CN108734103A - The detection of moving target and tracking in satellite video - Google Patents
The detection of moving target and tracking in satellite video Download PDFInfo
- Publication number
- CN108734103A CN108734103A CN201810358177.7A CN201810358177A CN108734103A CN 108734103 A CN108734103 A CN 108734103A CN 201810358177 A CN201810358177 A CN 201810358177A CN 108734103 A CN108734103 A CN 108734103A
- Authority
- CN
- China
- Prior art keywords
- target
- vehicle
- tracking
- detection
- detected
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/207—Analysis of motion for motion estimation over a hierarchy of resolutions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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)
- Image Analysis (AREA)
Abstract
The invention belongs to images and technical field of video processing, the detection of moving target and tracking specially in satellite video.The method of the present invention includes：Using the detection technique based on movement and local Clutter modeling, to the moving target in the city in satellite video -- vehicle is detected；Using the authentication technique increased based on region with multi thread, the moving target detected is differentiated；It is associated with a plurality of track tracked using the target following technology based on Kalman filter, and by the multiple vehicle targets detected, further to ground moving object into line trace.The present invention is highly practical, and can be used for the fields such as Traffic monitoring, supervision of the cities, military monitoring, has a extensive future.
Description
Invention field
The invention belongs to image procossings and technical field of video processing, and in particular to a kind of specifically for being transported in satellite video
The detection of moving-target and tracking.
Background technology
Earth observation (EO) has wide visual field, can reach the observation scope of city size, be the heat of current research
Point and following research tendency.It is in Traffic Systems, maritime surveillance, the civilian and militaries such as spy-in-the-sky and battlefield surveillance neck
Domain plays vital effect.Current many observation technologies include Optical satellite images, satellite-borne synthetic aperture radar and boat
Empty field image gradually develops and is of great significance.Optical satellite images and satellite-borne SAR can observe high-resolution
The broad horizon of rate.But Optical satellite images and SAR image are static images, cannot observe the dynamic change of object, limit
Its application range.
As unmanned plane (UAV) develops, the aviation image that can be used is also more and more.However aerial images are because of platform
The influence of unstability fluctuation is handled to have to rely on complicated stablizing.For target following, need to realize that image registration comes
Camera motion and target movement are separated.The limited height taken photo by plane, and according to new civil aviation regulation, unmanned plane is applicable in
Range is compressed so that the range of video for acquisition of taking photo by plane greatly reduces.
Satellite video is a kind of advanced monitoring technology, it utilizes optical sensor company of acquisition on the satellite platform of movement
Continuous image.Compared with the aerial images of unmanned plane, other than the broader visual field, since the stability of satellite motion to defend
Star video is more stablized compared with video of taking photo by plane.Another advantage of satellite video, which will be exactly it, always will by the flight of itself
Alignment lens area-of-interest (ROI).
Invention content
The purpose of the present invention is to provide one kind specifically in satellite imagery video (hereinafter referred to as " satellite video ")
The detect and track method of moving target.
The present invention is the first algorithm handled specifically for satellite video, can be realized automatically to vehicle mesh in satellite video
Target detect and track.Wherein, the present invention proposes new moving object detection algorithm, and new false-alarm targets reject algorithm.
The detection of moving target and tracking in satellite video proposed by the present invention, moving target here mainly for
Vehicle is as follows：
(1) using the detection technique based on movement and local Clutter modeling, to the movement mesh in the city in satellite video
Mark -- vehicle is detected；
(2) using the authentication technique increased based on region with multi thread, the moving target detected is differentiated；
(3) Multitarget Tracking based on Kalman filter is used, and the multiple vehicle target shapes that will be detected
State is associated with a plurality of track tracked, further to ground moving object into line trace；Here, target refers to arbitrary
The vehicle detected in frame, dbjective state are that its position, speed and acceleration tie up state vector the 6 of composition, and track refers to tool
There is set of some vehicle target in the state vector of one section of continuous time, arbitrary trajectory gathers around that there are one entire video is unique
Number.
