CN108664930A - A kind of intelligent multi-target detection tracking - Google Patents
A kind of intelligent multi-target detection tracking Download PDFInfo
- Publication number
- CN108664930A CN108664930A CN201810449676.7A CN201810449676A CN108664930A CN 108664930 A CN108664930 A CN 108664930A CN 201810449676 A CN201810449676 A CN 201810449676A CN 108664930 A CN108664930 A CN 108664930A
- Authority
- CN
- China
- Prior art keywords
- target
- tracking
- frame image
- max
- candidate region
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
Abstract
The present invention relates to object detecting and tracking fields, and in particular to a kind of intelligent multi-target detection tracking.A kind of intelligent multi-target detection tracking, this method are as follows：Acquisition detection target information is detected to the current frame image of acquisition；All detection target informations are subjected to data correlation, number line trace of going forward side by side；Centered on the central point of the target position of number, candidate region is chosen；The target location corresponding to candidate target is obtained in candidate region using sorter model；Judge whether candidate target is tracking target.The present invention can accurately judge whether target encounters and the situations such as block, loses or obscure；And continue to track after capable of accurately detecting target in the case where target is lost, realize the detect and track of multiple targets.
Description
Technical field
The present invention relates to object detecting and tracking fields, and in particular to a kind of intelligent multi-target detection tracking.
Background technology
It is being total to for the fields such as photoelectric guidance, target acquisition, automatic identification in mobile target in complex background detect and track
Same key technology.How automatically, accurately and quickly moving target is detected from complex background and carry out tenacious tracking, in army
Thing and civil field all have highly important status and wide application prospect.
Currently, the application environment of military field is often more complicated, as the spray, sleety weather and trees of the water surface shake meeting
Noise is generated, the difference of sunshine condition also results in image and different variations occurs, these factors all make the detection of moving target
It becomes difficult, but system requirements quickly detects and tenacious tracking.In addition, existing most of passive type radars are in target reconnaissance mistake
Cheng Zhong is easy, by electronic interferences, not detecting target position information and be supplied to tracking equipment, tracing task is caused to lose
It loses, then needing a kind of active interference-free object detection method to provide accurate coordinates of targets information.In civilian neck
Domain still has the unmanned plane detect and track under similar problem, such as low latitude background, be easy by building member, treetop,
The influence of UFO etc. leads to not the tracking for accurately detecting and stablizing.
For above-mentioned condition, traditional object detection method is high there are target in complex environment detection difficulty and accuracy is low
The problem of, lead to not normally be tracked；Occur in object tracking process simultaneously tracking lose or when blocking cannot and
When judge, or even when target reappears, there is also the problem that detection is more difficult, lead to not it is continual and steady with
Track.Therefore, it is necessary to a kind of efficient target detection tracking method be proposed, to realize that the intelligence of mobile target in complex background is examined
It surveys and tracks.
Invention content
The present invention is directed in view of the above-mentioned problems, propose one kind can to mobile target in complex background carry out intelligent measurement and
The method of tracking.
Technical program of the present invention lies in：
A kind of intelligent multi-target detection tracking, this method are as follows：
Current frame image is obtained, and acquisition detection target information is detected to the current frame image of acquisition；By all inspections
Target information is surveyed to carry out data correlation, number line trace of going forward side by side；Centered on the central point of the target position of number, choose
Candidate region；The target location corresponding to candidate target is obtained in candidate region using sorter model；
Judge whether candidate target is tracking target：
If tracking target, then use the coordinate information of tracking target in current frame image into line trace, and update grader
Model completes multi-target detection tracking；If non-tracking target, recycles above-mentioned steps, correctly target is tracked simultaneously until finding
Update sorter model.
A kind of intelligent multi-target detection tracking, this method are as follows：
Obtain current frame image；
Current frame image is detected using deep learning target detection model, and to every after current frame image detection
A target presets encumbrance and the number of interconnection；The detection target information for meeting preset condition is obtained, the preset condition includes length
Width ratio and duty ratio；
All detection target informations of acquisition are subjected to data correlation, judge to detect whether target information meets coding strip
Part；
If detection target information meets number condition, into line trace：With the center of the target position of preset numbers
Centered on point, candidate region is chosen with 2~5 times of range of target sizes；
The response diagram of candidate region is sought with sorter model, obtains the maximum response in response diagram, the peak response
Value position is the target location corresponding to candidate target；
Judge whether the target in candidate region is tracking target according to tracking failure Rule of judgment；If tracking target,
Then use the coordinate information of tracking target in current frame image into line trace, and update sorter model, complete multi-target detection with
Track；If non-tracking target, repeats the above steps, until finding correctly tracking target and updating sorter model, intelligence is completed
It can multi-target detection and tracking.
Described be detected to current frame image is completed by deep learning target detection sorter model, and this method is such as
Under：
Current frame image is detected with deep learning target detection model, obtain multiple targets and meets preset condition
Detection target correspondence probability,
Detection target of the maximum detection target of probability value as present frame in multiple targets is taken, a left side for detection target is obtained
Upper angular coordinate, width and elevation information.
The method of the target data association is as follows：
Target registration：By the location information and affinity information of target, by all targets and former frame in present frame
All targets are matched in image, if there are the target of successful match, the encumbrance of the target is constant, and the number of interconnection is incremental, then
Target information in the previous frame image of successful match is substituted for target information in current frame image；If the mesh without successful match
Mark, then the encumbrance of the target successively decreases, and the number of interconnection remains unchanged, then preserves the target letter matched in unsuccessful current frame image
Target information in breath and previous frame image；
The foundation of target registration is as follows：
In conjunction with the information of image itself, a similarity criterion is set, solves the normalized crosscorrelation between two targets
Value NCC；NCC values are the normalized crosscorrelation value between two targets；The similar journey between two targets is described using NCC values
Degree, to further determine that whether two targets match；
The calculation formula of NCC is as follows：
Wherein I1And I2Indicate that the corresponding image-region of two targets, ⊙ indicate point multiplication operation respectively；
Target is rejected：When matching unsuccessful, if the encumbrance of the target is decremented to preset value, the target is rejected；
Judge whether target meets number and require：If the number of interconnection of the target is incremented to preset value, meet label requirement,
Then the target is numbered, number order from small to large, number 0 to 9.
