3) detecting a foreground; (4) tracking an object; (5) judging an event; and (6) storing an image. The method is a background model non-real-time updating technology and solves the problem that the static foreground is easily updated to the background model. In addition, the invention also provides a static foreground extraction method based on the position characteristics of the static foreground and the random distribution characteristics of light noise. The invention has the advantages of stably detecting the static foreground, furthest eliminating light noise and judging the static foreground from a foreground.
Description
Technical field
The present invention relates to a kind of static foreground detection and tracking, belong to image processing field and computer vision field.
Background technology
1, about static foreground detection correlation technique
So-called foreground detection, the zone with the target object correspondence extracts from image sequence exactly, at concrete traffic monitoring, exactly pedestrian in the scene and vehicle etc. is separated from the monitoring image sequence.The foreground detection algorithm can be divided three classes substantially at present: (1) background subtraction point-score, (2) time differencing method, (3) optical flow method.From the angle of real-time monitoring, generally adopt the background subtraction separating method to come the extraction prospect.
The conventional method that sport foreground detects is earlier background image to be carried out modeling, obtains foreground image by the background subtraction separating method, gets rid of background disturbance and shade in the foreground image then, just can extract foreground target more exactly.By foreground object being carried out tracking, obtain the movable information of prospect again, thereby judge whether foreground object is moving object (Stauffer, C based on the BLOB piece; Grimson, W. E. L. " Adaptive background mixture models for real-time tracking, " Proc. of CVPR 1999, vol. 2, pp.2246-2252.).This method also can be used for static foreground detection, but this method is not strong for the antijamming capability of light noise, and static prospect can be updated in the background along with the renewal of background model and goes, can not detect with being stabilized, when target object blocked, the detection accuracy rate of this method reduced greatly in addition.
Porikli(Porikli, F.; Ivanov, Y.; Haga, T. " Robust Abandoned Object Detection Using Dual Foregrounds ", Journal on Advances in Signal Processing, art.30,11 pp., 2008.) proposed to utilize the method for double-background model, this method is carried out foreground extraction to picture frame respectively based on the background difference of different frame per second, these two kinds of background subtraction separating methods all are based on mixed Gauss model, one of them model is used for short detection model, be every frame update, another model is used for long detection model, and promptly every n frame update once, short background model is upgraded fast, the change of scene can be reflected in the background model quickly, and long background model adapts to the variation of scene with a lower learning rate, therefore will compare by the prospect mask that two background models obtain, if certain prospect exists in long prospect mask, and does not exist in short prospect mask, then this object may be static foreground object.Double-background model can utilize long background model, turnover rate slows down, thereby solve static prospect to a certain extent and be dissolved into the problem of going in the background model easily, and the short background model of utilization, can realize the real-time update of background again, but slow these characteristics of the renewal of long background model have determined this model can introduce a large amount of light noises, thereby cause false detection rate to raise.
Liao, H-H(Liao, H-H.; Chang, J-Y.; Chen, L-G. " A localized Approach to abandoned luggage detection with Foreground-Mask sampling ", Proc. of AVSS 2008, pp. 132-139.) method based on the sampling of prospect mask is proposed, promptly on the basis of background difference, obtain the foreground image sequence, then the foreground image sequence is sampled, the binaryzation prospect mask figure of per 6 samples that come sampling carry out with, utilize with the result decide the value of pixel among the prospect mask S, each point on the S, if the point (value is 1) of white, show that then this point existed in 30 seconds in the past always, and probably be stationary body.This method utilizes stationary body to be present in characteristics in the prospect always, prospect mask image is sampled, and whether certain a bit belongs to static prospect in the analysis image by the image of sampling is added up.Its essence is constant these characteristics of the position feature that has utilized static prospect.This method can detect static prospect to a great extent, but the weak point of this method in, thereby it does not have to solve and how could allow static prospect can not be dissolved into and go to obtain the stable and problem foreground image that noise is little in the background image along with the variation of time.
Summary of the invention
The invention provides the technology that a kind of background model non real-time upgrades, solve static prospect and be updated to problem in the background model easily.In addition, also provide a kind of based on the position feature of static prospect and the static foreground extracting method of light noise stochastic distribution characteristics; Having stable detection goes out static prospect, farthest eliminates light noise and judge the advantage of static prospect from prospect is arranged.