In step (1), the detection technique based on movement and local Clutter modeling assumes that original image is image
The sum of signal and noise signal, so inter-frame difference image (image that two field pictures take absolute value after subtracting each other) indicates noise
Signal, the abnormal pixel in difference image indicate vehicle target position that may be present, exponential distribution are selected to be fitted noise clutter
Distribution calculates binary-state threshold, realizes the binaryzation to inter-frame difference image, and the foreground in bianry image indicates vehicle target,
Context table in bianry image shows image background regions, and the present bit of target is determined using the intersection of continuous two width bianry image
It sets.
In step (2), the authentication technique increased based on region with multi thread, is the geometric form of reconstitution movement target
Shape differentiates moving target and false-alarm targets using variform clue.
In step (3), the target following technology based on Kalman filter, is to describe vehicle using linear model
Movement, in conjunction with observation and predicted value, estimates the current state of moving target, and update system mode.
In step (3), the multiple vehicle targets detected are associated with a plurality of track that is tracking, are to utilize
Hungary Algorithm realizes all Optimum Matchings to vehicle target and track, and partial target matches part rail after the completion of matching
Mark, to not matching one new pursuit path of object initialization of track, the track to not matching target utilizes arest neighbors
Search technique finds matched target again.
The algorithm of each step is described further below：
1, the detection technique based on movement and local Clutter modeling, the specific steps are：
(1) rasterizing, i.e.,：In each frame, along two-dimensional grid is vertically and horizontally divided an image into, each
Include several pixels in grid cell, executes detection algorithm respectively in each cell；
(1a) by image regard as original image signal and noise signal and, i.e.,：
Gi(x, y)=gi(x,y)+ni(x,y) (1)
Wherein, Gi(x, y) is the amplitude size of pixel (x, y) in the i-th frame video, gi(x, y) is the i-th frame video Central Plains
The amplitude size at beginning image midpoint (x, y), niThe amplitude size of (x, y) corresponding noise signal；
(1b) selects exponential distribution to be fitted the distribution of noise signal, and the cumulative density function (CDF) of exponential distribution is：
Wherein, cE(*) is the CDF of exponential distribution, and λ is the parameter of exponential distribution.
(1c) finds out the big region of grey scale change as target, presets false alarm rate, calculates the threshold value of binaryzation：
Wherein,It is inverse function, pfaIt is pre-set probability value；
(1d) calculates binary-state threshold by pixel classifications, before pixel amplitudes value is more than th using pre-defined probability value
Scape, it is background that pixel amplitudes value, which is less than th, and in bianry image, ' 1 ' is foreground, and ' 0 ' is background；
Obtain two bianry images are taken intersection figure by (1e), that is, carry out "AND" Boolean calculation pixel-by-pixel, before elimination
The scene target at face moment and later point, to uniquely retain moving vehicle target current location.
2, the authentication technique with multi thread is increased based on region, the specific steps are：
(2a) utilizes the region growing algorithm based on Gaussian Profile, utilizes the institute for the connected region that detector detects
There is the parameter of foreground pixel estimation Gaussian Profile；The PDF of Gaussian Profile is；
Wherein, μ and σ indicates that average value and standard deviation, the CDF of Gaussian Profile are respectively：
Wherein, erf (*) indicates dependent error functions；
(2b) predefined lower boundary probabilityWith coboundary probabilityCalculate separately lower boundary and the top of range value
Boundary, i.e.,：
Wherein,WithRespectively represent lower bound and the upper bound of threshold value；
(2c) rejudges the foreground target in this region, and range value existsPixel in range is foreground；
(2d) retains all pixels being connected with original foreground, is built into a new connected region, that is, region
The result of growth；
(2e) finally, using area, duty ratio, length and eccentricity as morphology clue, to the result of region growing
Carry out morphology discriminating, it is therefore an objective to reject the non-vehicle target in the target detected, the definition of wherein duty ratio is connected region
The ratio of the area of domain pixel and connected region bounding box.