Into when line trace, centered on the central point of the target position of preset numbers, in current frame image, in mesh
It marks and chooses 3-7 candidate region in the range of 2-5 times of size, method is as follows：
Centered on the central point for detecting target position, the first candidate region is chosen in current frame image, first
The width and height of candidate region are respectively wide and high in previous frame image 2-2.5 times of tracking target；
On the basis of the first candidate region range size, centered on its central point, using k as scale factor, 1-3 are chosen
Candidate region, wherein 1 k≤1.5 ＜；
On the basis of the first candidate region range size, centered on its central point, selected in current frame image with 1/k times
Take 1-3 candidate region.
The method that the response diagram of candidate region is sought with sorter model is as follows：
Before training sorter model, the tracking target in initial pictures is extended, i.e., in initial pictures
2-2.5 times of range of target area is extended, the Hog feature vectors after extraction extension corresponding to target area；
According to the corresponding Hog feature vectors in target area after extension, training sorter model；The training of sorter model
Formula is as follows：
Wherein,Indicate the Fourier transformation to α,Indicate that the sorter model that training obtains, y indicate in initial pictures
The corresponding label of training sample, k indicate that kernel function, x indicate that the Hog feature vectors of extension rear region, λ are a regularization ginsengs
Number is constant, value 0.000001；Then training sample is marked using continuous label during training sorter model,
Numerical value within the scope of 0-1, and Gaussian distributed are assigned respectively to the distance at center of a sample's distance objective center, it is closer from target,
Value is more intended to 1, remoter from target, and value is more intended to 0；
Using object classifiers model, the corresponding response diagram in candidate region of multiple scales in present frame is obtained；
Wherein,Indicate that the Fourier transformation to f (z), f (z) indicate that the corresponding response diagrams of candidate region z, z expressions are worked as
The corresponding Hog feature vectors in one of candidate region in previous frame, x indicate the corresponding Hog features in target area after extension to
Amount,Indicate the sorter model that training obtains.
The determination method of target location corresponding to candidate target is as follows：
The maximum response in response diagram corresponding to 3-7 candidate region is calculated separately by sorter model, wherein the
The maximum response of one candidate region is denoted as FmaxA, using k as scale factor, the maximum response of the candidate region of selection is denoted as
FmaxA ', using 1/k as scale factor, the maximum response of the candidate region of selection is denoted as FmaxA ", wherein A are the first candidate regions
Domain, A ' are the candidate region chosen by scale factor of k, and A " is the candidate region chosen by scale factor of 1/k；
Meanwhile scale weight factor scale_weight is introduced, its value range is set between 0.9-1；
Judge FmaxWhether A is more than scale_weight and FmaxThe product of A '；
Work as FmaxA>scale_weight×FmaxWhen A ', then F is assertmaxA is maximum response Fmax’, into sentencing in next step
It is disconnected；Otherwise assert FmaxA ' is maximum response Fmax’, judge into next step, while updating the information of candidate region；
Judge Fmax’Whether scale_weight and F is more thanmaxThe product of A "；
Work as Fmax’>scale_weight×FmaxWhen A ", then F is assertmax’For maximum response Fmax, then it is directly entered next
Step；Otherwise assert FmaxA ' is maximum response Fmax, while updating the information of candidate region；
Maximum response FmaxThe position that the candidate region at place, as present frame target most probable occur.
The determination method for tracking target is as follows：
Judge candidate region maximum response FmaxWhether default response is more than, wherein the default response refers to waiting
The minimum value of maximum response in favored area, value range is between 0-1, and preferably 0.3；
As maximum response FmaxWhen more than default response, then calculates present frame and can react candidate region response diagram and shake
The APCE values for swinging degree, are denoted as APCEcurrentAnd the average APCE of target is tracked in previous frame image to the second frame image
Value, is denoted as APCEaverage；
Wherein：APCE values to seek formula as follows：
Judge the APCE of present frame candidate regioncurrentWhether the APCE of default concussion ratio is more thanaverage；
Work as APCEcurrentMore than the average APCE of default concussion ratioaverageWhen, it is believed that the candidate mesh in current frame image
It is designated as tracking target, updates sorter model；Otherwise, judge that candidate target occurs blocking, lose or ambiguity, under
One frame image carries out target detection；The default concussion ratio is between 0-1, and preferably 0.4.
The method for updating sorter model is as follows：
Target information is tracked in the information update previous frame image for tracking target in current frame image, and calculates present frame
The APCE of target is tracked in imageaverage；
Judge the F of tracking targetmaxWhether the average F of default response ratio times is more thanmax-average, set the preset ratio
Between 0-1, preferably 0.7；
In the F for judging tracking targetmaxMore than the average F of default response ratio timesmax-averageWhen, then it is directly entered next
Step judges to be determined；Otherwise, current frame image is updated without sorter model；
Judge the APCE of tracking targetaverageWhether value is more than the average APCE values of default averagely concussion ratio times, setting
Default averagely concussion ratio is between 0-1, and preferably 0.45；
When judging that the APCE values of tracking target are more than the average APCE values of default averagely concussion ratio times, then to present frame
Image carries out sorter model update；Otherwise, current frame image is updated without sorter model；
Model modification is carried out to current frame image；
Wherein：Fmax-averageFor the maximum response F of response diagram in current frame imagemaxWith response diagram in previous frame image
Maximum response FmaxAverage value；Wherein default response ratio refers to the maximum response phase of present frame tracking target area
For tracking the floating degree of target histories average response value, value range is between 0-1, and preferably 0.7；
Default averagely concussion ratio refer to by the obtained average concussion value of present frame candidate region response diagram relative to
The severe degree of target histories average response figure concussion value is tracked, value range is between 0-1, and preferably 0.45.
The formula of the model modification is as described below：
Wherein,Indicate the sorter model parameter of n-th frame image,Indicate the sorter model of the (n-1)th frame image
Parameter, η indicate Study rate parameter, value 0.015.
The technical effects of the invention are that：
When the present invention handles video image, data pass is carried out to all targets after continuous multiple frames image detection
Connection, is numbered to meeting the target of condition of data correlation, according only to numbering into line trace when the later stage tracks；Another party
Face, the present invention accurately detect target using the target detection model of deep learning, solve conventional target detection difficult and accurate
Spend low problem；Rule of judgment is lost by the tracking that target response figure oscillatory condition is set, so as to accurately judge mesh
It marks whether to encounter and the situations such as blocks, loses or obscure；And after accurately capable of detecting target in the case where target is lost
Continue to track, realizes the detect and track of multiple targets.
