CN107122715A - It is a kind of based on frequency when conspicuousness combine moving target detecting method - Google Patents

It is a kind of based on frequency when conspicuousness combine moving target detecting method Download PDF

Info

Publication number
CN107122715A
CN107122715A CN201710196524.6A CN201710196524A CN107122715A CN 107122715 A CN107122715 A CN 107122715A CN 201710196524 A CN201710196524 A CN 201710196524A CN 107122715 A CN107122715 A CN 107122715A
Authority
CN
China
Prior art keywords
mrow
msub
moving target
munderover
conspicuousness
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
Application number
CN201710196524.6A
Other languages
Chinese (zh)
Inventor
王辉
高菁
于立君
董泽全
魏智红
张雪
胡羽坤
丁莹
王正安
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.)
Harbin Engineering University
Original Assignee
Harbin Engineering 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 Harbin Engineering University filed Critical Harbin Engineering University
Priority to CN201710196524.6A priority Critical patent/CN107122715A/en
Publication of CN107122715A publication Critical patent/CN107122715A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/254—Analysis of motion involving subtraction of images
    • 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30232—Surveillance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/08—Detecting or categorising vehicles

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Traffic Control Systems (AREA)

Abstract

The present invention propose it is a kind of based on frequency when conspicuousness combine moving target detecting method.This method is pre-processed to image first, then the conspicuousness moving target information in frequency tuning algorithm detection image is passed through, dynamic notable figure is calculated again, and final detection result is finally obtained to doing further detection to dynamic notable figure using gauss hybrid models algorithm.The inventive method has merged frequency tuning and gauss hybrid models method, and amount of calculation is smaller, can suppress the ghost of moving target in detection image.