3, the target following technology based on Kalman filter, the specific steps are：
(3a) KF applies in general to the target following of linear movement, and nonlinear motion can generally be decomposed into a series of lines
Property movement, only consider the KF tracking problems of a moving target first, the dynamical equation of forecasting system is：
Wherein, FiFor the Evolution Matrix at satellite video the i-th frame moment, viFor noise vector, xi-1It is the shape at the (i-1)-th frame moment
State vector, xiFor the state vector at current time；
State vector xiIt is defined as:
xi=[x, y, vx,vy,ax,ay]T (9)
Wherein, x, y, vx,vy,ax,ayThe position of expression X-direction respectively, the position of Y direction, the speed of X-direction,
The speed of Y direction, the acceleration of X-direction, the acceleration of Y direction；
Assuming that the linear motion of even acceleration is carried out at the time of vehicle target is between two frames, so Evolution Matrix FiFor:
Wherein, τ indicates that the time interval between two frames in video, unit are " second ", that is, the inverse of frame per second.
(3b) using the area of detection vehicle and position as observation, obtaining observational equation is：
Wherein, yi,HiAnd niIt indicates to measure vector, calculation matrix and measurement noise respectively；
Observing matrix HiFor：
(3c) exports state vector x by Bayes' theoremiPosterior probability density function (PDF), recycle chain method
Abbreviation then is carried out to it, i.e.,：
Wherein, Probability p (x is exported by Chapman-Kolmogorov equationsi|y1:i-1), i.e.,：
p(xi|y1:i-1)=∫ p (xi|xi-1)p(xi-1|y1:i-1)dxi-1 (14)
(3d) utilizes formula (13) and (14), obtains state vector xiPosterior probability density function PDF:
(3e) combines observation and predicted value, using formula (15), estimates current state, and constantly update the shape of system
State.
4, the association of multiple moving vehicles detection and trackings, the specific steps are：
(4a) by movement locus and is moved in the target following technology based on Kalman filter by Hungary Algorithm
Target is associated, and to associated target into line trace；
(4b) utilizes the Nearest Neighbor Search based on SSIM (structural similarity index), by not associated mesh
Moving target associated therewith is found in target track again, if being successfully associated, continues to complete tracking after calibration, if the rail
Mark remains unchanged no associated target, then trail termination.
Inventive algorithm precision is high, has practical value, can be used for the fields such as Traffic monitoring, supervision of the cities, military monitoring,
It has a extensive future.
Description of the drawings
Fig. 1 is the schematic diagram of an example of satellite video of the present invention.Wherein, (a) video place is located at Spain Wahlen
West Asia (b) is the respective optical map downloaded from Google Earth.
Fig. 2 is the flow chart of the detection algorithm of utilization index fitting of distribution noise clutter distribution of the present invention.
Fig. 3 is the schematic diagram of the foreground segmentation result and region growth results of moving target of the present invention and false-alarm.Wherein,
(a), (b), (c) and (d) indicates four it is assumed that the first row, the second row and the third line indicate cromogram, foreground segmentation respectively respectively
As a result with region growth results.For the ease of observation, original gray-scale map is accordingly converted into image shown in the first row.
Fig. 4 is moving vehicle sample of the present invention and non-athletic vehicle sample schematic diagram.Wherein (a), (b) and (c) be true
Moving target, still the black vehicle in (b), no significant difference compared with background are (d) false-alarm targets, it can
It can be a stationary vehicle for being similar to real moving target.
Fig. 5 is the schematic diagram of the comment field of the satellite video used in present invention experiment.
Fig. 6 is the design sketch after the annotation of the satellite video used in present invention experiment.Wherein, (a) shows the position of vehicle
It sets, (b) shows its corresponding title.Annotated in figure is representative scene.
Fig. 7 is the schematic diagram of vehicle detection and tracking result in four frame of satellite video of the present invention：Respectively the 50th frame, the
100 frames, the 150th frame, the 200th frame, each car have the number of oneself.
Fig. 8 is flow diagram of the present invention.
Specific implementation mode
First, the moving target in satellite video is detected, using the method based on movement and local Clutter modeling,
Specific implementation mode is：
Step 1：In each frame, along vertically and horizontally dividing an image into two-dimensional grid, each grid list
Include several pixels in first；
Step 2：By image regard as original image signal and noise signal and, select exponential distribution to be fitted noise
The distribution of signal；
Step 3：The big region of grey scale change is found out as target, it is 1 × 10 to preset false alarm rate-2, calculate two-value
The threshold value of change, with the threshold value by pixel classifications, i.e.,：It is foreground that pixel amplitudes value, which is more than the threshold value, anyway it is background (in two-value
In image, ' 1 ' is foreground, and ' 0 ' is background)；
Step 4：Obtain two bianry images are taken into intersection figure, that is, carry out "AND" Boolean calculation pixel-by-pixel, are eliminated
The scene target of previous instant and later point obtains testing result to uniquely retain moving vehicle target current location.