Description of the drawings
Fig. 1 is a kind of implementation flow chart of intelligent multi-target detection tracking of the present invention；
Fig. 2 is a kind of specific implementation block diagram of intelligent multi-target detection tracking of the present invention；
Fig. 3 is that the target after continuous multiple frames image detection carries out result figure after data correlation；
Fig. 4 is to proceed by tracing figure to the driving vehicle of number 0；
Fig. 5 is the continual and steady tracing figure of driving vehicle of number 0；
Fig. 6 targets loss causes tracking failure to reenter detection figure.
Specific implementation mode
A kind of intelligent multi-target detection tracking, this method are as follows：
Obtain current frame image；
When tracking for the first time, the initial pictures for including tracking target information, and the tracking comprising initial pictures is needed to regard
Frequently.
Current frame image is detected using deep learning target detection model, and to every after current frame image detection
A target presets encumbrance and the number of interconnection；The detection target information for meeting preset condition is obtained, the preset condition includes length
Width ratio and duty ratio；
In order to solve traditional target in complex environment detection difficult and the low situation of accuracy, the present invention mainly utilizes depth
Learn to realize to the target detection under the complex background of low latitude.Deep learning algorithm of target detection model is mainly by specified
The target image sample data of classification carries out model training, is realized pair using the powerful target's feature-extraction ability of deep learning
The detection of such target identifies, to achieve the purpose that module of target detection accurately detects.
Under normal circumstances, multiple targets in current frame image are obtained by detection model, each target correspondence belongs to finger
Determine the probability of classification.In order to improve the reliability of tracking, the detection target for meeting predetermined probabilities value is selected as present frame most
Final inspection surveys target.
Method is as follows：
Current frame image is detected with deep learning target detection model, obtain multiple targets and meets preset condition
Detection target correspondence probability；
Detection target of the maximum detection target of probability value as present frame in multiple targets is taken, a left side for detection target is obtained
Upper angular coordinate, width and elevation information.
All detection target informations of acquisition are subjected to data correlation, judge to detect whether target information meets coding strip
Part；
Target registration：By the location information and affinity information of target, by all targets and former frame in present frame
All targets are matched in image, if there are the target of successful match, the encumbrance of the target is constant, and the number of interconnection is incremental, then
Target information in the previous frame image of successful match is substituted for target information in current frame image；If the mesh without successful match
Mark, then the encumbrance of the target successively decreases, and the number of interconnection remains unchanged, then preserves the target letter matched in unsuccessful current frame image
Target information in breath and previous frame image；
The foundation of target registration is as described below：
In conjunction with the information of image itself, a similarity criterion is set, solves the normalized crosscorrelation between two targets
Value NCC；NCC values are the normalized crosscorrelation value between two targets；The similar journey between two targets is described using NCC values
Degree, to further determine that whether two targets match；
The calculation formula of NCC is as follows：
Wherein I1And I2Indicate that the corresponding image-region of two targets, ⊙ indicate point multiplication operation respectively；
Target is rejected：Using target elimination method, judge whether all targets meet rejecting foundation respectively；It matches unsuccessful
When, if the encumbrance of the target is decremented to preset value, reject the target；
Target reject foundation be：
It is not to detect between consecutive frame although the target of detection is more per frame in Multiple Targets Data Association processing
Target can successful match, it is understood that there may be the case where target or target of long-term non-successful match are lost.Therefore, it usually gives
Each target after current frame image detection presets encumbrance, which is usually a constant.If the mesh of current frame image
When mark and the success of the object matching of former frame, then the encumbrance of the target is constant；Otherwise, the encumbrance of the target successively decreases, and works as mesh
When target encumbrance is decremented to preset value, then the target is rejected.
Judge whether target meets number and require：Using target designation method, judge whether all targets meet volume respectively
Number require；If the number of interconnection of the target is incremented to preset value, meet label requirement, then the target is numbered, number is suitable
Sequence from small to large, number 0 to 9.
The foundation of target designation is：
The present invention is numbered by the way that whether the number of interconnection of target meets number condition.Usually detected to current frame image
Each target afterwards presets the number of interconnection, which is usually a constant.If the target of current frame image and the mesh of former frame
When marking successful match, then the number of interconnection of the target is incremented by；Otherwise, the number of interconnection of the target remains unchanged.When the number of interconnection of target
When being incremented to preset value, then the target is numbered.In addition, carried out successively according to sequence from small to large when target designation,
Number is from 0 to 9.
Method is as follows：
First, NCC value of the target with target in previous frame image in correspondence image region in current frame image is solved respectively,
It is most to match target to pick out the maximum target of NCC values in former frame with current block, i.e. present frame object matching success, with working as
The coordinate information of preceding target replaces the coordinate information with the maximum target of NCC values of current block in former frame；
Secondly, the encumbrance that all targets after present frame has detected are respectively set is 5.Wait for front and back frame target registration processing
After the completion, if the object matching of the target of current frame image and former frame is successful, the encumbrance of the target is still initial value；
Otherwise, the encumbrance of the target successively decreases 1.Whether the encumbrance of all targets of cycle criterion is less than or equal to 0.If the guarantor of some target
Residual is less than or equal to 0, then rejects the target.
Finally, the number of interconnection that all targets after present frame has detected are respectively set is 0.Wait for front and back frame target rejecting processing
After the completion, if the object matching of the target of current frame image and former frame is successful, the number of interconnection of the target is incremented by 1；It is no
Then, the number of interconnection of the target is still initial value.Whether the number of interconnection of all targets of cycle criterion is more than 4.If the interconnection of some target
Number is more than 4, then it is numbered.
If detection target information meets number condition, into line trace：With the center of the target position of preset numbers
Centered on point, 3-7 candidate region of selection, method are as follows in the range of 2-5 times of target sizes：
Centered on the central point for detecting target position, the first candidate region is chosen in current frame image, first
The wide and high of candidate region is respectively wide and high in previous frame image 2 times or 2.5 times of tracking target；
On the basis of the first candidate region range size, centered on its central point, using k as scale factor, 1-3 are chosen
Candidate region, wherein 1 k≤1.5 ＜；
On the basis of the first candidate region range size, centered on its central point, selected in current frame image with 1/k times
Take 1-3 candidate region.
