CN106204594A - A kind of direction detection method of dispersivity moving object based on video image - Google Patents
A kind of direction detection method of dispersivity moving object based on video image Download PDFInfo
- Publication number
- CN106204594A CN106204594A CN201610548836.4A CN201610548836A CN106204594A CN 106204594 A CN106204594 A CN 106204594A CN 201610548836 A CN201610548836 A CN 201610548836A CN 106204594 A CN106204594 A CN 106204594A
- Authority
- CN
- China
- Prior art keywords
- pixel
- image
- region
- motion
- moving
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to the direction detection method of a kind of dispersivity moving object based on video image, including: the gray processing first carrying out video image processes；Extract foreground image；Foreground image is carried out binary conversion treatment and morphologic filtering；Circumscribed rectangular region after being merged；The characteristic point of the circumscribed rectangular region after merging in front and back two frame is tracked, calculates dense optical flow, each pixel optical flow field in the circumscribed rectangular region of doubtful moving target；The motion excursion amount of same pixel in two frames before and after contrast, after rejecting the pixel that motion excursion amount is less than 2, the pixel number that contrast remains accounts for the ratio of original pixel, if ratio is more than 60%, then judges that rectangular area meets doubtful moving region；Obtain the direction of motion of object.The present invention has fast and convenient feature.
Description
Technical field
The present invention relates to the moving object detection in computer vision field and Orientation, especially to dispersivity moving object
Body angle detecting has good effect.
Background technology
The detection of image motion target and process are important research contents in computer vision, contain and move
Image sequence produced by the information of the whole motor process of object can provide, for the detection of moving target, the data that many is abundant,
The detection making moving target is possibly realized.For whole computer vision subject, this is an extremely important and range of application
The widest object of study.
The process that realizes of whole moving object detection is: obtains image sequence from photographic head, is carried out smoothing denoising, face
The a series of images such as color space transformation, morphological image process process, and utilize the information that image sequence frame is contained, and use mixing
The method of Gaussian Background modeling sets up a background model, and all with this Background, each two field picture in image sequence is done difference
Point, thus obtain difference image, finally difference image is carried out binary conversion treatment, obtain containing the binary map of moving target information
Picture, is finally completed the detection of moving target.And the detection of moving target is motion target tracking, Classification and Identification, walking direction
Etc. the basis of a series of subsequent treatment, the missing inspection of moving target or flase drop will result in the accuracy of these subsequent treatment.
For the detection of moving target, a lot of scholars both domestic and external propose multiple method to this, and currently a popular is main
There are three kinds: optical flow method, frame difference method and background subtraction method.
In some occasion owing to needing to detect the direction of motion of target, judgement is to enter detection region or leave detection zone
Territory, it is thus desirable to set warning line. and the judgement of the direction of motion is typically judged by the direction of motion of detection target. namely transports
By the thought of estimation. according to tradition evaluation method, need each pixel in image is calculated, calculate image every
The sports ground of a bit, then obtains the sports ground of entire image, such algorithm, and amount of calculation is quite big, cannot complete in real time.
Summary of the invention
The present invention proposes a kind of method calculating and simply can quickly detecting dispersivity movement direction of object.The skill of the present invention
Art scheme is as follows:
A kind of direction detection method of dispersivity moving object based on video image, the step including following:
1) gray processing first carrying out video image processes, and coloured image is converted into gray level image.
2) method using background modeling extracts the foreground image of video image.
3) foreground image is carried out binary conversion treatment and morphologic filtering, then by filtered bianry image is carried out
The lookup of profile, then stores this profile in one sequence, splits profile according to distance between element in sequence
Thus profile is divided into a few class, finally the types of profiles boundary rectangle of same type is outlined.
4) for step 3) in obtain each scattered boundary rectangle, carry out the judgement of distance, when two boundary rectangles away from
From during less than certain value, then two boundary rectangles are merged into a rectangle, if the rectangular area after He Binging is less than certain threshold value,
I.e. get rid of this rectangle, the circumscribed rectangular region after being merged.