Description

A kind of moving target detecting method combined based on frequency-when conspicuousness
Technical field
The present invention relates to pattern-recognition and computer vision field, more particularly to a kind of combined based on frequency-when conspicuousness Moving target detecting method.
Background technology
Moving object detection in video sequence image, as the main research of video monitoring, is that current computer is regarded The study hotspot in feel field, its other research field to computer vision have important impetus.Moving object detection It is the basis for realizing the tasks such as target following, intelligent monitoring, behavioural analysis.Therefore, detect that moving target has in video image Important Research Significance and application prospect.
Moving object detection refers in sequence of video images moving target interested and background or other does not feel emerging The moving target of interest makes a distinction.Traditional moving target detecting method mainly has：Optical flow method, frame differential method and background difference Method etc..Wherein, optical flow method is to enter Mobile state by the velocity feature to each pixel in image to analyze, and obtains moving mesh Mark, it can carry out moving object detection under mobile camera, but its is computationally intensive, and need special hardware to support, It is difficult to ensure that real-time verification and measurement ratio；Frame differential method is carried out by the difference between adjacent two field picture and a kind of dynamic calculation threshold value Compare and obtain detecting target, the algorithm is preferable to environmental suitability, it may have compared with stiff stability, but during its extraction information, easily Influenceed by illumination Strength Changes, cause to detect image blurring；Background subtraction is to be entered by current frame image with background image Row difference, obtains moving target, due to background difference algorithm registration, and algorithm is simple, has also obtained wide in actual applications General application.How to build background model to play an important role in the algorithm, the method for building up of background model mainly has：Mixing is high This model, code book and moving average filter method, wherein practical application are more successfully mixed Gauss models.Mixed Gauss model The distribution of each pixel is exactly regarded as the weighting of multiple Gaussian Profiles.It can remove the environment such as most of light, noise Influence, solves the problems, such as the moving object detection under complex background.But the existing moving object detection based on mixed Gauss model The problems such as method still suffers from the ghost phenomenon caused to light sudden change, and computationally intensive under complex environment can not solve well Certainly.
The content of the invention
It is an object of the invention to provide a kind of amount of calculation is smaller, detection rates are faster based on the conspicuousness combination of frequency-when Moving target detecting method.This method is examined by fusion frequency tuning and mixed Gauss model algorithm to moving target Survey, well suppress the ghost phenomenon caused by light sudden change.
The object of the present invention is achieved like this：
A kind of moving target detecting method combined based on frequency-when conspicuousness, is comprised the following steps：
Step one：The video sequence for including multiple image is obtained, the first frame original image is extracted；
Step 2：Gaussian smoothing is carried out to original image, then will be changed by the image of Gaussian smoothing to Lab Color space, obtains the image of Lab color spaces；
Described Gaussian smoothing includes first being averaged to pixel grey scale in Image neighborhood, then to each position picture Element is weighted processing；
Step 3：By the image of frequency tuning algorithm process Lab color spaces, carry out conspicuousness moving target information and carry Take；If extracting conspicuousness moving target information, enter the calculating of Mobile state notable figure；If failing to extract conspicuousness moving target Information, then go to step 2, handles next two field picture；
Described dynamic notable figure SmExpression formula is：
Sm=(L-Lm)2+(a-Am)2+(b-Bm)2
In formula, Lm、Am、BmRespectively characteristic mean of the image in L, a, b triple channel, its expression formula is
In formula, H, W are respectively the height of original image, width, and L is brightness, and a, b are color characteristic, and (i, j) represents that pixel is sat Mark；
Step 4：Using mixed Gauss model method carry out before, background separation, detect moving target.
Described mixed Gauss model method includes following sub-step：
1. to dynamic notable figure SiIn each pixel set up k Gaussian Profile, calculating probability density function；
2. to t SiIn pixel XtMatching judgment is carried out with j-th of Gaussian Profile, wherein, j is integer and 1≤j ≤k；If matching judgment expression formula is set up, parameter renewal is carried out；If matching judgment expression formula is invalid, weights are reduced；Institute The matching judgment expression formula stated is
|Xt-μj,t-1|≤2.5σj,t-1
In formula, μj,t-1For the average of j-th of Gaussian Profile of t-1 moment；σj,t-1For j-th of Gaussian Profile of t-1 moment Standard deviation；
Described parameter more new-standard cement is
ωj,t=(1- α) × ωj,t-1+α
μj,t=(1- ρ) × μt-1+ρ×Xt
Described reduction weights expression formula is
ωj,t=(1- α) × ωj,t-1
Wherein, α is learning rate, and ρ is right value update rate, ωj,t-1The weights of j-th of Gaussian Profile of t-1 moment are represented, on Mark T is background threshold；
3. chosen by background dot, detect moving target.
The inventive method has the advantages that：
(1) the inventive method is extracted using the frequency tuning algorithm of frequency domain and is partitioned into moving target information, so as to be moved State notable figure, then as auxiliary information merged with the mixed Gaussian method in time domain progress before, background separation, moving target Accuracy of detection is high；And this method remains most frequency domain information, the ghost that can effectively suppress in moving object detection is asked Topic, the integrality of moving target is guaranteed, and can detect static target；
(2) by fusion frequency tuning and mixed Gauss model, mixed Gauss model computational complexity problem is taken into full account, Detection and parameter more new task of the mixed Gauss model algorithm to background pixel point has largely been shared, i.e., has been adjusted using frequency Humorous algorithm has shared a part of evaluation work of mixed Gauss model, reduces the amount of calculation of moving object detection, improves fortune The detection efficiency of moving-target detection.