The first candidate region is existed with tracking target centered on the central point for detecting target position in the present embodiment
Wide and high 2.5 times 1 candidate regions chosen for range in previous frame image, the second candidate region is with the first candidate region
On the basis of range size, centered on its central point, with a candidate region of 1.05 times of scale factor selection；Third is candidate
Region centered on its central point, is chosen with 1/1.05 times of scale factor on the basis of the first candidate region range size
A candidate region.
During the motion in view of target, it may occur that dimensional variation, it can be big with the range of the first candidate region
Based on small, 2 or 3 scale factors are chosen again in the range of 1 k≤1.5 ＜, such as 1.1,1.5 multiple candidates are determined with this
Region, to help sorter model accurately to obtain the precision target position corresponding to candidate target in more candidate regions
It sets.The response diagram of candidate region is sought with sorter model, obtains the maximum response in response diagram, maximum response place
Position is the target location corresponding to candidate target；
The method for calculating the response diagram of candidate region is as follows：
Before training sorter model, the tracking target in initial pictures is extended, i.e., in initial pictures
2 times of target area or 2.5 times are extended, the Hog feature vectors after extraction extension corresponding to target area；
According to the corresponding Hog feature vectors in target area after extension, training sorter model；
In view of Hog features are a kind of multidimensional characteristics, the illumination variation and dimensional variation to target have robustness, because
This trains grader by extracting Hog features to the target area after extension using this feature vector.In addition, by target
The problem of tracking, is converted into the problem of solving ridge regression model, by building the circular matrix of training sample, utilizes circular matrix
Diagonalizable characteristic in Fourier, greatly simplifies the solution procedure of ridge regression model parameter, is obtained to more quick
To object classifiers.
The training formula of sorter model is as follows：
Wherein,Indicate the Fourier transformation to α,Indicate that the sorter model that training obtains, y indicate in initial pictures
The corresponding label of training sample, k indicate that kernel function, x indicate that the Hog feature vectors of extension rear region, λ are a regularization ginsengs
Number is constant, value 0.000001；
Further, since at present major part algorithm be all using it is non-just bear by the way of mark training sample, i.e. positive sample
Label is 1, negative sample 0.The method of this marker samples has a problem in that cannot react each negative sample well
Weight, the i.e. close sample to the sample remote from target's center and from target's center are put on an equal footing.
Then training sample is marked using continuous label during training sorter model, to center of a sample apart from mesh
The far and near numerical value assigned respectively within the scope of 0-1 at mark center, and Gaussian distributed, closer from target, value is more intended to 1, from mesh
Mark is remoter, and value is more intended to 0；
Using object classifiers model, the corresponding response diagram in candidate region of multiple scales in present frame is obtained；
Wherein,Indicate that the Fourier transformation to f (z), f (z) indicate that the corresponding response diagrams of candidate region z, z expressions are worked as
The corresponding Hog feature vectors in one of candidate region in previous frame, x indicate the corresponding Hog features in target area after extension to
Amount,Indicate the sorter model that training obtains.
The determination method of target location corresponding to candidate target is as follows：
The present invention chooses three candidate regions, and the first candidate region is 1 times, the second candidate region is 1.05 times, third is waited
Favored area is 1/1.05 times of three scale size, is denoted as F respectivelymax-1.05, Fmax-1, Fmax-1/1.05；
Candidate region under three scales, which is calculated separately, by sorter model corresponds to maximum response in response diagram；
Scale weight factor scale_weight is introduced, value is set as 0.95；
Judge Fmax-1Whether scale_weight and F is more thanmax-1.05Product；Work as Fmax-1>scale_weight×
Fmax-1.05When, then by Fmax-1Regard as maximum response Fmax’, then it is directly entered and judges to be determined in next step；Otherwise will
Fmax-1.05Regard as maximum response Fmax’, also it is determined into judgement in next step, while updating the information of candidate region；
Judge Fmax’Whether scale_weight and F is more thanmax-1/1.05Product；Work as Fmax’>scale_weight×
Fmax-1/1.05When, then by Fmax’Regard as maximum response Fmax, then it is directly entered and judges to be determined in next step；Otherwise will
Fmax-1.05Regard as maximum response Fmax, while updating the information of candidate region；
It is final to determine maximum response FmaxThe candidate region at place, i.e. its position are present frame target location.
Judge whether the target in candidate region is tracking target according to tracking failure Rule of judgment；If tracking target,
Then use the coordinate information of tracking target in current frame image into line trace, and update sorter model, complete multi-target detection with
Track；If non-tracking target, repeats the above steps, until finding correctly tracking target and updating sorter model, intelligence is completed
It can multi-target detection and tracking.
The determination method for tracking target is as follows：
How to judge the degree of strength of tracker tracking stability, how accurately to judge current frame image in other words
Middle target occurs blocking or even target is lost, and loses the quality of judgment method by tracking during tracking to be assessed
Once can judge this point, the accuracy of model modification can have a distinct increment, and the stability of tracking is also strengthened.
Accurate, the maximum value of candidate target response diagram in tracking, that is, peak value, are an apparent waves
Peak, close to ideal dimensional gaussian distribution.And in the case of tracking bad, especially encounter and blocks, loses or obscure etc.
Violent oscillation can occur for the response diagram of situation, candidate target.At this point, response diagram will will appear the case where multiple peak values, cause
The center of target can not be determined by peak value of response.But it is current that target can be timely reacted by degree of oscillation
State the situations such as blocks, loses or obscure to accurately judge whether target encounters.Therefore it can react loud using one
Should figure degree of oscillation criterion APCE (average peak correlation energy)：
Judge candidate region maximum response FmaxWhether default response is more than, wherein the default response refers to waiting
The minimum value of maximum response in favored area, value range is between 0-1, and preferably 0.3；
As maximum response FmaxWhen more than default response, then calculates present frame and can react candidate region response diagram and shake
The APCE values for swinging degree, are denoted as APCEcurrentAnd the average APCE of target is tracked in previous frame image to the second frame image
Value, is denoted as APCEaverage；
Wherein：APCE values to seek formula as follows：
Wherein FmaxFor the maximum response in response diagram, FminFor the minimum response value in response diagram, Fw,hFor in response diagram
The response of corresponding position (w, h), mean are to seek mean value.It is exactly that target is blocked or mesh when APCE values reduce suddenly
The case where mark loss or even objective fuzzy.