5) characteristic point of the circumscribed rectangular region after merging in front and back two frame is tracked, calculates dense optical flow, obtain
Each pixel optical flow field in the circumscribed rectangular region of doubtful moving target.
6) by step 5) obtain each pixel optical flow field in the circumscribed rectangular region of doubtful moving target, before and after contrast two
The motion excursion amount of same pixel in frame, after rejecting the pixel that motion excursion amount is less than 2, the pixel that contrast remains
Counting and account for the ratio of original pixel, if ratio is more than 60%, then judging that rectangular area meets doubtful moving region, if being less than
60%, then get rid of this rectangular area.
7) to the pixel retained, calculate its mean motion side-play amount, i.e. obtain the object in this doubtful moving region
The average distance that front and back two frames move, then gets rid of this doubtful moving region when the average distance of movement is less than 4 pixels, no
Then this doubtful moving region, extends its average moving vector and rectangle intersection point, i.e. can get the direction of motion of object.
Accompanying drawing explanation
Accompanying drawing is the testing result using inventive algorithm to obtain:
Fig. 1 is the extraction of moving region；
Fig. 2 for carrying out optical flow method to moving region, and obtains after using mean deviation amount and Mean-Shift algorithm process
Optical flow field.
Fig. 3 is the result figure of direction of motion detection.
Detailed description of the invention
The direction detection method of the dispersivity moving object based on video image of the present invention, technical scheme is as follows:
1) gray processing first carrying out image processes, and coloured image is converted into gray level image.Each picture of coloured image
Vegetarian refreshments has tri-components of R, G, B, and the span of each component is 0 to 255, and the most each component has 16,000,000 (256*
256*256) situation.By contrast, the value of tri-components of R, G, B of gray level image is identical, consequently only that 256 kinds of situations.Gray scale
The features such as it is little that figure has view data, and treatment effeciency is high, the complexity of reduction algorithm time.
2) method utilizing moving object detection extracts the foreground image of dispersivity moving object, and the present invention uses background to build
The method of mould extracts foreground image.
3) foreground image is carried out binary conversion treatment and morphologic filtering, then by filtered bianry image is carried out
The lookup of profile (this profile surrounds binary image), then stores this profile in one sequence, according to element in sequence
Between distance profile is split thus profile is divided into a few class, finally by the types of profiles of same type boundary rectangle frame
Go out, obtain the boundary rectangle of each profile.The boundary rectangle now obtained is the most discrete, due to dispersivity moving object campaign relatively
Slowly, and move there is waving property, generally a moving meshes is become multiple little rectangle.
4) step 3) in obtain each scattered boundary rectangle, every boundary rectangle is carried out the judgement of distance, when two is external
When the distance of rectangle is less than certain value, then two boundary rectangles are merged into a rectangle.Rectangle after being combined has certain face
Long-pending requirement, when rectangular area is less than certain threshold value, i.e. gets rid of this rectangle.Such as, the most obtained video comprises the time, aobvious
Show that beating of time font also can be detected, set by certain area, then can get rid of font and beat etc. and to bring
Impact.
5) algorithm utilizing moving object detection can obtain the rectangular area of motion, and to same rectangle in front and back two frame
The characteristic point in region is tracked, and obtains its motion excursion amount, i.e. uses in obtained rectangle
The algorithm of calcOpticalFlowFarneback calculates dense optical flow, and (i.e. on image, the light stream of all pixels all calculates
Come), thus obtain the optical flow field of each pixel in rectangular area.
6) in steps 5) each pixel optical flow field in rectangular area is obtained, the fortune of same pixel in two frames before and after contrast
Dynamic side-play amount, when the motion excursion of this pixel is less than 2 pixels, then rejects this pixel, if more than the distance of 2 pixels
Then retain.The pixel number that contrast remains accounts for the ratio of original pixel, if ratio is more than 60%, then rectangular area meets
Doubtful moving region, if less than 60%, then getting rid of this rectangular area.