Brief description of the drawings
Fig. 1 is the moving object detection flow chart of the inventive method.
Fig. 2 is the frame original image of monitor video the 346th at traffic light intersection.
Fig. 3 is to handle the frame original image of monitor video the 346th at traffic light intersection using common mixed Gauss model method to obtain The vehicle detection design sketch arrived.
Fig. 4 is the frame artwork of monitor video the 346th at utilization " residual spectra+mixed Gauss model method " processing traffic light intersection As obtained vehicle detection design sketch.
Fig. 5 is to handle the vehicle that the frame original image of monitor video the 346th is obtained at traffic light intersection using the inventive method to examine Survey design sketch.
Fig. 6 is the frame original image of monitor video the 938th at traffic light intersection.
Fig. 7 is to handle the frame original image of monitor video the 938th at traffic light intersection using common mixed Gauss model method to obtain The vehicle detection design sketch arrived.
Fig. 8 is the frame artwork of monitor video the 938th at utilization " residual spectra+mixed Gauss model method " processing traffic light intersection As obtained vehicle detection design sketch.
Fig. 9 is to handle the vehicle that the frame original image of monitor video the 938th is obtained at traffic light intersection using the inventive method to examine Survey design sketch.
Embodiment
The present invention is described in more detail below in conjunction with the accompanying drawings：
The present invention propose it is a kind of based on frequency when conspicuousness combine moving target detecting method, its flow as described in Figure 1, tool Body comprises the following steps：
Step one：The video sequence for including multiple image is obtained, the first two field picture is extracted；
Step 2：Gaussian smoothing is carried out to image, then will be changed by the image of Gaussian smoothing to Lab face The colour space, obtains the image of Lab color spaces；
Described Gaussian smoothing includes first being averaged to pixel grey scale in Image neighborhood, then to each position picture Element is weighted processing；
Step 3：By the image of frequency tuning algorithm process Lab color spaces, carry out conspicuousness moving target information and carry Take；If extracting conspicuousness moving target information, enter the calculating of Mobile state notable figure；If failing to extract conspicuousness moving target Information, then go to step 2, handles next two field picture；
Described dynamic notable figure SmExpression formula is：
Sm=(L-Lm)2+(a-Am)2+(b-Bm)2
In formula, Lm、Am、BmRespectively characteristic mean of the image in L, a, b triple channel, its expression formula is
In formula, H, W are respectively the height of original image, width, and L is brightness, and a, b are color characteristic, and (i, j) represents pixel Coordinate；
Step 4：Using mixed Gauss model method carry out before, background separation, detect moving target.
Described mixed Gauss model method includes following sub-step：
1. to dynamic notable figure SiIn each pixel set up k Gaussian Profile, calculating probability density function；
2. to t SiIn pixel XtMatching judgment is carried out with j-th of Gaussian Profile, wherein, j is integer and 1≤j ≤k；If matching judgment expression formula is set up, parameter renewal is carried out；If matching judgment expression formula is invalid, weights are reduced；
Described matching judgment expression formula is
|Xt-μj,t-1|≤2.5σj,t-1
In formula, μj,t-1For the average of j-th of Gaussian Profile of t-1 moment；σj,t-1For j-th of Gaussian Profile of t-1 moment Standard deviation；
Described parameter more new-standard cement is
ωj,t=(1- α) × ωj,t-1+α
μj,t=(1- ρ) × μt-1+ρ×Xt
Described reduction weights expression formula is
ωj,t=(1- α) × ωj,t-1
Wherein, α is learning rate, and ρ is right value update rate, ωkThe weights of k-th of Gauss model are represented, subscript T is background threshold Value；
3. chosen by background dot, detect moving target.
It is monitoring at 320 × 240 traffic light intersection to one section of RGB24 resolution ratio to verify the effect of the inventive method Driving vehicle in video is detected.This video totalframes is 3921 frames.Preferred parameter：Learning rate α is preferably 0.5, background Threshold value T is preferably 0.7.346th frame original image is as shown in Figure 2.The testing result of 346th frame original image is as shown in Fig. 3~Fig. 5. 938th frame original image is as shown in Figure 6.The testing result of 938th frame original image is as shown in Fig. 7~Fig. 9.
With elementary mixing Gauss model method, " residual spectra+mixed Gauss model method " as a comparison, for verifying this hair The advantage of bright method.
Fig. 3 and Fig. 7 are the moving object detection result based on mixed Gauss model method, it can be seen that this method detection the 346 frames and the 938th frame all occur in that obvious ghost；3rd row image is the motion of " residual spectra+mixed Gauss model method " The design sketch at target detection correspondence moment, hence it is evident that Detection results are very bad, can not detect moving vehicle profile completely；The Four row graphical representation the inventive method are to the 346th frame of video and the Detection results figure of the moving vehicle of the 938th frame, and the present invention is not Obvious inhibitory action is only served to ghost, the vehicle detected is also than more complete.To sum up analysis can prove that the present invention is right Ghost serves good inhibitory action.
Average time needed for the every frame of this section of video of three kinds of algorithm process is tested simultaneously：Elementary mixing Gauss model Method, " residual spectra+mixed Gauss model method " and the inventive method handle this section of video, and often average time needed for frame is respectively： 47.3125,15.4285 and 34.5072.It is " residual spectra+mixed Gauss model method " algorithm that detection speed is most fast, but by Although Fig. 4 and Fig. 8 can be seen that this blending algorithm detection rates has obvious advantage compared with other two kinds of algorithms, significantly sacrificial Domestic animal moving object detection effect, or even occur in that detection leakage phenomenon.Not only relatively conventional mixing is high in speed for this paper algorithms This model is improved, and ensure that than more fully detecting moving target.