Judge the APCE of present frame candidate regioncurrentWhether the APCE of default concussion ratio is more thanaverage；
Work as APCEcurrentMore than the average APCE of default concussion ratioaverageWhen, it is believed that the candidate mesh in current frame image
It is designated as tracking target, updates sorter model；Otherwise, judge that candidate target occurs blocking, lose or ambiguity, under
One frame image carries out target detection；The default concussion ratio is between 0-1, and preferably 0.4.
The method for updating sorter model is as follows：
The judgement of tracking result reliability is carried out by previous step, it is determined whether the tracking result of each frame is all used for more
Newly.Especially when target is blocked or tracker with it is bad when, if removing more new model again, only can make
Tracker increasingly None- identified target, to cause model drifting problem.
In addition, due to wanting to ensure tracking velocity, it is necessary to a kind of simple and effective model modification strategy, it is desirable to logical
Judged after some resources obtained, without carrying out too many complicated calculating.
Target information is tracked in the information update previous frame image for tracking target in current frame image, and calculates present frame
The APCE of target is tracked in imageaverage；
Judge the F of tracking targetmaxWhether the average F of default response ratio times is more thanmax-average, set the preset ratio
Between 0-1, preferably 0.7；
In the F for judging tracking targetmaxMore than the average F of default response ratio timesmax-averageWhen, then it is directly entered next
Step judges to be determined；Otherwise, current frame image is updated without sorter model；
Judge the APCE of tracking targetaverageWhether value is more than the average APCE values of default averagely concussion ratio times, setting
Default averagely concussion ratio is between 0-1, and preferably 0.45；
When judging that the APCE values of tracking target are more than the average APCE values of default averagely concussion ratio times, then to present frame
Image carries out sorter model update；Otherwise, current frame image is updated without sorter model；
Model modification is carried out to current frame image；
Wherein：Fmax-averageFor the maximum response F of response diagram in current frame imagemaxWith response diagram in previous frame image
Maximum response FmaxAverage value；Wherein default response ratio refers to the maximum response phase of present frame tracking target area
For tracking the floating degree of target histories average response value, value range is between 0-1, and preferably 0.7；
Default averagely concussion ratio refer to by the obtained average concussion value of present frame candidate region response diagram relative to
The severe degree of target histories average response figure concussion value is tracked, value range is between 0-1, and preferably 0.45.
The formula of the model modification is as described below：
Wherein,Indicate the sorter model parameter of n-th frame image,Indicate the sorter model of the (n-1)th frame image
Parameter, η indicate Study rate parameter, value 0.015.
If it is when tracking target, to re-start multi-target detection to judge candidate target not；Cycle executes the above method, complete
At the detection and tracking of multiple target.
Here is the verification for combining photo to carry out effect with tracking to mobile target in complex background of the present invention detection：
This video is certain traffic surveillance videos of outfield acquisition, mainly for complex scene (such as building, grove and interference
Object etc.) real-time detect and track is carried out to driving vehicle.
When starting to handle video image, data correlation mainly is carried out to the target after continuous multiple frames image detection,
Object block to meeting various criterions in data correlation is numbered, as shown in Figure 3.It can be seen from the figure that first fit is compiled
The driving vehicle of number condition has 8, numbers from 0 to 7.According to tracking condition, only the driving vehicle for meeting 0 condition of number is carried out
Tracking.
When driving vehicle meets the condition of number 0, start to the driving vehicle of number 0 into line trace, as shown in Figure 4.
At this point, this method is directly transferred to tracking, the intelligence switching of open detection and tracking from detection.After starting tracking, the traveling of number 0
Vehicle carries out stable tracking.
This method can ensure continual and steady tracking, until blocking, lose or obscuring occur in target, as shown in Figure 5.
The driving vehicle of number 0 enters continual and steady tracking mode, the terminal interface successful mark 1 of returning tracking (i=1) always.This
Outside, when being lost suddenly in object tracking process, this method can accurately judge the case where target is lost, at this time contact lost
Failure, restarting detection, as shown in Figure 6.The vehicle of number 0 is lost at turning, and this method judges in time, at this time eventually
The mark 0 (i=0) for holding interface returning tracking to lose；Start detection simultaneously, starts to be detected target to current frame image, directly
It is just transferred to tracking to the condition for meeting number 0.
Claims (10)
1. a kind of intelligent multi-target detection tracking, which is characterized in that this method is as follows：
Current frame image is obtained, and acquisition detection target information is detected to the current frame image of acquisition；
All detection target informations are subjected to data correlation, number line trace of going forward side by side；With the center of the target position of number
Centered on point, candidate region is chosen；The target location corresponding to candidate target is obtained in candidate region using sorter model；
Judge whether candidate target is tracking target：If tracking target, then the coordinate letter of tracking target in current frame image is used
It ceases into line trace, and updates sorter model, complete multi-target detection tracking；If non-tracking target recycles above-mentioned steps, directly
Target and sorter model is updated to correctly tracking is found.
2. intelligent multi-target detection tracking according to claim 1, which is characterized in that this method is as follows：
Obtain current frame image；
Current frame image is detected using deep learning target detection model, and to each mesh after current frame image detection
Mark presets encumbrance and the number of interconnection；The detection target information for meeting preset condition is obtained, the preset condition includes length-width ratio
And duty ratio；
All detection target informations of acquisition are subjected to data correlation, judge to detect whether target information meets number condition；
If detection target information meets number condition, into line trace：Central point with the target position of preset numbers is
Candidate region is chosen in center with 2-5 times of range of target sizes；
The response diagram of candidate region is sought with sorter model, obtains the maximum response in response diagram, the maximum response institute
It is the target location corresponding to candidate target in position；
Judge whether the target in candidate region is tracking target according to tracking failure Rule of judgment；If tracking target, then use
The coordinate information of target is tracked in current frame image into line trace, and updates sorter model, completes multi-target detection tracking；If
Non-tracking target, then recycle above-mentioned steps, until finding correctly tracking target and updating sorter model, completes the more mesh of intelligence
Mark detect and track.
3. the multi-target detection tracking of intelligence as claimed in claim 2, which is characterized in that described to be carried out to current frame image
Detection is completed by deep learning target detection sorter model, and this method is as follows：
Current frame image is detected with deep learning target detection model, obtain multiple targets and meets the inspection of preset condition
The correspondence probability of target is surveyed,
Detection target of the maximum detection target of probability value as present frame in multiple targets is taken, the upper left corner of detection target is obtained
Coordinate, width and elevation information.