Then, the moving target in the satellite video that detects is differentiated, using being increased based on region and multi thread
Method, specific implementation mode is：
Step 1：Using the region growing algorithm based on Gaussian Profile, the connected region detected using detector
All foreground pixels estimation Gaussian Profile parameter, and then obtain the PDF of Gaussian Profile；
Step 2：The numerical value of predefined upper explanation probability and lower boundary probability, design lower bound probability is 1 × 10-2, in design
Boundary's probability is 1-1 × 10-2, the coboundary and the lower boundary that calculate separately range value rejudge this using new amplitude boundary value
The foreground target in a region；
Step 3：Retain all pixels being connected with original foreground, is built into a new connected region, obtains region
The result of growth；
Step 4：Using area, duty ratio, length and eccentricity as morphology clue, the parameter of morphology clue can
To be designed according to actual conditions, morphology discriminating is carried out to the result of region growing, rejects the non-vehicle in the target detected
Target obtains identification result.
Finally, to through detection and differentiate satellite video in moving target into line trace be associated with, utilize Kalman
The method of the target following of filter, specific implementation mode are：
Step 1：First only consider that the KF tracking problems of a moving target, the dynamical equation of forecasting system are sweared using state
Amount, obtains Evolution Matrix；
Step 2：Using the area of detection vehicle and position as observation, observational equation is obtained, and utilize measurement vector
Obtain observing matrix；
Step 3：The posterior probability density function of state vector is derived by bayes rule, and utilizes chain rule
It is simplified；
Step 4：In conjunction with observation and predicted value, current state is estimated, and update system mode；
Step 5：The vehicle target detected is associated with the track tracked using Hungary Algorithm, if
It can be associated with the state that corresponding track is so updated with dbjective state, if target initializes not as being associated with upper track
One new track is gone forward side by side line trace, if track searches again for corresponding mesh not as being associated with upper target using SSIM algorithms
Mark, if it is possible to the state for searching the state update track so with the target searched, if cannot search so eventually
Only to the tracking of this track.
It is tested on algorithm video shown in Fig. 1, which is 20 frames/second, and resolution ratio is 1 meter, when
29 seconds long, 3072 pixel of picture altitude position, picture traverse is 4096 pixels.Some regions are manually labelled with as shown in fig. 6, using this
The data of a little regional artificial's marks and the result of algorithm output carry out comparison to Quantitative Evaluation Algorithm performance, and evaluation test shows
Algorithm recall rate 63.06%, accuracy rate 81.04%, arithmetic accuracy is higher, has practical value.
Claims (5)
1. the detection of moving target and tracking in a kind of satellite video, which is characterized in that be as follows：
(1) using the detection technique based on movement and local Clutter modeling, to the moving target in the city in satellite video --
Vehicle is detected；
(2) using the authentication technique increased based on region with multi thread, the moving target detected is differentiated；
(3) Multitarget Tracking based on Kalman filter is used, and by the multiple vehicle targets detected and
The a plurality of track of tracking is associated, further to ground moving object into line trace；Here, target refers to being detected in arbitrary frame
The vehicle arrived, dbjective state refer to 6 dimension state vectors of its position, speed and acceleration composition, and track refers to having some
In the set of the state vector of one section of continuous time, arbitrary trajectory is gathered around uniquely numbers vehicle target there are one entire video；
In step (1), the detection technique based on movement and local Clutter modeling, to the fortune in the city in satellite video
Moving-target -- vehicle is detected, and assumes that original image is the sum of picture signal and noise signal, inter-frame difference image i.e. two
The graphical representation noise signal to take absolute value after frame image subtraction, the abnormal pixel in difference image indicate that vehicle target can
Position existing for energy selects exponential distribution fitting noise clutter distribution, calculates binary-state threshold, realize to inter-frame difference image
Binaryzation, the foreground in bianry image indicate that vehicle target, the context table in bianry image show image background regions, using continuous
The intersection of two width bianry images determines the current location of target；
In step (2), the authentication technique increased based on region with multi thread reflects to the moving target detected
Not, it is the geometry of reconstitution movement target, using variform clue, differentiates moving target and false-alarm targets；
In step (3), the target following technology based on Kalman filter is to describe vehicle using linear model to transport
It is dynamic, in conjunction with observation and predicted value, estimate the current state of moving target, and update system mode；
In step (3), the multiple vehicle targets detected are associated with a plurality of track that is tracking, are to utilize breast tooth
Sharp algorithm realizes all Optimum Matchings to vehicle target and track, and partial target matches partial traces after the completion of matching, right
The one new pursuit path of object initialization for not matching track, the track to not matching target utilize nearest neighbor search skill
Art finds matched target again.
2. the detection of moving target and tracking in satellite video according to claim 1, which is characterized in that step
(1) detection based on movement and local Clutter modeling in the specific steps are：
(2a) rasterizing, i.e.,：In each frame, along vertically and horizontally dividing an image into two-dimensional grid, each net
Include several pixels in lattice unit, executes detection algorithm respectively in each cell；
(2b) by image regard as original image signal and noise signal and, i.e.,：
Gi(x, y)=gi(x,y)+ni(x,y) (1)
Wherein, Gi(x, y) is the amplitude size of pixel (x, y) in the i-th frame video, gi(x, y) is original graph in the i-th frame video
As the amplitude size of midpoint (x, y), niThe amplitude size of (x, y) corresponding noise signal；
(2c) selects exponential distribution to be fitted the distribution of noise signal, and the cumulative density function (CDF) of exponential distribution is：
Wherein, cE(*) is the CDF of exponential distribution, and λ is the parameter of exponential distribution；
(2d) finds out the big region of grey scale change as target, presets false alarm rate, calculates the threshold value of binaryzation：
Wherein,It is inverse function, pfaIt is pre-set probability value；
(2e) calculates binary-state threshold by pixel classifications using pre-defined probability value, and it is foreground that pixel amplitudes value, which is more than th,
It is background that pixel amplitudes value, which is less than th, and in bianry image, ' 1 ' is foreground, and ' 0 ' is background；
Obtain two bianry images are taken intersection figure by (2f), that is, carry out "AND" Boolean calculation pixel-by-pixel, when eliminating front
The scene target with later point is carved, to uniquely retain moving vehicle target current location.
3. the detection of moving target and tracking in satellite video according to claim 1, which is characterized in that step
(2) in based on region increase with the discriminating of multi thread the specific steps are：
(3a) utilizes the region growing algorithm based on Gaussian Profile, the connected region detected using detector it is all before
Scene element estimates the parameter of Gaussian Profile；The PDF of Gaussian Profile is；
Wherein, μ and σ indicates that average value and standard deviation, the CDF of Gaussian Profile are respectively：
Wherein, erf (*) indicates dependent error functions；
(3b) predefined lower boundary probabilityWith coboundary probabilityLower boundary and the coboundary of range value are calculated separately,
I.e.：
Wherein,WithRespectively represent lower bound and the upper bound of threshold value；
(3c) rejudges the foreground target in this region, and range value existsPixel in range is foreground；
(3d) retains all pixels being connected with original foreground, is built into a new connected region, that is, region growing
Result；
(3e) finally, using area, duty ratio, length and eccentricity as morphology clue, carries out the result of region growing
Morphology differentiates, it is therefore an objective to reject the non-vehicle target in the target detected, the definition of wherein duty ratio is connected region picture
The ratio of the area of element and connected region bounding box.
4. the detection of moving target and tracking in satellite video according to claim 1, which is characterized in that step
(3) in based on Kalman filter (KF) target following the specific steps are：
(4a) KF is suitable for the target following of linear movement, and nonlinear motion is decomposed into a series of linear movements, is only examined first
Consider the KF tracking problems of a moving target, the dynamical equation of forecasting system is：
Wherein, FiFor the Evolution Matrix at satellite video the i-th frame moment, viFor noise vector, xi-1Be the (i-1)-th frame moment state to
Amount, xiFor the state vector at current time；
State vector xiIt is defined as:
xi=[x, y, vx,vy,ax,ay]T (9)
Wherein, x, y, vx,vy,ax,ayThe position of X-direction, the position of Y direction, the speed of X-direction, Y-axis side are indicated respectively
To speed, the acceleration of X-direction, the acceleration of Y direction；
Assuming that the linear motion of even acceleration is carried out at the time of vehicle target is between two frames, so Evolution Matrix FiFor:
Wherein, τ indicates that the time interval between two frames in video, unit are " second ", that is, the inverse of frame per second；
(4b) using the area of detection vehicle and position as observation, obtaining observational equation is：
Wherein, yi,HiAnd niIt indicates to measure vector, calculation matrix and measurement noise respectively；
Observing matrix HiFor：
(4c) exports state vector x by Bayes' theoremiPosterior probability density function (PDF), recycle chain rule to it
Abbreviation is carried out, i.e.,：
Wherein, Probability p (x is exported by Chapman-Kolmogorov equationsi|y1:i-1), i.e.,：
p(xi|y1:i-1)=∫ p (xi|xi-1)p(xi-1|y1:i-1)dxi-1 (14)
(4d) utilizes formula (13) and (14), obtains state vector xiPosterior probability density function PDF:
(4e) combines observation and predicted value, using formula (15), estimates current state, and constantly update the state of system.
5. the detection of moving target and tracking in satellite video according to claim 1, which is characterized in that step
(3) the multiple vehicle targets detected in and a plurality of track tracked be associated the specific steps are：
(5a) in the target following technology based on Kalman filter, by Hungary Algorithm by movement locus and moving target
It is associated, and to associated target into line trace；
(5b) utilizes the Nearest Neighbor Search based on SSIM, and movement mesh associated therewith is found in the track of not associated target again
Mark, if being successfully associated, continues to complete tracking after calibration, no associated target if the track remains unchanged, trail termination.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810358177.7A CN108734103B (en) | 2018-04-20 | 2018-04-20 | Method for detecting and tracking moving target in satellite video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810358177.7A CN108734103B (en) | 2018-04-20 | 2018-04-20 | Method for detecting and tracking moving target in satellite video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108734103A true CN108734103A (en) | 2018-11-02 |
| CN108734103B CN108734103B (en) | 2021-08-20 |
Family
ID=63939136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810358177.7A Active CN108734103B (en) | 2018-04-20 | 2018-04-20 | Method for detecting and tracking moving target in satellite video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108734103B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109708627A (en) * | 2018-11-15 | 2019-05-03 | 上海航天控制技术研究所 | A kind of moving platform down space dynamic point target rapid detection method |
| CN109712169A (en) * | 2018-11-15 | 2019-05-03 | 上海卫星工程研究所 | Moving-target motion profile prediction technique and method for autonomous tracking based on EO-1 hyperion |
| CN109816606A (en) * | 2019-01-18 | 2019-05-28 | 中国科学院电子学研究所 | A method of target following is carried out using Optical remote satellite |
| CN110057449A (en) * | 2019-04-29 | 2019-07-26 | 上海卫星工程研究所 | Aerial target height and speed calculation method based on visual field light splitting EO-1 hyperion detection |
| CN110969643A (en) * | 2019-12-18 | 2020-04-07 | 中国人民解放军国防科技大学 | On-satellite autonomous prediction method for ground target moving track |