Claims (3)

1. a kind of moving target detecting method combined based on frequency-when conspicuousness, it is characterised in that comprise the following steps：
Step one：Video sequence is obtained, the first frame original image is extracted；
Step 2：Gaussian smoothing is carried out to original image, then will be changed by the image of Gaussian smoothing to Lab colors Space, obtains the image of Lab color spaces；
Step 3：By the image of frequency tuning algorithm process Lab color spaces, conspicuousness moving target information extraction is carried out； If extracting conspicuousness moving target information, enter the calculating of Mobile state notable figure；If failing to extract conspicuousness moving target letter Breath, then go to step 2, handles next two field picture；
Step 4：Using mixed Gauss model method carry out before, background separation, detect moving target；
Described mixed Gauss model method includes following sub-step：
1. to dynamic notable figure SiIn each pixel set up k Gaussian Profile, calculating probability density function；
2. to t SiIn pixel XtMatching judgment is carried out with j-th of Gaussian Profile, wherein, j is integer and 1≤j≤k；
If matching judgment expression formula is set up, parameter renewal is carried out；If matching judgment expression formula is invalid, weights are reduced；
Described matching judgment expression formula is
|Xt-μj,t-1|≤2.5σj,t-1
In formula, μj,t-1For the average of j-th of Gaussian Profile of t-1 moment；σj,t-1For the mark of j-th of Gaussian Profile of t-1 moment It is accurate poor；
Described parameter more new-standard cement is
ωj,t=(1- α) × ωj,t-1+α
μj,t=(1- ρ) × μt-1+ρ×Xt
<mrow> <msubsup> <mi>&amp;sigma;</mi> <mrow> <mi>j</mi> <mo>,</mo> <mi>t</mi> </mrow> <mn>2</mn> </msubsup> <mo>=</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>-</mo> <mi>&amp;rho;</mi> <mo>)</mo> </mrow> <mo>&amp;times;</mo> <msubsup> <mi>&amp;sigma;</mi> <mi>t</mi> <mn>2</mn> </msubsup> <mo>+</mo> <mi>&amp;rho;</mi> <mo>&amp;times;</mo> <msup> <mrow> <mo>(</mo> <msub> <mi>X</mi> <mi>t</mi> </msub> <mo>-</mo> <msub> <mi>&amp;mu;</mi> <mrow> <mi>j</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> <mi>T</mi> </msup> <mrow> <mo>(</mo> <msub> <mi>X</mi> <mi>t</mi> </msub> <mo>-</mo> <msub> <mi>&amp;mu;</mi> <mrow> <mi>j</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>)</mo> </mrow> </mrow>
Described reduction weights expression formula is
ωj,t=(1- α) × ωj,t-1
Wherein, α is learning rate, and ρ is right value update rate, ωj,t-1Represent the weights of j-th of Gaussian Profile of t-1 moment, subscript T For background threshold；
3. chosen by background dot, detect moving target.
2. a kind of moving target detecting method combined based on frequency-when conspicuousness as claimed in claim 1, described Gauss is put down Sliding processing includes first being averaged to pixel grey scale in Image neighborhood, is then weighted processing to each position pixel.
3. a kind of moving target detecting method combined based on frequency-when conspicuousness as claimed in claim 1, described dynamic is shown Writing graph expression formula is：
Sm=(L-Lm)2+(a-Am)2+(b-Bm)2
In formula, SmFor dynamic notable figure, Lm、Am、BmRespectively characteristic mean of the image in L, a, b triple channel, its expression formula is
<mrow> <msub> <mi>L</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mi>H</mi> <mo>&amp;times;</mo> <mi>W</mi> </mrow> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>H</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>W</mi> </munderover> <mi>L</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow>
<mrow> <msub> <mi>A</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mi>H</mi> <mo>&amp;times;</mo> <mi>W</mi> </mrow> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>H</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>W</mi> </munderover> <mi>a</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow>
<mrow> <msub> <mi>B</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mi>H</mi> <mo>&amp;times;</mo> <mi>W</mi> </mrow> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>H</mi> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>W</mi> </munderover> <mi>b</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow>
In formula, H, W are respectively the height of original image, width, and L is brightness, and a, b are color characteristic, and (i, j) represents pixel coordinate.
CN201710196524.6A 2017-03-29 2017-03-29 It is a kind of based on frequency when conspicuousness combine moving target detecting method Pending CN107122715A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710196524.6A CN107122715A (en) 2017-03-29 2017-03-29 It is a kind of based on frequency when conspicuousness combine moving target detecting method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710196524.6A CN107122715A (en) 2017-03-29 2017-03-29 It is a kind of based on frequency when conspicuousness combine moving target detecting method

Publications (1)

Publication Number Publication Date
CN107122715A true CN107122715A (en) 2017-09-01

Family

ID=59718257

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710196524.6A Pending CN107122715A (en) 2017-03-29 2017-03-29 It is a kind of based on frequency when conspicuousness combine moving target detecting method

Country Status (1)