7) to the pixel retained, calculate its mean deviation amount, i.e. can obtain before and after the object in this rectangular area two
The average distance that frame moves.This rectangular area is then got rid of when the average distance of movement is less than 4 pixels.Extend this averagely to move
Dynamic vector and rectangle intersection point, i.e. can obtain the direction of motion of object in this rectangular area.
It is embodied as step below in conjunction with what embodiment illustrated to detect diffusivity moving object direction:
1. the gray processing of image
The method that coloured image changes into gray level image has:
A) meansigma methods of RGB component is taken
B) the maximum value of numerical value in RGB component is taken as gray value.
GRAY=Max (R, G, B)
C) RGB component is taken weighted average, obtain gray value.
GRAY=ωrR+ωgG+ωbB
2. the extraction of sport foreground
In method, the extraction of sport foreground uses background mean value relief method.
Assume fk(i j) represents the gray value of video sequence kth frame, Bk(i, j) represents the gray value of video sequence kth frame,
Dk(i, j) represents difference:
Dk(i, j)=| fk(i, j)-Bk(i, j) |
Wherein, (i, j) coordinate position in representative image, for the difference of foreground and background, a selected suitable threshold
Value T judges that this point is prospect or background, if this difference is more than threshold value T, then judges that this point is foreground pixel, and no person is judged to
Background pixel point.Use Rk(i, j) represents the pixel of moving region, then:
The principle of background mean value relief method is: compare fk(i, gray value j) and Bk(i, size j), if fk(i, j)
Value is more than Bk(i, j), then updates the background model of next frame at this gray value pointed out, and the gray value of this pixel will add 1；
If fk(i, value j) is less than Bk(i, j), then update the background model of next frame at this gray value pointed out, will this pixel
Gray value subtracts 1；If fk(i, value j) is equal to Bk(i, j), then the background model updating next frame is constant at this gray value pointed out.
It is formulated as:
In above formula, Bk+1(i, j) represents the gray value that the background model of next frame is pointed out in this pixel,
3. pair motion pixel carries out connected domain extraction
By the sport foreground obtained is carried out morphologic filtering, remove some little interference.Re-use expansion algorithm, make
The region of next-door neighbour merges.The mode finally utilizing freeman chain code describes the profile of each foreground area detected.
4. the classification of moving region with merge
Due to the impact of various factors, a moving object is often divided into several pieces, and this is accomplished by carrying out these regions
Merge.Foreground image is carried out binaryzation, then (this profile surrounds binary picture by filtered bianry image carries out profile
Picture) lookup, then this profile is stored in one sequence, according to distance between element in sequence profile split from
And profile is divided into a few class, finally the types of profiles boundary rectangle of same type is outlined, i.e. obtain moving region.This method
Process is the video (352*288) of CIF form, so when the distance of two rectangles is less than 40, then two rectangles being merged into one
Individual rectangle.
5. direction of motion detection
The first step: to rectangular area obtained above, front and back two frame is used in obtained rectangular area
The algorithm of calcOpticalFlowFarneback calculates dense optical flow, obtains in rectangular area about each pixel light stream
, i.e. displacement vector.(i.e. on image, the light stream of all pixels is all to calculate dense optical flow with the algorithm of Gunnar Farneback
Calculate), its correlative theses is: " Two-Frame Motion Estimation Based on
PolynomialExpansion"
Second step: owing to optical flow method is more sensitive to factors such as illumination, and other factors will to obtained optical flow field
Can produce impact, this is accomplished by processing obtained optical flow field.The inventive method uses mean deviation amount: utilize fortune
Dynamic partitioning algorithm obtains suspicious motion target, and is tracked the characteristic point of front and back two frame moving target, obtains its motion excursion
Amount.Assuming that there be n characteristic point in a certain region, before and after ith feature point, the coordinate of two frames is respectively (xi, yi)、(x′i, y 'i)。
Side-play amount is Li, thenJudge LiSize, work as LiLess than during 2 pixel distances then
Reject this pixel；Work as LiThis pixel is then retained during more than 2 pixel distances.