Technical solution of the present invention is
:
A kind of static foreground detection and tracking, its method step is as follows:
1. the reception of real time data stream; 2. the foundation of background model; 3. foreground detection; 4. target following; 5. incident is judged; 6. image is preserved.
The reception of described real time data stream is the real-time streaming data by high definition web camera acquisition monitoring scene; Data stream arrives local computer through Network Transmission; Local computer utilizes the libvlc interface function to realize decoding, multiplexing to data stream; At last data stream is become exercisable picture data type.
The foundation of described background model is meant the mixed Gauss model that utilizes initial multi-frame video picture construction to go out initial background, follow-up interim again update background module.The value that mixed Gauss model (GMM) comes each pixel in the token image by a plurality of Gauss models is carried out modeling to each pixel exactly.This background model to video before 200 two field pictures carry out modeling;
The method step of described foreground detection is:
1. morphologic filtering is handled, and gets rid of the noise that most of environmental factor causes;
2. every M frame image sequence is carried out rim detection, the edge contour figure that obtains; Foreground pixel point to aforementioned edge contour figure is added up, if the probability that the some pixels of discovery existed in a period of time of setting then is defined as static prospect greater than setting threshold; If less than this threshold value, then it is filtered out;
3. utilize aforementioned foundation background model the N*M two field picture is carried out the background difference, obtain the foreground image of N*M two field picture; Step 2. middle edge statistics foreground image that obtains and the foreground image that obtains through the background subtraction branch are compared analysis, and filtering noise and moving object obtain static foreground image;
M, N are natural number, and preferred, M is 20.
Described target following is based on the shape and the motion feature of static prospect, and the BLOB piece in the foreground image is followed the tracks of, and obtains the relevant location information of static prospect and has the statistical information of duration in this position.
Described incident judges, is meant when time that certain static prospect exists in prospect during greater than assign thresholds, judges that this static prospect belongs to shed thing (or legacy etc.), and then implements warning.
Described image is preserved, be meant start report to the police after, the key images sequence of the L frame image sequence before reporting to the police is preserved; L is a natural number; Preferably, the L empirical value is 200.
In general motion detection, upgrade mixed Gauss model with each two field picture of newly obtaining, mate with each pixel in the present image and mixed Gauss model, if the match is successful then judge that this point is a background dot, otherwise be the foreground point.Gauss model mainly is by variance and two parameter decisions of average, for the study of average and variance, takes different study mechanisms, will directly have influence on stability, accuracy and the convergence of model.Since to the background extracting modeling of moving target, therefore need be to the variance in the Gauss model and two parameter real-time update of average.But in static foreground detection, the renewal meeting of background model is updated to stationary body in the background goes.Therefore upgrade in the background modeling stage, model is not carried out real-time update after model has been built, but upgrade background interimly, can stably obtain static prospect.This background model non real-time renewal technology has solved static prospect and has been updated to problem in the background model easily.
Comprise static prospect, dynamic prospect and light noise in the foreground image, computing machine can't be recognized the false prospect of worry that static prospect, dynamic prospect or light noise cause from a width of cloth foreground image.But, static prospect has the characteristics of invariant position, the regular in time variation in the position of sport foreground, then there are the characteristics of stochastic distribution in the light noise, by analyzing, adopt the method for statistics that the foreground pixel point in per 20 frame image sequence is added up,, illustrate that then it is that the possibility of static prospect is very big if find probability that some pixels exist in a period of time of setting greater than some threshold values; Otherwise,, then it is filtered out if less than that threshold value.Like this, just can get rid of a large amount of light noises, and static prospect is distinguished from sport foreground.
In addition, do not have the characteristics at edge in conjunction with light, former figure carried out rim detection, the foreground picture that will obtain by rim detection with compare analysis by the background difference foreground image after the statistics, almost can all noises of filtering and moving object, obtain cleaner static foreground image.This method testing result based on the static prospect of extraction of the position feature of static prospect and light noise stochastic distribution characteristics is accurate, has significantly reduced the interference of noise and moving object.
Description of drawings
Accompanying drawing is used to provide further understanding of the present invention, and constitutes the part of instructions, is used from explanation the present invention with embodiments of the invention one, is not construed as limiting the invention.In the accompanying drawings:
Fig. 1 is the system flow synoptic diagram of a kind of static foreground detection of the present invention and tracking;
Fig. 2 is a mixed Gaussian background modeling effect synoptic diagram of the present invention;
Fig. 3 is the foreground detection design sketch based on mixed Gauss model of the present invention;
Fig. 4 of the present inventionly sheds thing foreground detection algorithm flow chart based on rim detection;
Fig. 5 is a foreground detection design sketch of the present invention;
Fig. 6 is another group foreground detection design sketch of the present invention.
Embodiment
Below in conjunction with accompanying drawing the preferred embodiments of the present invention are described, should be appreciated that preferred embodiment described herein only is used for description and interpretation the present invention, and be not used in qualification the present invention.
Embodiment 1
Below by an embodiment who on special vehicle intelligent supervision platform, the vehicle incident of shedding is detected, be described in detail the present invention.
As shown in Figure 1, shed event detection procedure based on the vehicle of static foreground detection and mainly comprise several sections: the reception of real time data stream, the foundation of background model, foreground detection, target following, incident are judged, image is preserved.
1, the collection of real time data stream and reception
Real-time streaming data by high definition web camera acquisition monitoring scene.Data stream arrives local computer through Network Transmission.Local computer is by the reception of libvlc interface function realization to network data flow, but the conversion that utilizes opencv interface function realization flow data to arrive image data processing again; Handle for convenience, coloured image is converted to gray level image.
2, the foundation of background model
Before carrying out foreground detection, utilize the preceding N frame video image of beginning to construct initial background, the quality of its quality will directly have influence on the differential effect of back.
The value that mixed Gauss model (GMM) comes each pixel in the token image by a plurality of Gauss models can be carried out modeling to each pixel more accurately.Present embodiment obtains background model comparatively accurately, as shown in Figure 2 after 200 two field pictures carry out modeling before to video.
In general motion detection, upgrade mixed Gauss model with new each two field picture that obtains, mate with each pixel in the present image and mixed Gauss model, if the match is successful then judge that this point is a background dot, otherwise be the foreground point.But in static foreground detection, the renewal meeting of background model is updated to stationary body in the background goes, shown in (d) among Fig. 3, (e).Therefore present embodiment upgraded in the background modeling stage, model was not carried out real-time update after model has been built, and every two hours background was carried out modeling once more but set, and can stably obtain static prospect like this.
3, foreground detection
As shown in Figure 4 and Figure 5, method step is as follows:
1. will use the canny operator to carry out rim detection through the original image frame that morphologic filtering is handled, obtain the edge contour sequence chart;
2. the image after the rim detection is carried out binaryzation;
Whether the image of 3. judging above-mentioned binaryzation is N 20 frames, and N gets natural number; If 4. N 20 frames enter step, 1. carry out rim detection otherwise return step;
4. carry out following two operations, the one, judge that whether the time that pixel exists (representing with the frame number that exists) is greater than threshold value 18 in this 20 frame image sequence; If the result is for being, then there is this point among the statistics figure, otherwise does not exist, thereby obtain edge statistics figure; The 2nd, utilize the background model of above-mentioned foundation that the 20th two field picture is carried out the background difference, obtain the binaryzation foreground image of the 20th two field picture;
5. with step edge statistics 4. as a result figure with compare analysis through the binaryzation foreground image of background difference, filter out most of hot-tempered sound and moving object, obtain cleaner foreground object, as shown in Figure 5.
4, target following
Interested object before the purpose of target following is to find out from the follow-up frame of video flowing in certain frame.
The tracking of target is the shape and the kinetic characteristic of combining target, realizes the object of difformity, different motion characteristic is carried out the technology of Intelligent Recognition.In following the tracks of processing procedure, mainly solve the problems such as initialization, renewal and end of following the tracks of.
Shape and motion feature based on static prospect, BLOB piece in the foreground image is followed the tracks of, can be obtained relevant location information and some statistical informations of static prospect by tracking, as the positional information of target to the BLOB piece, motion state information, the temporal information of existence.
5, incident is judged
When time that certain static prospect exists in prospect greater than certain threshold value, can judge then that this static prospect belongs to shed thing (or legacy), and then implement to report to the police; If present embodiment target setting position does not become, movement velocity is zero, and the time of existence then is judged as static thing greater than 15s, implements to report to the police.Shown in (i) among Fig. 5, to shed quality testing and survey based on the foreground detection figure shown in (h) is carried out tracking and matching, testing result is accurate, is not subjected to the interference of hot-tempered sound and moving object.
5, image is preserved
After starting warning, promptly the key images sequence of 200 frame image sequence before reporting to the police is preserved.
Fig. 6 is another group foreground detection design sketch of the present invention.
It should be noted that at last: the above only is the preferred embodiments of the present invention, be not limited to the present invention, although the present invention is had been described in detail with reference to previous embodiment, for a person skilled in the art, it still can be made amendment to the technical scheme that aforementioned each embodiment put down in writing, and perhaps part technical characterictic wherein is equal to replacement.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. static foreground detection and tracking is characterized in that:
Its method step is as follows:
1. the reception of real time data stream; 2. the foundation of background model; 3. foreground detection; 4. target following; 5. incident is judged; 6. image is preserved.
2. static foreground detection according to claim 1 and tracking is characterized in that: the reception of described real time data stream is the real-time streaming data by high definition web camera acquisition monitoring scene; Data stream arrives local computer through Network Transmission; Local computer utilizes the libvlc interface function to realize decoding, multiplexing to data stream; At last data stream is become exercisable picture data type.
3. static foreground detection according to claim 1 and tracking, it is characterized in that: the foundation of described background model, be meant the mixed Gauss model that utilizes initial multi-frame video picture construction to go out initial background, follow-up interim again update background module.
4. static foreground detection according to claim 3 and tracking is characterized in that: utilize 200 initial frame video images to construct the mixed Gauss model of initial background.
5. static foreground detection according to claim 1 and tracking is characterized in that: the method step of described foreground detection is:
1. morphologic filtering is handled, and gets rid of the noise that most of environmental factor causes;
2. every M frame image sequence is carried out rim detection, the edge contour figure that obtains; Foreground pixel point to aforementioned edge contour figure is added up, if the probability that the some pixels of discovery existed in a period of time of setting then is defined as static prospect greater than setting threshold; If less than this threshold value, then it is filtered out;
3. utilize aforementioned foundation background model the N*M two field picture is carried out the background difference, obtain the foreground image of N*M two field picture; Step 2. middle edge statistics foreground image that obtains and the foreground image that obtains through the background subtraction branch are compared analysis, and filtering noise and moving object obtain static foreground image;
M, N are natural number.
6. static foreground detection according to claim 5 and tracking is characterized in that: the M value is 20.
7. static foreground detection according to claim 1 and tracking, it is characterized in that: described target following, be based on the shape and the motion feature of static prospect, BLOB piece in the foreground image is followed the tracks of, and the relevant location information that obtains static prospect reaches the statistical information that has duration in this position.
8. static foreground detection according to claim 1 and tracking, it is characterized in that: described incident is judged, be meant when time that certain static prospect exists in prospect during, judge that this static prospect belongs to shed thing or legacy etc., and then implement to report to the police greater than assign thresholds.
9. static foreground detection according to claim 1 and tracking is characterized in that: described image is preserved, be meant start report to the police after, the key images sequence of the L frame image sequence before reporting to the police is preserved; L is a natural number.
10. static foreground detection according to claim 9 and tracking is characterized in that: the empirical value of L is 200.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010601984 CN102034240A (en) | 2010-12-23 | 2010-12-23 | Method for detecting and tracking static foreground |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010601984 CN102034240A (en) | 2010-12-23 | 2010-12-23 | Method for detecting and tracking static foreground |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102034240A true CN102034240A (en) | 2011-04-27 |
Family
ID=43887093
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010601984 Pending CN102034240A (en) | 2010-12-23 | 2010-12-23 | Method for detecting and tracking static foreground |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102034240A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254396A (en) * | 2011-07-06 | 2011-11-23 | 北京国铁华晨通信信息技术有限公司 | Intrusion detection method and device based on video |
| CN102572518A (en) * | 2012-01-13 | 2012-07-11 | 河海大学 | VideoLan client (VLC)-based video image analysis platform and method |
| CN102708565A (en) * | 2012-05-07 | 2012-10-03 | 深圳市贝尔信智能系统有限公司 | Foreground detection method, device and system |
| CN103729613A (en) * | 2012-10-12 | 2014-04-16 | 浙江大华技术股份有限公司 | Method and device for detecting video image |
| CN103778785A (en) * | 2012-10-23 | 2014-05-07 | 南开大学 | Vehicle tracking and detecting method based on parking lot environment video monitoring |
| CN105210368A (en) * | 2013-05-09 | 2015-12-30 | 国立大学法人东京大学 | Background-differential extraction device and background-differential extraction method |
| CN106228572A (en) * | 2016-07-18 | 2016-12-14 | 西安交通大学 | The long inactivity object detection of a kind of carrier state mark and tracking |
| CN106297278A (en) * | 2015-05-18 | 2017-01-04 | 杭州海康威视数字技术股份有限公司 | A kind of method and system shedding thing vehicle for inquiry |
| CN106991674A (en) * | 2016-01-21 | 2017-07-28 | 深圳超多维光电子有限公司 | A kind of image processing method, device and electronic equipment |
| CN107918762A (en) * | 2017-10-24 | 2018-04-17 | 江西省高速公路投资集团有限责任公司 | A kind of highway drops thing rapid detection system and method |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
| CN109409315A (en) * | 2018-11-07 | 2019-03-01 | 浩云科技股份有限公司 | A kind of ATM machine panel zone remnant object detection method and system |
| US10515463B2 (en) | 2018-04-20 | 2019-12-24 | Sony Corporation | Object segmentation in a sequence of color image frames by background image and background depth correction |
| CN111968181A (en) * | 2020-08-21 | 2020-11-20 | 中科微至智能制造科技江苏股份有限公司 | Real-time parcel position detection method and system based on image processing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1801930A (en) * | 2005-12-06 | 2006-07-12 | 南望信息产业集团有限公司 | Dubious static object detecting method based on video content analysis |
| CN101119482A (en) * | 2007-09-28 | 2008-02-06 | 北京智安邦科技有限公司 | Overall view monitoring method and apparatus |
| CN101458871A (en) * | 2008-12-25 | 2009-06-17 | 北京中星微电子有限公司 | Intelligent traffic analysis system and application system thereof |
-
2010
- 2010-12-23 CN CN 201010601984 patent/CN102034240A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1801930A (en) * | 2005-12-06 | 2006-07-12 | 南望信息产业集团有限公司 | Dubious static object detecting method based on video content analysis |
| CN101119482A (en) * | 2007-09-28 | 2008-02-06 | 北京智安邦科技有限公司 | Overall view monitoring method and apparatus |
| CN101458871A (en) * | 2008-12-25 | 2009-06-17 | 北京中星微电子有限公司 | Intelligent traffic analysis system and application system thereof |
Non-Patent Citations (1)
| Title |
|---|
| 《中国科技论文在线》 20101216 向梅 一种基于多级框架的静态前景检测方法 摘要，第1页引言第1段，第2页第22-26行，第3页第15行-第9页第5行、图6 1-10 , * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254396A (en) * | 2011-07-06 | 2011-11-23 | 北京国铁华晨通信信息技术有限公司 | Intrusion detection method and device based on video |
| CN102254396B (en) * | 2011-07-06 | 2014-06-04 | 通号通信信息集团有限公司 | Intrusion detection method and device based on video |
| CN102572518A (en) * | 2012-01-13 | 2012-07-11 | 河海大学 | VideoLan client (VLC)-based video image analysis platform and method |
| CN102572518B (en) * | 2012-01-13 | 2014-04-09 | 河海大学 | VideoLan client (VLC)-based video image analysis platform and method |
| CN102708565A (en) * | 2012-05-07 | 2012-10-03 | 深圳市贝尔信智能系统有限公司 | Foreground detection method, device and system |
| CN103729613A (en) * | 2012-10-12 | 2014-04-16 | 浙江大华技术股份有限公司 | Method and device for detecting video image |
| CN103778785A (en) * | 2012-10-23 | 2014-05-07 | 南开大学 | Vehicle tracking and detecting method based on parking lot environment video monitoring |