Country Link
CN (1) CN107122715A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109190474A (en) * 2018-08-01 2019-01-11 南昌大学 Human body animation extraction method of key frame based on posture conspicuousness
CN109767454A (en) * 2018-12-18 2019-05-17 西北工业大学 Based on Space Time-frequency conspicuousness unmanned plane video moving object detection method
CN111507235A (en) * 2020-04-13 2020-08-07 北京交通大学 Video-based railway perimeter foreign matter intrusion detection method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102339393A (en) * 2011-09-14 2012-02-01 电子科技大学 Target search method
US8995793B1 (en) * 2009-10-09 2015-03-31 Lockheed Martin Corporation Moving object super-resolution systems and methods
CN105631898A (en) * 2015-12-28 2016-06-01 西北工业大学 Infrared motion object detection method based on spatio-temporal saliency fusion
CN105957054A (en) * 2016-04-20 2016-09-21 北京航空航天大学 Image change detecting method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8995793B1 (en) * 2009-10-09 2015-03-31 Lockheed Martin Corporation Moving object super-resolution systems and methods
CN102339393A (en) * 2011-09-14 2012-02-01 电子科技大学 Target search method
CN105631898A (en) * 2015-12-28 2016-06-01 西北工业大学 Infrared motion object detection method based on spatio-temporal saliency fusion
CN105957054A (en) * 2016-04-20 2016-09-21 北京航空航天大学 Image change detecting method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
牛金海: "《超声原理及生物医学工程应用 生物医学超声学》", 31 January 2017, 上海交通大学出版社 *
雷帮军等: "《视频目标跟踪***分步详解》", 31 December 2015, 国防工业出版社 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109190474A (en) * 2018-08-01 2019-01-11 南昌大学 Human body animation extraction method of key frame based on posture conspicuousness
CN109190474B (en) * 2018-08-01 2021-07-20 南昌大学 Human body animation key frame extraction method based on gesture significance
CN109767454A (en) * 2018-12-18 2019-05-17 西北工业大学 Based on Space Time-frequency conspicuousness unmanned plane video moving object detection method
CN109767454B (en) * 2018-12-18 2022-05-10 西北工业大学 Unmanned aerial vehicle aerial video moving target detection method based on time-space-frequency significance
CN111507235A (en) * 2020-04-13 2020-08-07 北京交通大学 Video-based railway perimeter foreign matter intrusion detection method
CN111507235B (en) * 2020-04-13 2024-05-28 北京交通大学 Railway perimeter foreign matter intrusion detection method based on video

Similar Documents

Publication Publication Date Title
CN103177446B (en) Based on the accurate extracting method of display foreground of neighborhood and non-neighborhood smoothing prior
US7187811B2 (en) Method for image resolution enhancement
CN105608456B (en) A kind of multi-direction Method for text detection based on full convolutional network
CN107220949A (en) The self adaptive elimination method of moving vehicle shade in highway monitoring video
CN101996401B (en) Target analysis method and apparatus based on intensity image and depth image
CN107169985A (en) A kind of moving target detecting method based on symmetrical inter-frame difference and context update
CN103955949B (en) Moving target detecting method based on Mean-shift algorithm
CN107944437B (en) A kind of Face detection method based on neural network and integral image
CN106991686B (en) A kind of level set contour tracing method based on super-pixel optical flow field
CN103914699A (en) Automatic lip gloss image enhancement method based on color space
CN107480607B (en) Method for detecting and positioning standing face in intelligent recording and broadcasting system
CN104123529A (en) Human hand detection method and system thereof
CN104182983B (en) Highway monitoring video definition detection method based on corner features
CN109657612A (en) A kind of quality-ordered system and its application method based on facial image feature
CN104966285A (en) Method for detecting saliency regions
CN109166137A (en) For shake Moving Object in Video Sequences detection algorithm
CN107122715A (en) It is a kind of based on frequency when conspicuousness combine moving target detecting method
CN112613579A (en) Model training method and evaluation method for human face or human head image quality and selection method for high-quality image
CN104599291B (en) Infrared motion target detection method based on structural similarity and significance analysis
CN105488475A (en) Method for detecting human face in mobile phone
CN103514595B (en) Detection method for image salient region
CN106127813B (en) The monitor video motion segments dividing method of view-based access control model energy sensing
CN110008834B (en) Steering wheel intervention detection and statistics method based on vision
CN107392211A (en) The well-marked target detection method of the sparse cognition of view-based access control model
CN107481257A (en) The image background minimizing technology of Fusion of Color and local ternary parallel pattern feature

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for 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: 20170901