3rd step: the pixel point being retained second step uses Mean-Shift algorithm, Mean-Shift is a kind of
The multi-model dividing method of imparametrization, its basic calculating module uses traditional pattern recognition program, i.e. by dividing
The feature space of analysis image and the method for cluster reach the purpose of segmentation.It is by direct estimation feature space probability density
The local maximum of function obtains the density mode of unknown classification, and determines the position of this pattern, is then allowed to cluster and arrives
In the middle of the classification relevant with this pattern.Use each pixel optical flow field that second step is retained by Mean-Shift algorithm
Process, can obtain converging to the place that probability density is maximum, retain the point close to the place of probability density maximum, thus
Reject the point comparing edge further.
4th step: set and meet the pixel of mean deviation amount and Mean-Shift algorithm in rectangular area and have k, if front and back
In two frames, the average of k point coordinates is
ByCan calculateExtend2 available intersection points with rectangle A and B represents terminal and the starting point of the direction of motion respectively, thus obtains the direction of motion.
Claims (1)
1. a direction detection method for dispersivity moving object based on video image, the step including following:
1) gray processing first carrying out video image processes, and coloured image is converted into gray level image；
2) method using background modeling extracts the foreground image of video image；
3) foreground image is carried out binary conversion treatment and morphologic filtering, then by filtered bianry image is carried out profile
Lookup, then this profile is stored in one sequence, according to distance between element in sequence profile split thus
Profile is divided into a few class, finally the types of profiles boundary rectangle of same type is outlined；
4) for step 3) in obtain each scattered boundary rectangle, carry out the judgement of distance, when the distance of two boundary rectangles is little
When certain value, then two boundary rectangles are merged into a rectangle, if the rectangular area after He Binging is less than certain threshold value, i.e. arrange
Except this rectangle, the circumscribed rectangular region after being merged；
5) characteristic point of the circumscribed rectangular region after merging in front and back two frame is tracked, calculates dense optical flow, obtain doubtful
Each pixel optical flow field in the circumscribed rectangular region of moving target；
6) by step 5) obtain each pixel optical flow field in the circumscribed rectangular region of doubtful moving target, before and after contrast in two frames
The motion excursion amount of same pixel, after rejecting the pixel that motion excursion amount is less than 2, the pixel number that contrast remains
Account for the ratio of original pixel, if ratio is more than 60%, then judge that rectangular area meets doubtful moving region, if less than 60%,
Then get rid of this rectangular area；
7) to the pixel retained, its mean motion side-play amount is calculated, before and after i.e. obtaining the object in this doubtful moving region
The average distance that two frames move, then gets rid of this doubtful moving region when the average distance of movement is less than 4 pixels, otherwise this
Doubtful moving region, extends its average moving vector and rectangle intersection point, i.e. can get the direction of motion of object.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610548836.4A CN106204594A (en) | 2016-07-12 | 2016-07-12 | A kind of direction detection method of dispersivity moving object based on video image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610548836.4A CN106204594A (en) | 2016-07-12 | 2016-07-12 | A kind of direction detection method of dispersivity moving object based on video image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106204594A true CN106204594A (en) | 2016-12-07 |
Family
ID=57478095
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610548836.4A Pending CN106204594A (en) | 2016-07-12 | 2016-07-12 | A kind of direction detection method of dispersivity moving object based on video image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106204594A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570891A (en) * | 2016-11-03 | 2017-04-19 | 天津大学 | Target tracking algorithm based on video image taken by fixed camera |