4. intelligent multi-target detection tracking according to claim 3, which is characterized in that the target data association
Method is as follows：
Target registration：By the location information and affinity information of target, by all targets and previous frame image in present frame
In all targets matched, if there are the target of successful match, the encumbrance of the target is constant, and the number of interconnection is incremental, then will
It is substituted for target information in current frame image with target information in successful previous frame image；If the target without successful match,
The encumbrance of the target successively decreases, and the number of interconnection remains unchanged, then preserve match target information in unsuccessful current frame image and
Target information in previous frame image；
The method of target registration is as follows：
In conjunction with the information of image itself, a similarity criterion is set, solves the normalized crosscorrelation value between two targets
NCC；NCC values are the normalized crosscorrelation value between two targets；The similarity degree between two targets is described using NCC values,
To further determine that whether two targets match；
The calculation formula of NCC is as follows：
Wherein I1And I2Indicate that the corresponding image-region of two targets, ⊙ indicate point multiplication operation respectively；
Target is rejected：When matching unsuccessful, if the encumbrance of the target is decremented to preset value, the target is rejected；
Judge whether target meets number and require：If the number of interconnection of the target is incremented to preset value, meet label requirement, then it is right
The target is numbered, number order from small to large, number 0 to 9.
5. intelligent multi-target detection tracking according to claim 4, which is characterized in that into when line trace, with default
Centered on the central point of the target position of number, 3-7 candidate region is chosen in the range of 2-5 times of target sizes, just
Method is as follows：
Centered on the central point for detecting target position, the first candidate region is chosen in current frame image, first is candidate
The width and height in region are respectively wide and high in previous frame image 2-2.5 times of tracking target；
On the basis of the first candidate region range size, centered on its central point, using k as scale factor, 1-3 candidate is chosen
Region, wherein 1 k≤1.5 ＜；
On the basis of the first candidate region range size, centered on its central point, 1- is chosen in current frame image with 1/k times
3 candidate regions.
6. intelligent multi-target detection tracking according to claim 5, which is characterized in that seek waiting with sorter model
The method of the response diagram of favored area is as follows：
Before training sorter model, the tracking target in initial pictures is extended, i.e., with the target in initial pictures
The range in 2-2.5 times of region is extended, the Hog feature vectors after extraction extension corresponding to target area；
According to the corresponding Hog feature vectors in target area after extension, training sorter model；
The training formula of sorter model is as follows：
Wherein,Indicate the Fourier transformation to α,Indicate that the sorter model that training obtains, y indicate training in initial pictures
The corresponding label of sample, k indicate that kernel function, x indicate the Hog feature vectors of extension rear region, and λ is a regularization parameter；
Then training sample is marked using continuous label during training sorter model, in center of a sample's distance objective
The far and near numerical value assigned respectively within the scope of 0-1 of the heart, and Gaussian distributed, closer from target, value is more intended to 1, is got over from target
Far, value is more intended to 0；
Using object classifiers model, the corresponding response diagram in candidate region of multiple scales in present frame is obtained；
Wherein,Indicate that the Fourier transformation to f (z), f (z) indicate that the corresponding response diagrams of candidate region z, z indicate present frame
In the corresponding Hog feature vectors in one of candidate region, x indicates the corresponding Hog feature vectors in target area after extension,
Indicate the sorter model that training obtains.
7. intelligent multi-target detection tracking according to claim 6, which is characterized in that the mesh corresponding to candidate target
The determination method of cursor position is as follows：
The maximum response in response diagram corresponding to 3-7 candidate region is calculated separately by sorter model, wherein first waits
The maximum response of favored area is denoted as FmaxA, using k as scale factor, the maximum response of the candidate region of selection is denoted as FmaxA′,
Using 1/k as scale factor, the maximum response of the candidate region of selection is denoted as FmaxA ", wherein A are the first candidate region, and A ' is
Using the candidate region that k chooses as scale factor, A " is the candidate region chosen by scale factor of 1/k；
Meanwhile scale weight factor scale_weight is introduced, its value range is set between 0.9-1；
Judge FmaxWhether A is more than scale_weight and FmaxThe product of A '；
Work as FmaxA>scale_weight×FmaxWhen A ', then F is assertmaxA is maximum response Fmax’, judge into next step；It is no
Then assert FmaxA ' is maximum response Fmax’, judge into next step, while updating the information of candidate region；
Judge Fmax’Whether scale_weight and F is more thanmaxThe product of A "；
Work as Fmax’>scale_weight×FmaxWhen A ", then F is assertmax’For maximum response Fmax, then it is directly entered in next step；It is no
Then assert FmaxA ' is maximum response Fmax, while updating the information of candidate region；
Maximum response FmaxThe position that the candidate region at place, as present frame target most probable occur.
8. intelligent multi-target detection tracking according to claim 7, which is characterized in that track the determination method of target
It is as follows：
Judge candidate region maximum response FmaxWhether default response is more than, wherein the default response refers to candidate regions
The minimum value of maximum response in domain, value range is between 0-1；
As maximum response FmaxWhen more than default response, then candidate region response diagram oscillation journey can be reacted by calculating present frame
The APCE values of degree, are denoted as APCEcurrentAnd the average APCE values of target, note are tracked in previous frame image to the second frame image
For APCEaverage；
Wherein：APCE values to seek formula as follows：
Judge the APCE of present frame candidate regioncurrentWhether the APCE of default concussion ratio is more thanaverage；
Work as APCEcurrentMore than the average APCE of default concussion ratioaverageWhen, it is believed that the candidate target in current frame image is
Target is tracked, sorter model is updated；Otherwise, judge that candidate target occurs blocking, lose or ambiguity, into next frame
Image carries out target detection；The default concussion ratio is between 0-1.
