CN108492318A - A method of the target following based on bionics techniques - Google Patents
A method of the target following based on bionics techniques Download PDFInfo
- Publication number
- CN108492318A CN108492318A CN201810171787.6A CN201810171787A CN108492318A CN 108492318 A CN108492318 A CN 108492318A CN 201810171787 A CN201810171787 A CN 201810171787A CN 108492318 A CN108492318 A CN 108492318A
- Authority
- CN
- China
- Prior art keywords
- target
- characteristic
- probability
- vision
- bionics techniques
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of methods of the target following based on bionics techniques, the present invention utilizes bionics techniques five kinds of visual characteristics of the motion mode of eyes and search strategy and biological vision system in object tracking process, including light and shade characteristic, color characteristics, sensitivity characteristic, persistence of vision and memory characteristic, the searching algorithm that changing features design is suitble to target following is carried out to target with these characteristics.It includes structure target template, structure candidate target template, three step of update of target template.Method for tracking target based on bionics techniques has merged intelligent processing method of the biological vision characteristic to target, it is proposed that the method for replacing color characteristic center of gravity with maximum probability center of gravity.
Description
Technical field
The invention belongs to field of machine vision, are related to mimic biology visual characteristic and carry out target signature detection and extraction, special
It is not to target based on bionical visual characteristic into line trace and identification.
Background technology
The tracking problem of moving target is always the research hotspot of scientific research field, and target following is influenced in actual life
Effect factor is very more, and current most target tracking algorism all can only be just applicable under particular circumstances, traditional movement
Target tracking algorism lack the characteristics of biological intelligence can not adapt to complex environment transformation so that tracking effect be not so good as people's will, deposit
Limitation is used prodigious, achievees the effect that biology tracking target, is always a challenge on target following direction.It is based on
The target following research of bionics techniques is the high efficiency and intelligence showed in object tracking process by biological vision system
Property enlightenment, incorporate human-eye visual characteristic thought and make up the deficiency of existing track algorithm.
This patent is played by analyzing human-eye visual characteristic in target's feature-extraction from human visual system's angle
Effect summarize color characteristic transformation, susceptibility transformation, memory characteristic convert three kinds of image conversion methods, it is proposed that Yi Zhongji
In the method for tracking target of bionics techniques.
Invention content
The processing method to extraneous visual information and bionical visual characteristic it is an object of the invention to mimic biology vision
Effect during target's feature-extraction and tracking, and memory function when to target deformation, it is proposed that imitative biological vision
The method for tracking target of characteristic.
The present invention establish bionical vision system in object tracking process the motion mode of eyes and search strategy with
And five kinds of visual characteristics of bionical vision system, including light and shade characteristic, color characteristics, sensitivity characteristic, persistence of vision and note
Recall characteristic, the searching algorithm that changing features design is suitble to target following is carried out to target with these characteristics.
The present invention is according to the light and shade characteristic of vision, color characteristics, sensitivity characteristic, persistence of vision, this five kinds of memory characteristic
Visual characteristic has simultaneously carried out eigentransformation with these characteristics to target, and clarification of objective is made to be become from RGB color feature
Destination probability feature.To target template, candidate target template and background in method for tracking target based on biological vision characteristic
Eigentransformation is done, to be scanned for best candidate model.Basic scheme is as follows:
1. first frame chooses target y0, to target using visual characteristic synthesis transformation, obtain target template P0, target organism
Probability characteristics coefficient W.
2. calculating candidate target statistical model Z using target organism probability characteristics coefficient W in next frame1, and calculate
Candidate target probability center P1, probability characteristics coefficient iterations T is set, T nearest probability characteristics coefficient iteration is selected.Such as
Fruit probability characteristics coefficient W is empty set, judges that the coefficient fails, first probability coefficent is selected to do intersection processing.Calculate background
Coefficient of variation h and given threshold h*If h<h*, new probability characteristics coefficient need to be recalculated and the probability calculated with front is special
Sign coefficient does intersection processing, if h >=h*, then into such as step 3.
3. setting iterations N, iteration threshold ε*, calculating ε=| P1-P0|.If ε>ε*, update destination probability center P0=
P1If ε≤ε*, then target area y is best candidate target.
4. being tracking target by best candidate target designation.
The background information remembered by short-term memory can help us preferably to track target.We utilize biological vision
When feature extraction target signature with background information be exactly by short-term memory store.When in face of complex environment,
With the change of environment, the background in short-term memory is also constantly changing, the most apparent highest feature of susceptibility of target
Changing, it would be desirable to which the template of a transformation replaces pervious template.Color characteristics in biological vision characteristic are also mentioned
This point, color that our human eyes can be seen according to the different adjust automaticallies of background color.