| CN107048890A (en) * | 2017-05-25 | 2017-08-18 | 美的集团股份有限公司 | Automatic defogging method, controller, Intelligent mirror and computer-readable recording medium |
| CN109431681A (en) * | 2018-09-25 | 2019-03-08 | 吉林大学 | A kind of intelligent eyeshade and its detection method detecting sleep quality |
| US20190176333A1 (en) * | 2017-12-13 | 2019-06-13 | Disney Enterprises, Inc. | Robot navigation in context of obstacle traffic including movement of groups |
| CN110796035A (en) * | 2019-10-14 | 2020-02-14 | 上海复瞰科技有限公司 | People entering and exiting counting method based on human shape detection and speed calculation |
| CN110827313A (en) * | 2019-09-19 | 2020-02-21 | 深圳云天励飞技术有限公司 | Fast optical flow tracking method and related equipment |
| CN111507977A (en) * | 2020-04-28 | 2020-08-07 | 同济大学 | Method for extracting barium agent information in image |
| CN111724416A (en) * | 2020-06-20 | 2020-09-29 | 马鞍山职业技术学院 | Moving object detection method and trajectory tracking method based on background subtraction |
| CN112610905A (en) * | 2020-12-25 | 2021-04-06 | 中法渤海地质服务有限公司 | Offshore platform pipeline gas leakage identification method based on image identification and infrared thermal imaging technology |
| CN113542868A (en) * | 2021-05-26 | 2021-10-22 | 浙江大华技术股份有限公司 | Video key frame selection method and device, electronic equipment and storage medium |
| CN113570546A (en) * | 2021-06-16 | 2021-10-29 | 北京农业信息技术研究中心 | Fan running state detection method and device |
| CN110796035B (en) * | 2019-10-14 | 2024-05-24 | 上海复瞰科技有限公司 | People entering and exiting number counting method based on human shape detection and speed calculation |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009069996A (en) * | 2007-09-11 | 2009-04-02 | Sony Corp | Image processing device and image processing method, recognition device and recognition method, and program |
| CN102663869A (en) * | 2012-04-23 | 2012-09-12 | 国家消防工程技术研究中心 | Indoor fire detection method based on video monitoring platform |
| CN103065124A (en) * | 2012-12-24 | 2013-04-24 | 成都国科海博计算机系统有限公司 | Smoke detection method, device and fire detection device |
| CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
| CN104966045A (en) * | 2015-04-02 | 2015-10-07 | 北京天睿空间科技有限公司 | Video-based airplane entry-departure parking lot automatic detection method |
-
2016
- 2016-07-12 CN CN201610548836.4A patent/CN106204594A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009069996A (en) * | 2007-09-11 | 2009-04-02 | Sony Corp | Image processing device and image processing method, recognition device and recognition method, and program |
| CN102663869A (en) * | 2012-04-23 | 2012-09-12 | 国家消防工程技术研究中心 | Indoor fire detection method based on video monitoring platform |
| CN103065124A (en) * | 2012-12-24 | 2013-04-24 | 成都国科海博计算机系统有限公司 | Smoke detection method, device and fire detection device |
| CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
| CN104966045A (en) * | 2015-04-02 | 2015-10-07 | 北京天睿空间科技有限公司 | Video-based airplane entry-departure parking lot automatic detection method |
Non-Patent Citations (1)
| Title |
|---|
| 宋佳声: "视频序列图像中运动目标检测与跟踪算法研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570891A (en) * | 2016-11-03 | 2017-04-19 | 天津大学 | Target tracking algorithm based on video image taken by fixed camera |
| CN107048890A (en) * | 2017-05-25 | 2017-08-18 | 美的集团股份有限公司 | Automatic defogging method, controller, Intelligent mirror and computer-readable recording medium |
| US10688662B2 (en) * | 2017-12-13 | 2020-06-23 | Disney Enterprises, Inc. | Robot navigation in context of obstacle traffic including movement of groups |
| US20190176333A1 (en) * | 2017-12-13 | 2019-06-13 | Disney Enterprises, Inc. | Robot navigation in context of obstacle traffic including movement of groups |
| US11072073B2 (en) * | 2017-12-13 | 2021-07-27 | Disney Enterprises, Inc. | System for communicating and using traffic analysis in a space with moving obstacles |
| CN109431681A (en) * | 2018-09-25 | 2019-03-08 | 吉林大学 | A kind of intelligent eyeshade and its detection method detecting sleep quality |
| CN109431681B (en) * | 2018-09-25 | 2023-12-19 | 吉林大学 | Intelligent eye patch for detecting sleep quality and detection method thereof |