9. intelligent multi-target detection tracking according to claim 8, which is characterized in that update the side of sorter model
Method is as follows：
Target information is tracked in the information update previous frame image for tracking target in current frame image, and calculates current frame image
The APCE of middle tracking targetaverage；
Judge the F of tracking targetmaxWhether the average F of default response ratio times is more thanmax-average, the preset ratio is set in 0-1
Between；
In the F for judging tracking targetmaxMore than the average F of default response ratio timesmax-averageWhen, then it is directly entered and sentences in next step
It is disconnected to be determined；Otherwise, current frame image is updated without sorter model；
Judge the APCE of tracking targetaverageWhether value is more than the average APCE values of default averagely concussion ratio times, sets default flat
Concussion ratio is between 0-1；
When judging that the APCE values of tracking target are more than the average APCE values of default averagely concussion ratio times, then to current frame image
Carry out sorter model update；Otherwise, current frame image is updated without sorter model；
Model modification is carried out to current frame image；
Wherein：Fmax-averageFor the maximum response F of response diagram in current frame imagemaxMost with response diagram in previous frame image
Big response FmaxAverage value；Wherein default response ratio refer to present frame tracking target area maximum response relative to
The floating degree of target histories average response value is tracked, value range is between 0-1；
Default averagely concussion ratio refers to the obtained average concussion value by present frame candidate region response diagram relative to tracking
The severe degree of target histories average response figure concussion value, value range is between 0-1.
10. intelligent multi-target detection tracking according to claim 9, which is characterized in that the public affairs of the model modification
Formula is as described below：
Wherein,Indicate the sorter model parameter of n-th frame image,Indicate the sorter model parameter of the (n-1)th frame image,
η indicates Study rate parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810449676.7A CN108664930A (en) | 2018-05-11 | 2018-05-11 | A kind of intelligent multi-target detection tracking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810449676.7A CN108664930A (en) | 2018-05-11 | 2018-05-11 | A kind of intelligent multi-target detection tracking |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108664930A true CN108664930A (en) | 2018-10-16 |
Family
ID=63779286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810449676.7A Pending CN108664930A (en) | 2018-05-11 | 2018-05-11 | A kind of intelligent multi-target detection tracking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108664930A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492537A (en) * | 2018-10-17 | 2019-03-19 | 桂林飞宇科技股份有限公司 | A kind of object identification method and device |
| CN109615641A (en) * | 2018-11-23 | 2019-04-12 | 中山大学 | Multiple target pedestrian tracking system and tracking based on KCF algorithm |
| CN109743497A (en) * | 2018-12-21 | 2019-05-10 | 创新奇智(重庆)科技有限公司 | A kind of dataset acquisition method, system and electronic device |
| CN110021032A (en) * | 2019-03-04 | 2019-07-16 | 五邑大学 | A kind of multi-object tracking method, device, equipment and storage medium |
| CN110148153A (en) * | 2019-04-03 | 2019-08-20 | 深圳云天励飞技术有限公司 | A kind of tracking and relevant apparatus of multiple target |
| CN110245643A (en) * | 2019-06-21 | 2019-09-17 | 上海摩象网络科技有限公司 | Target following image pickup method, device, electronic equipment |
| CN110363165A (en) * | 2019-07-18 | 2019-10-22 | 深圳大学 | Multi-object tracking method, device and storage medium based on TSK fuzzy system |
| CN110414439A (en) * | 2019-07-30 | 2019-11-05 | 武汉理工大学 | Anti- based on multi-peak detection blocks pedestrian tracting method |
| CN110533690A (en) * | 2019-08-08 | 2019-12-03 | 河海大学 | The core correlation filtering Method for Underwater Target Tracking of Fusion Features and learning rate optimization |
| CN111161323A (en) * | 2019-12-31 | 2020-05-15 | 北京理工大学重庆创新中心 | Complex scene target tracking method and system based on correlation filtering |
| CN111512317A (en) * | 2018-10-24 | 2020-08-07 | 深圳鲲云信息科技有限公司 | Multi-target real-time tracking method and device and electronic equipment |
| CN111932583A (en) * | 2020-06-05 | 2020-11-13 | 西安羚控电子科技有限公司 | Space-time information integrated intelligent tracking method based on complex background |
| CN112287794A (en) * | 2020-10-22 | 2021-01-29 | 中国电子科技集团公司第三十八研究所 | Method for managing number consistency of video image automatic identification target |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100124358A1 (en) * | 2008-11-17 | 2010-05-20 | Industrial Technology Research Institute | Method for tracking moving object |
| CN106203513A (en) * | 2016-07-08 | 2016-12-07 | 浙江工业大学 | A kind of based on pedestrian's head and shoulder multi-target detection and the statistical method of tracking |
| CN106204638A (en) * | 2016-06-29 | 2016-12-07 | 西安电子科技大学 | A kind of based on dimension self-adaption with the method for tracking target of taking photo by plane blocking process |
| CN107330920A (en) * | 2017-06-28 | 2017-11-07 | 华中科技大学 | A kind of monitor video multi-target tracking method based on deep learning |
| CN107424171A (en) * | 2017-07-21 | 2017-12-01 | 华中科技大学 | A kind of anti-shelter target tracking based on piecemeal |
| CN107516321A (en) * | 2017-07-04 | 2017-12-26 | 深圳大学 | A kind of video multi-target tracking and device |
| CN107563387A (en) * | 2017-09-14 | 2018-01-09 | 成都掌中全景信息技术有限公司 | Frame method is selected in a kind of image object detection based on Recognition with Recurrent Neural Network |
-
2018
- 2018-05-11 CN CN201810449676.7A patent/CN108664930A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100124358A1 (en) * | 2008-11-17 | 2010-05-20 | Industrial Technology Research Institute | Method for tracking moving object |
| CN106204638A (en) * | 2016-06-29 | 2016-12-07 | 西安电子科技大学 | A kind of based on dimension self-adaption with the method for tracking target of taking photo by plane blocking process |
| CN106203513A (en) * | 2016-07-08 | 2016-12-07 | 浙江工业大学 | A kind of based on pedestrian's head and shoulder multi-target detection and the statistical method of tracking |