| CN105210368A (en) * | 2013-05-09 | 2015-12-30 | 国立大学法人东京大学 | Background-differential extraction device and background-differential extraction method |
| CN105210368B (en) * | 2013-05-09 | 2018-10-26 | 国立大学法人东京大学 | Background difference extraction element and background difference extracting method |
| CN106297278A (en) * | 2015-05-18 | 2017-01-04 | 杭州海康威视数字技术股份有限公司 | A kind of method and system shedding thing vehicle for inquiry |
| CN106991674A (en) * | 2016-01-21 | 2017-07-28 | 深圳超多维光电子有限公司 | A kind of image processing method, device and electronic equipment |
| CN106228572A (en) * | 2016-07-18 | 2016-12-14 | 西安交通大学 | The long inactivity object detection of a kind of carrier state mark and tracking |
| CN106228572B (en) * | 2016-07-18 | 2019-01-29 | 西安交通大学 | A kind of the long inactivity object detection and tracking of carrier state mark |
| CN107918762A (en) * | 2017-10-24 | 2018-04-17 | 江西省高速公路投资集团有限责任公司 | A kind of highway drops thing rapid detection system and method |
| CN107918762B (en) * | 2017-10-24 | 2022-01-14 | 江西省高速公路投资集团有限责任公司 | Rapid detection system and method for road scattered objects |
| CN108257347A (en) * | 2018-01-10 | 2018-07-06 | 安徽大学 | A kind of flame image sequence sorting technique and device using convolutional neural networks |
| CN108257347B (en) * | 2018-01-10 | 2020-09-29 | 安徽大学 | Flame image sequence classification method and device by using convolutional neural network |
| US10515463B2 (en) | 2018-04-20 | 2019-12-24 | Sony Corporation | Object segmentation in a sequence of color image frames by background image and background depth correction |
| CN109409315A (en) * | 2018-11-07 | 2019-03-01 | 浩云科技股份有限公司 | A kind of ATM machine panel zone remnant object detection method and system |
| CN109409315B (en) * | 2018-11-07 | 2022-01-11 | 浩云科技股份有限公司 | Method and system for detecting remnants in panel area of ATM (automatic Teller machine) |
| CN111968181A (en) * | 2020-08-21 | 2020-11-20 | 中科微至智能制造科技江苏股份有限公司 | Real-time parcel position detection method and system based on image processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102034240A (en) | Method for detecting and tracking static foreground | |
| CN109872341B (en) | High-altitude parabolic detection method and system based on computer vision | |
| CN109309811B (en) | High-altitude parabolic detection system and method based on computer vision | |
| CN104303193B (en) | Target classification based on cluster | |
| CN102184552B (en) | Moving target detecting method based on differential fusion and image edge information | |
| CN109101944B (en) | Real-time video monitoring method for identifying garbage thrown into river channel | |
| WO2022001961A1 (en) | Detection method, detection device and detection system for moving target thrown from height | |
| CN108038415B (en) | Unmanned aerial vehicle automatic detection and tracking method based on machine vision | |
| CN102663362B (en) | Moving target detection method based on gray features | |
| CN111401311A (en) | High-altitude parabolic recognition method based on image detection | |
| CN101799968B (en) | Detection method and device for oil well intrusion based on video image intelligent analysis | |
| WO2021139049A1 (en) | Detection method, detection apparatus, monitoring device, and computer readable storage medium | |
| CN103824070A (en) | Rapid pedestrian detection method based on computer vision | |
| CN103279737A (en) | Fight behavior detection method based on spatio-temporal interest point | |
| CN102332167A (en) | Target detection method for vehicles and pedestrians in intelligent traffic monitoring | |
| CN110490043A (en) | A kind of forest rocket detection method based on region division and feature extraction | |
| CN108022249A (en) | A kind of remote sensing video satellite moving vehicle target region of interest extraction method | |
| CN104200466A (en) | Early warning method and camera | |
| CN114639075B (en) | Method and system for identifying falling object of high altitude parabola and computer readable medium | |
| CN108898042B (en) | Method for detecting abnormal user behavior in ATM cabin | |
| Brahme et al. | An implementation of moving object detection, tracking and counting objects for traffic surveillance system | |
| CN103593679A (en) | Visual human-hand tracking method based on online machine learning | |
| CN109948474A (en) | AI thermal imaging all-weather intelligent monitoring method | |
| CN105554462A (en) | Remnant detection method | |
| Han et al. | A method based on multi-convolution layers joint and generative adversarial networks for vehicle detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20110427 |