| CN110827313A (en) * | 2019-09-19 | 2020-02-21 | 深圳云天励飞技术有限公司 | Fast optical flow tracking method and related equipment |
| CN110827313B (en) * | 2019-09-19 | 2023-03-03 | 深圳云天励飞技术股份有限公司 | Fast optical flow tracking method and related equipment |
| CN110796035A (en) * | 2019-10-14 | 2020-02-14 | 上海复瞰科技有限公司 | People entering and exiting counting method based on human shape detection and speed calculation |
| CN110796035B (en) * | 2019-10-14 | 2024-05-24 | 上海复瞰科技有限公司 | People entering and exiting number counting method based on human shape detection and speed calculation |
| CN111507977A (en) * | 2020-04-28 | 2020-08-07 | 同济大学 | Method for extracting barium agent information in image |
| CN111507977B (en) * | 2020-04-28 | 2024-04-02 | 同济大学 | Method for extracting barium agent information in image |
| CN111724416A (en) * | 2020-06-20 | 2020-09-29 | 马鞍山职业技术学院 | Moving object detection method and trajectory tracking method based on background subtraction |
| CN112610905A (en) * | 2020-12-25 | 2021-04-06 | 中法渤海地质服务有限公司 | Offshore platform pipeline gas leakage identification method based on image identification and infrared thermal imaging technology |
| CN113542868A (en) * | 2021-05-26 | 2021-10-22 | 浙江大华技术股份有限公司 | Video key frame selection method and device, electronic equipment and storage medium |
| CN113570546B (en) * | 2021-06-16 | 2023-12-05 | 北京农业信息技术研究中心 | Fan running state detection method and device |
| CN113570546A (en) * | 2021-06-16 | 2021-10-29 | 北京农业信息技术研究中心 | Fan running state detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106204594A (en) | A kind of direction detection method of dispersivity moving object based on video image | |
| CN104392468B (en) | Based on the moving target detecting method for improving visual background extraction | |
| CN106023257B (en) | A kind of method for tracking target based on rotor wing unmanned aerial vehicle platform | |
| CN105404847B (en) | A kind of residue real-time detection method | |
| CN106022231A (en) | Multi-feature-fusion-based technical method for rapid detection of pedestrian | |
| CN108229458A (en) | A kind of intelligent flame recognition methods based on motion detection and multi-feature extraction | |
| CN107169985A (en) | A kind of moving target detecting method based on symmetrical inter-frame difference and context update | |
| CN103310194A (en) | Method for detecting head and shoulders of pedestrian in video based on overhead pixel gradient direction | |
| CN108346160A (en) | The multiple mobile object tracking combined based on disparity map Background difference and Meanshift | |
| CN103035013A (en) | Accurate moving shadow detection method based on multi-feature fusion | |
| CN110298297A (en) | Flame identification method and device | |
| CN109919053A (en) | A kind of deep learning vehicle parking detection method based on monitor video | |
| CN106023155A (en) | Online object contour tracking method based on horizontal set | |
| Lian et al. | A novel method on moving-objects detection based on background subtraction and three frames differencing | |
| CN104599291B (en) | Infrared motion target detection method based on structural similarity and significance analysis | |
| Liu et al. | Multi-type road marking recognition using adaboost detection and extreme learning machine classification | |
| CN102663778B (en) | A kind of method for tracking target based on multi-view point video and system | |
| CN106056078B (en) | Crowd density estimation method based on multi-feature regression type ensemble learning | |
| CN113822352A (en) | Infrared dim target detection method based on multi-feature fusion | |
| CN107103301B (en) | Method and system for matching discriminant color regions with maximum video target space-time stability | |
| Mahapatra et al. | Background subtraction and human detection in outdoor videos using fuzzy logic | |
| CN103077383B (en) | Based on the human motion identification method of the Divisional of spatio-temporal gradient feature | |
| Abdullah et al. | Vehicles detection system at different weather conditions | |
| CN106446832B (en) | Video-based pedestrian real-time detection method | |
| CN105139358A (en) | Video raindrop removing method and system based on combination of morphology and fuzzy C clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20161207 |