| CN107330920A (en) * | 2017-06-28 | 2017-11-07 | 华中科技大学 | A kind of monitor video multi-target tracking method based on deep learning |
| CN107516321A (en) * | 2017-07-04 | 2017-12-26 | 深圳大学 | A kind of video multi-target tracking and device |
| CN107424171A (en) * | 2017-07-21 | 2017-12-01 | 华中科技大学 | A kind of anti-shelter target tracking based on piecemeal |
| CN107563387A (en) * | 2017-09-14 | 2018-01-09 | 成都掌中全景信息技术有限公司 | Frame method is selected in a kind of image object detection based on Recognition with Recurrent Neural Network |
Non-Patent Citations (3)
| Title |
|---|
| FAN LI 等: "Scene-Aware Adaptive Updating for Visual Tracking via Correlation Filters", 《SENSORS》 * |
| MENGMENG WANG 等: "Large Margin Object Tracking with Circulant Feature Maps", 《ARXIV》 * |
| 吴慧玲: "基于检测的在线多目标跟踪算法研究", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109492537A (en) * | 2018-10-17 | 2019-03-19 | 桂林飞宇科技股份有限公司 | A kind of object identification method and device |
| CN109492537B (en) * | 2018-10-17 | 2023-03-14 | 桂林飞宇科技股份有限公司 | Object identification method and device |
| CN111512317A (en) * | 2018-10-24 | 2020-08-07 | 深圳鲲云信息科技有限公司 | Multi-target real-time tracking method and device and electronic equipment |
| CN111512317B (en) * | 2018-10-24 | 2023-06-06 | 深圳鲲云信息科技有限公司 | Multi-target real-time tracking method and device and electronic equipment |
| CN109615641A (en) * | 2018-11-23 | 2019-04-12 | 中山大学 | Multiple target pedestrian tracking system and tracking based on KCF algorithm |
| CN109615641B (en) * | 2018-11-23 | 2022-11-29 | 中山大学 | Multi-target pedestrian tracking system and tracking method based on KCF algorithm |
| CN109743497A (en) * | 2018-12-21 | 2019-05-10 | 创新奇智(重庆)科技有限公司 | A kind of dataset acquisition method, system and electronic device |
| CN110021032A (en) * | 2019-03-04 | 2019-07-16 | 五邑大学 | A kind of multi-object tracking method, device, equipment and storage medium |
| CN110148153A (en) * | 2019-04-03 | 2019-08-20 | 深圳云天励飞技术有限公司 | A kind of tracking and relevant apparatus of multiple target |
| CN110245643A (en) * | 2019-06-21 | 2019-09-17 | 上海摩象网络科技有限公司 | Target following image pickup method, device, electronic equipment |
| CN110363165B (en) * | 2019-07-18 | 2023-04-14 | 深圳大学 | Multi-target tracking method and device based on TSK fuzzy system and storage medium |
| CN110363165A (en) * | 2019-07-18 | 2019-10-22 | 深圳大学 | Multi-object tracking method, device and storage medium based on TSK fuzzy system |
| CN110414439B (en) * | 2019-07-30 | 2022-03-15 | 武汉理工大学 | Anti-blocking pedestrian tracking method based on multi-peak detection |
| CN110414439A (en) * | 2019-07-30 | 2019-11-05 | 武汉理工大学 | Anti- based on multi-peak detection blocks pedestrian tracting method |
| CN110533690B (en) * | 2019-08-08 | 2022-02-11 | 河海大学 | Nuclear correlation filtering underwater target tracking method based on feature fusion and learning rate optimization |
| CN110533690A (en) * | 2019-08-08 | 2019-12-03 | 河海大学 | The core correlation filtering Method for Underwater Target Tracking of Fusion Features and learning rate optimization |
| CN111161323A (en) * | 2019-12-31 | 2020-05-15 | 北京理工大学重庆创新中心 | Complex scene target tracking method and system based on correlation filtering |
| CN111161323B (en) * | 2019-12-31 | 2023-11-28 | 北京理工大学重庆创新中心 | Complex scene target tracking method and system based on correlation filtering |
| CN111932583A (en) * | 2020-06-05 | 2020-11-13 | 西安羚控电子科技有限公司 | Space-time information integrated intelligent tracking method based on complex background |
| CN112287794A (en) * | 2020-10-22 | 2021-01-29 | 中国电子科技集团公司第三十八研究所 | Method for managing number consistency of video image automatic identification target |
| CN112287794B (en) * | 2020-10-22 | 2022-09-16 | 中国电子科技集团公司第三十八研究所 | Method for managing number consistency of video image automatic identification target |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108664930A (en) | A kind of intelligent multi-target detection tracking | |
| CN108765452A (en) | A kind of detection of mobile target in complex background and tracking | |
| CN108694724A (en) | A kind of long-time method for tracking target | |
| CN112132042B (en) | SAR image target detection method based on contrast domain adaptation | |
| CN112325883B (en) | Indoor positioning method for mobile robot with WiFi and visual multi-source integration | |
| Levinson et al. | Traffic light mapping, localization, and state detection for autonomous vehicles | |
| CN104112282B (en) | A method for tracking a plurality of moving objects in a monitor video based on on-line study | |
| CN108921873B (en) | Markov decision-making online multi-target tracking method based on kernel correlation filtering optimization | |
| CN110782481A (en) | Unmanned ship intelligent decision method and system | |
| CN101701818B (en) | Method for detecting long-distance barrier | |
| CN109341694A (en) | A kind of autonomous positioning air navigation aid of mobile sniffing robot | |
| CN110275153A (en) | A kind of waterborne target detection and tracking based on laser radar | |
| CN103268616A (en) | Multi-feature multi-sensor method for mobile robot to track moving body | |
| CN106570490B (en) | A kind of pedestrian's method for real time tracking based on quick clustering | |
| CN104880160B (en) | Two-dimensional-laser real-time detection method of workpiece surface profile | |
| CN105427342A (en) | Method and system for detecting and tracking underwater small-target sonar image target | |
| CN105160649A (en) | Multi-target tracking method and system based on kernel function unsupervised clustering | |
| CN110006444B (en) | Anti-interference visual odometer construction method based on optimized Gaussian mixture model | |
| CN104778699B (en) | A kind of tracking of self adaptation characteristics of objects | |
| CN108052887A (en) | A kind of doubtful illegal land automatic recognition system and method for merging SLAM/GNSS information | |
| CN109859202A (en) | A kind of deep learning detection method based on the tracking of USV water surface optical target | |
| CN112580479A (en) | Geomagnetic indoor positioning system based on cavity convolution neural network | |
| Khan et al. | An IR-UWB multi-sensor approach for collision avoidance in indoor environments | |
| CN113988228B (en) | Indoor monitoring method and system based on RFID and vision fusion | |
| CN111354016A (en) | Unmanned aerial vehicle ship tracking method and system based on deep learning and difference value hashing |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181016 |