Under complex environment, the feature of target template its feature-sensitive degree can change with the variation of environment, work as target
When template characteristic disappears or weakens relative to the susceptibility of background, the tracking of target will fail.Higher organism utilizes brain
Store function and optic nerve system processing.By constantly updating target template, target signature and background is made to keep forever
The method of maximum susceptibility tracks object.
Compared with the prior art, the present invention has the following advantages:
Bionical vision system motion mode of eyes during to the tracking of target is calculated better than MeanShift target followings
The search strategy of method, in MeanShift track algorithms, the influence due to kernel function and weight function to target location so that
MeanShift is poor to the tracking effect of non-rigid object in target following, once target deforms upon, tracking is easy to lose
Effect.Method for tracking target based on bionical visual characteristic has merged biological vision characteristic to the intelligent processing method of target, carries
The method for replacing color characteristic center of gravity with maximum probability center of gravity is gone out.
Description of the drawings
Fig. 1 is the method for tracking target flow chart of the present invention based on bionics techniques.
Specific implementation mode
1. building target template, it converts the biological probability characteristics coefficient W of target to destination probability figure.In destination probability figure
In we can be apparent discovery target signature.The position of centre of gravity of destination probability figure is calculated using destination probability figure,
I.e. in target template at maximum probability center.The coordinate position of target template region x pixels is { xi *, i=
1...n },
Object pixel probability value is
Destination probability total amount:
Destination probability center:
Target organism probability characteristics coefficient:
2. building candidate target template.Candidate target template is similar with target template, i.e., selects former frame in the next frame
The coordinate position of target location region y, region y pixels isWith the target pixel probability value I of present frameyi,
Then candidate family:
Candidate target probability total amount:
Candidate target probability center:
3. finding target maximum probability position, that is, find candidate target probability center P1With the relative position of image, calculate
Then the offset vector of candidate target and target finds out best candidate target location.
Offset vector:
Bs=P1-P0 (6)
Best candidate target area:
Y=x+bs (7)
4. calibration tracking target, after finding out target location with offset vector, more accurate position is acquired with iterative processing
And it demarcates into target location.
Claims (5)
1. a kind of method of the target following based on bionics techniques, it is characterised in that:
(1) biological vision the system motion mode of eyes and search strategy and biology in object tracking process has been used to regard
Five kinds of visual characteristics of feel system, including light and shade characteristic, color characteristics, sensitivity characteristic, persistence of vision and memory characteristic, fortune
The searching algorithm that changing features design is suitble to target following is carried out to target with these characteristics;
(2) according to the light and shade characteristic of biological vision, color characteristics, sensitivity characteristic, persistence of vision, this five kinds of visions of memory characteristic
Characteristic has simultaneously carried out eigentransformation with these characteristics to target, and clarification of objective is made to be become for mesh by RGB color feature
Mark probability characteristics.Spy has been done to target template, candidate target template and background in method for tracking target based on imitative biotechnology
Sign transformation, to be scanned for best candidate model.
2. the method for a kind of target following based on bionics techniques according to claim 1, it is characterized in that by the general of target
Rate characteristic coefficient W is converted into destination probability figure, and then finds in target template at maximum probability center.
3. the method for a kind of target following based on bionics techniques according to claim 1, it is characterized in that according to candidate mesh
Mark probability total amount finds candidate target probability center, and then finds candidate target template.
4. a kind of method of target following based on bionics techniques according to claim 1, it is characterised in that:By continuous
Update target template, make target signature and background that the method for maximum susceptibility be kept to track object forever.