| CN111369570A (en) * | 2020-02-24 | 2020-07-03 | 成都空御科技有限公司 | Multi-target detection tracking method for video image |
| CN111462488A (en) * | 2020-04-01 | 2020-07-28 | 北京工业大学 | Intersection safety risk assessment method based on deep convolutional neural network and intersection behavior characteristic model |
| CN112182133A (en) * | 2020-09-29 | 2021-01-05 | 南京北斗创新应用科技研究院有限公司 | AIS data-based ship loitering detection method |
| CN112990124A (en) * | 2021-04-26 | 2021-06-18 | 湖北亿咖通科技有限公司 | Vehicle tracking method and device, electronic equipment and storage medium |
| CN113283330A (en) * | 2021-05-20 | 2021-08-20 | 南京航空航天大学 | Video SAR moving target detection method based on deep learning and multi-target tracking algorithm |
| CN116030633A (en) * | 2023-02-21 | 2023-04-28 | 天津汉云工业互联网有限公司 | Vehicle tunnel early warning method and device |
| CN117315498A (en) * | 2023-10-10 | 2023-12-29 | 中国人民解放军战略支援部队航天工程大学 | False alarm discrimination method based on space target detection result |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105374026A (en) * | 2015-10-09 | 2016-03-02 | 上海海事大学 | A maritime infrared small target detection method suitable for coastal defense monitoring |
| WO2016085552A1 (en) * | 2014-10-31 | 2016-06-02 | Raytheon Company | System for real-time moving target detection using vision based image segmentation |
| CN106846374A (en) * | 2016-12-21 | 2017-06-13 | 大连海事大学 | The track calculating method of vehicle under multi-cam scene |
| CN107292911A (en) * | 2017-05-23 | 2017-10-24 | 南京邮电大学 | A kind of multi-object tracking method merged based on multi-model with data correlation |
| CN107730515A (en) * | 2017-10-12 | 2018-02-23 | 北京大学深圳研究生院 | Panoramic picture conspicuousness detection method with eye movement model is increased based on region |
-
2018
- 2018-04-20 CN CN201810358177.7A patent/CN108734103B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016085552A1 (en) * | 2014-10-31 | 2016-06-02 | Raytheon Company | System for real-time moving target detection using vision based image segmentation |
| CN105374026A (en) * | 2015-10-09 | 2016-03-02 | 上海海事大学 | A maritime infrared small target detection method suitable for coastal defense monitoring |
| CN106846374A (en) * | 2016-12-21 | 2017-06-13 | 大连海事大学 | The track calculating method of vehicle under multi-cam scene |
| CN107292911A (en) * | 2017-05-23 | 2017-10-24 | 南京邮电大学 | A kind of multi-object tracking method merged based on multi-model with data correlation |
| CN107730515A (en) * | 2017-10-12 | 2018-02-23 | 北京大学深圳研究生院 | Panoramic picture conspicuousness detection method with eye movement model is increased based on region |
Non-Patent Citations (3)
| Title |
|---|
| C.SRINIVAS RAO,ET AL: "Frame Difference And Kalman Filter Techniques For Detection Of Moving Vehicles In Video Surveillance", 《INTERNATIONAL JOURNAL OF ENGINEERING RESEARCH AND APPLICATIONS》 * |
| 李冰 等: ""基于帧差法和自适应阂值区域生长的红外运动目标检测"", 《光电技术应用》 * |
| 车少帅: "基于O2O学习的多目标检测与跟踪技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109708627A (en) * | 2018-11-15 | 2019-05-03 | 上海航天控制技术研究所 | A kind of moving platform down space dynamic point target rapid detection method |
| CN109712169A (en) * | 2018-11-15 | 2019-05-03 | 上海卫星工程研究所 | Moving-target motion profile prediction technique and method for autonomous tracking based on EO-1 hyperion |
| CN109816606A (en) * | 2019-01-18 | 2019-05-28 | 中国科学院电子学研究所 | A method of target following is carried out using Optical remote satellite |
| CN109816606B (en) * | 2019-01-18 | 2022-01-04 | 中国科学院空天信息创新研究院 | Method for tracking target by using optical remote sensing satellite |
| CN110057449A (en) * | 2019-04-29 | 2019-07-26 | 上海卫星工程研究所 | Aerial target height and speed calculation method based on visual field light splitting EO-1 hyperion detection |
| CN110969643A (en) * | 2019-12-18 | 2020-04-07 | 中国人民解放军国防科技大学 | On-satellite autonomous prediction method for ground target moving track |
| CN110969643B (en) * | 2019-12-18 | 2023-04-25 | 中国人民解放军国防科技大学 | On-board autonomous prediction method for ground target movement track |
| CN111369570A (en) * | 2020-02-24 | 2020-07-03 | 成都空御科技有限公司 | Multi-target detection tracking method for video image |
| CN111369570B (en) * | 2020-02-24 | 2023-08-18 | 成都空御科技有限公司 | Multi-target detection tracking method for video image |
| CN111462488A (en) * | 2020-04-01 | 2020-07-28 | 北京工业大学 | Intersection safety risk assessment method based on deep convolutional neural network and intersection behavior characteristic model |
| CN111462488B (en) * | 2020-04-01 | 2021-09-10 | 北京工业大学 | Intersection safety risk assessment method based on deep convolutional neural network and intersection behavior characteristic model |
| CN112182133B (en) * | 2020-09-29 | 2022-02-15 | 南京北斗创新应用科技研究院有限公司 | AIS data-based ship loitering detection method |
| CN112182133A (en) * | 2020-09-29 | 2021-01-05 | 南京北斗创新应用科技研究院有限公司 | AIS data-based ship loitering detection method |
| CN112990124B (en) * | 2021-04-26 | 2021-08-06 | 湖北亿咖通科技有限公司 | Vehicle tracking method and device, electronic equipment and storage medium |
| CN112990124A (en) * | 2021-04-26 | 2021-06-18 | 湖北亿咖通科技有限公司 | Vehicle tracking method and device, electronic equipment and storage medium |
| CN113283330A (en) * | 2021-05-20 | 2021-08-20 | 南京航空航天大学 | Video SAR moving target detection method based on deep learning and multi-target tracking algorithm |
| CN116030633B (en) * | 2023-02-21 | 2023-06-02 | 天津汉云工业互联网有限公司 | Vehicle tunnel early warning method and device |
| CN116030633A (en) * | 2023-02-21 | 2023-04-28 | 天津汉云工业互联网有限公司 | Vehicle tunnel early warning method and device |
| CN117315498A (en) * | 2023-10-10 | 2023-12-29 | 中国人民解放军战略支援部队航天工程大学 | False alarm discrimination method based on space target detection result |
| CN117315498B (en) * | 2023-10-10 | 2024-05-24 | 中国人民解放军战略支援部队航天工程大学 | False alarm discrimination method based on space target detection result |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108734103B (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108734103A (en) | The detection of moving target and tracking in satellite video | |
| CN109409283B (en) | Method, system and storage medium for tracking and monitoring sea surface ship | |
| CN110866887A (en) | Target situation fusion sensing method and system based on multiple sensors | |
| CN106204629A (en) | Space based radar and infrared data merge moving target detection method in-orbit | |
| CN103149939A (en) | Dynamic target tracking and positioning method of unmanned plane based on vision | |
| CN107144839A (en) | Pass through the long object of sensor fusion detection | |
| CN108957453A (en) | A kind of high-precision pre-filter method and recognition methods based on multiple target tracking | |
| Ahmadi et al. | A simple method for detecting and tracking vehicles and vessels from high resolution spaceborne videos | |
| CN105160649A (en) | Multi-target tracking method and system based on kernel function unsupervised clustering | |
| CN111913177A (en) | Method and device for detecting target object and storage medium | |
| CN112464933A (en) | Intelligent recognition method for small dim target of ground-based staring infrared imaging | |
| Kaya et al. | Car detection from very high-resolution UAV images using deep learning algorithms | |
| Wang et al. | A deep deformable residual learning network for SAR image segmentation | |
| CN111311640B (en) | Unmanned aerial vehicle identification and tracking method based on motion estimation | |
| CN103688289A (en) | Method and system for estimating a similarity between two binary images | |
| Yang et al. | Sar images target detection based on yolov5 | |
| CN116862832A (en) | Three-dimensional live-action model-based operator positioning method | |
| Farooq et al. | Efficient object proposals extraction for target detection in VHR remote sensing images | |
| Pavlove et al. | Efficient Deep Learning Methods for Automated Visibility Estimation at Airports | |
| Dudek et al. | Cloud detection system for uav sense and avoid: First results of cloud segmentation in a simulation environment | |
| Feng et al. | Research on information fusion method of millimetre wave radar and monocular camera for intelligent vehicle | |
| Chen et al. | Research on optimization of visual navigation algorithm for real-time obstacle avoidance of uav moving target tracking | |
| Mohammadzadeh et al. | A self-organizing fuzzy segmentation (SOFS) method for road detection from high resolution satellite images | |
| Dudek et al. | Cloud Detection System for UAV Sense and Avoid: Challenges and Findings in Flight Experiments | |
| CN115994934B (en) | Data time alignment method and device and domain controller |
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 |