5. a kind of method of target following based on bionics techniques according to claim 1 and 2, it is characterised in that:Fusion
Intelligent processing method of the biological vision characteristic to target, with maximum probability center of gravity replaces color characteristic center of gravity.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810171787.6A CN108492318B (en) | 2018-03-01 | 2018-03-01 | Target tracking method based on bionic technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810171787.6A CN108492318B (en) | 2018-03-01 | 2018-03-01 | Target tracking method based on bionic technology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108492318A true CN108492318A (en) | 2018-09-04 |
CN108492318B CN108492318B (en) | 2022-04-26 |
Family
ID=63341146
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810171787.6A Active CN108492318B (en) | 2018-03-01 | 2018-03-01 | Target tracking method based on bionic technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108492318B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060078163A1 (en) * | 2002-06-07 | 2006-04-13 | Microsoft Corporation | Mode- based multi-hypothesis tracking using parametric contours |
JP5076971B2 (en) * | 2008-02-29 | 2012-11-21 | 富士通株式会社 | Image target detection and tracking apparatus, image target detection and tracking method, and image target detection and tracking program |
CN103824305A (en) * | 2014-03-17 | 2014-05-28 | 天津工业大学 | Improved Meanshift target tracking method |
CN105335986A (en) * | 2015-09-10 | 2016-02-17 | 西安电子科技大学 | Characteristic matching and MeanShift algorithm-based target tracking method |
CN106446833A (en) * | 2016-09-27 | 2017-02-22 | 湖南商学院 | Multichannel bionic vision method for recognizing complex scene image |
CN106780560A (en) * | 2016-12-29 | 2017-05-31 | 北京理工大学 | A kind of feature based merges the bionic machine fish visual tracking method of particle filter |
CN107368802A (en) * | 2017-07-14 | 2017-11-21 | 北京理工大学 | Motion target tracking method based on KCF and human brain memory mechanism |
CN107657627A (en) * | 2017-08-24 | 2018-02-02 | 北京理工大学 | Space-time contextual target tracking based on human brain memory mechanism |
-
2018
- 2018-03-01 CN CN201810171787.6A patent/CN108492318B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060078163A1 (en) * | 2002-06-07 | 2006-04-13 | Microsoft Corporation | Mode- based multi-hypothesis tracking using parametric contours |
JP5076971B2 (en) * | 2008-02-29 | 2012-11-21 | 富士通株式会社 | Image target detection and tracking apparatus, image target detection and tracking method, and image target detection and tracking program |
CN103824305A (en) * | 2014-03-17 | 2014-05-28 | 天津工业大学 | Improved Meanshift target tracking method |
CN105335986A (en) * | 2015-09-10 | 2016-02-17 | 西安电子科技大学 | Characteristic matching and MeanShift algorithm-based target tracking method |
CN106446833A (en) * | 2016-09-27 | 2017-02-22 | 湖南商学院 | Multichannel bionic vision method for recognizing complex scene image |
CN106780560A (en) * | 2016-12-29 | 2017-05-31 | 北京理工大学 | A kind of feature based merges the bionic machine fish visual tracking method of particle filter |
CN107368802A (en) * | 2017-07-14 | 2017-11-21 | 北京理工大学 | Motion target tracking method based on KCF and human brain memory mechanism |
CN107657627A (en) * | 2017-08-24 | 2018-02-02 | 北京理工大学 | Space-time contextual target tracking based on human brain memory mechanism |
Non-Patent Citations (2)
Title |
---|
QINGBIN WANG等: "Motion Control in Saccade and Smooth Pursuit for Bionic Eye Based on Three-dimensional Coordinates", 《JOURNAL OF BIONIC ENGINEERING》 * |
董慧芬等: "基于相关区域分层的改进Meanshift目标跟踪算法", 《计算机应用》 * |
Also Published As
Publication number | Publication date |
---|---|
CN108492318B (en) | 2022-04-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111797716B (en) | Single target tracking method based on Siamese network | |
CN106096561B (en) | Infrared pedestrian detection method based on image block deep learning features | |
CN105869178B (en) | A kind of complex target dynamic scene non-formaldehyde finishing method based on the convex optimization of Multiscale combination feature | |
CN104200495B (en) | A kind of multi-object tracking method in video monitoring | |
CN110060277A (en) | A kind of vision SLAM method of multiple features fusion | |
CN112784736B (en) | Character interaction behavior recognition method based on multi-modal feature fusion | |
CN111199556B (en) | Indoor pedestrian detection and tracking method based on camera | |
CN103870843B (en) | Head posture estimation method based on multi-feature-point set active shape model (ASM) | |
CN108765452A (en) | A kind of detection of mobile target in complex background and tracking | |
CN107564035B (en) | Video tracking method based on important area identification and matching | |
CN107563323A (en) | A kind of video human face characteristic point positioning method | |
CN110490913A (en) | Feature based on angle point and the marshalling of single line section describes operator and carries out image matching method | |
CN111563452A (en) | Multi-human body posture detection and state discrimination method based on example segmentation | |
CN111291701B (en) | Sight tracking method based on image gradient and ellipse fitting algorithm | |
CN105631899B (en) | A kind of ultrasound image motion target tracking method based on gray scale textural characteristics | |
CN111476077A (en) | Multi-view gait recognition method based on deep learning | |
CN112651998A (en) | Human body tracking algorithm based on attention mechanism and double-current multi-domain convolutional neural network | |
CN112364881B (en) | Advanced sampling consistency image matching method | |
CN111199245A (en) | Rape pest identification method | |
CN113608663A (en) | Fingertip tracking method based on deep learning and K-curvature method | |
CN116935133B (en) | Cardiovascular disease classification method and system based on SPECT image recognition | |
CN113327269A (en) | Unmarked cervical vertebra movement detection method | |
Schulz et al. | Object-class segmentation using deep convolutional neural networks | |
CN116777956A (en) | Moving target screening method based on multi-scale track management | |
CN116721398A (en) | Yolov5 target detection method based on cross-stage route attention module and residual information fusion module |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |