CN109993771B - Modulation domain infrared target tracking method based on state vector increment - Google Patents
Modulation domain infrared target tracking method based on state vector increment Download PDFInfo
- Publication number
- CN109993771B CN109993771B CN201711480518.XA CN201711480518A CN109993771B CN 109993771 B CN109993771 B CN 109993771B CN 201711480518 A CN201711480518 A CN 201711480518A CN 109993771 B CN109993771 B CN 109993771B
- Authority
- CN
- China
- Prior art keywords
- target
- frame
- infrared
- template
- tracking
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/251—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
-
- 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
- G06T2207/10048—Infrared image
Abstract
The invention discloses a modulation domain infrared target tracking method based on state vector increment. Aiming at the problems of low contrast, low signal-to-noise ratio, complex background change and the like in the infrared moving target tracking, the method adopts the state vector increment to expand the state vector of the moving target and stably tracks the infrared target in a modulation domain. Firstly, analyzing an input infrared video through an 18-channel Gabor filter bank, and establishing an AM-FM characteristic model of a moving target; secondly, extracting AM (amplitude modulation) significance characteristics of the moving target by adopting a significance component analysis (DCA) algorithm; then, stably tracking the moving target in the infrared video under a particle filter framework by using the AM significance characteristics of the moving target, and eliminating the interference of a complex background to the target by using the state vector increment of the target in the tracking process; secondly, judging whether the currently used target template needs to be updated, and if the target template meets the template updating condition, updating the current target template; and finally, outputting a tracking result of the current frame to realize stable tracking of the infrared target.
Description
Technical Field
The invention belongs to the technical field of infrared moving target tracking in computer vision, and particularly relates to a modulation domain infrared target tracking method based on state vector increment.
Background
The infrared target tracking technology has the characteristics of low contrast, low signal-to-noise ratio, complex background change, target mutation and the like, and has greater challenge compared with the visible light moving target tracking. In the actual tracking process, the above characteristics often cause the phenomena of deviation of tracking result, target loss and the like, and the infrared target cannot be stably tracked (1.Han, Tony X., Ming Liu, and Thomas S.Huang. "A rendering-process for tracking and connecting and obtaining." Applications of Computer Vision,2007.WACV'07.IEEE works hop on. IEEE, 2007; 2.Pan, Jiyan, and Bo Hu. "road object tracking and obtaining template driver." Image Processing,2007.ICIP 2007.IEEE International Conference on. Vol.3.IEEE, 2007.). Therefore, the moving target tracking algorithm suitable for the infrared video is still a great research hotspot in the field of computer vision.
As the modulation domain characteristics of the infrared target can distinguish a moving target from a complex background, an amplitude modulation-frequency modulation (AM-FM) target characteristic model theory can be applied to the field of infrared target tracking. The salient component analysis (DCA) algorithm can extract AM salient features from an AM-FM feature model of a target, and is therefore widely applied to the fields of Image segmentation, fingerprint Recognition, target tracking, and the like (1. havlick, J.P., P.C.Tay, and A.C.Bovik. "AM-FM Image models: functional techniques and engineering trees." Handbook of Image and Video Processing 2 (2005); 2.Prakash, R.Senthil, and Rangarajan array index for modeling-domain feature mapping. Pattern Recognition,2008. PR 2008.19 International patent application IEEE 2008, Conn.).
Chuong et al noted that in infrared target tracking, infrared moving targets and complex backgrounds can be effectively distinguished by target characteristics in the Modulation domain (Nguyen, Chuong t., and Joseph p. havlicek. "Modulation domain defects for transforming in target targets and background." Image Processing,2006IEEE International Conference on. IEEE, 2006.). The Modulation domain infrared target tracking method based on the normalized cross-correlation template proposed according to the theoretical idea can track the infrared target (Nguyen, Chuong t., Joseph p.havlicek, and Mark year. "Modulation domain mapping." Computer Vision and Pattern Recognition,2007.CVPR'07.IEEE reference on.ieee, 2007.). However, the method does not consider the influence of factors such as sudden change of the target form in the actual tracking process of the target, and has certain defects. Jesyca et al propose a target tracking method based on state vector increment, which adds speed information of a target into a target state estimation process to track the target (Bello, Jesyca C.Fuermayor, and Joseph P.Havlik. "A state vector estimation technique for estimating the localization information integration of the localization tracking filter." Image Analysis and Interpretation (SSIAI),2016IEEE Southwest Symposium on.IEEE, 2016.).
Disclosure of Invention
The invention aims to provide a modulation domain infrared target tracking method based on state vector increment. And (3) extracting the AM significance characteristics in an AM-FM characteristic model by adopting a significance component analysis (DCA) algorithm, and eliminating the interference of a complex background on the target by using the state vector increment of the target under a particle filter frame to realize the stable tracking of the infrared target. Compared with the traditional pixel domain infrared target tracking algorithm, the method can overcome the defect that the traditional pixel domain infrared target tracking is easy to lose, stably tracks the infrared target under the complex background, and has feasibility.
In order to solve the technical problem, the method for tracking the infrared target in the modulation domain based on the state vector increment effectively tracks the moving target in the infrared video sequence by adopting the following steps:
the method comprises the following steps: shooting an infrared video sequence I containing a moving object by using an infrared camera k (x, y), wherein k is 1,2, a.
Step two: input infrared video sequence I through 18-channel Gabor filter group k (x, y), k is 1,2, n, and an AM-FM characteristic model s of the moving object is established k (x, y), k ═ 1, 2.., n; the model is shown in formula (1):
in the formula (1), s k (x, y) represents an input imageI k (x, y) complex extended image, M18 represents the number of channels of the filter in the Gabor filter bank, a k,m (x, y) represents the AM function, i.e. AM characteristic, at the point (x, y) obtained by the mth Gabor filter in the kth frame image,
the FM function at the point (x, y) obtained by the mth Gabor filter in the kth frame image, i.e., the FM feature, exp represents the exponential operation, and j represents the imaginary number.
Step three, adopting a significant component analysis (DCA) algorithm to obtain an AM-FM characteristic model s of the moving target k Extracting AM significant features a of the moving target from (x, y), k 1,2 D,k (x, y), where k is 1, 2.., n, which is the AM characteristic value corresponding to the channel with the largest channel response among the 18 channels of the Gabor filter bank; defining AM significance characteristics a D And (x, y) is the AM eigenvalue corresponding to the channel with the largest channel response in 18 channels of the Gabor filter bank. The channel selection criteria are defined as:
in formula (2), Γ (x, y) represents a channel selection criterion, y m (x, y), (m ═ 1, 2.., 18) denotes the channel response of the Gabor filter bank, G m (. cndot.) denotes the frequency response of the mth Gabor filter, max denotes the maximum value, and | cndot | denotes the absolute value.
Step four: inputting the first frame I in the infrared video sequence image 1 (x, y) based on the target center coordinates (x) 1 ,y 1 ) And target width w 1 High h, h 1 True value initialization target template T 1 (m,n)；
Step five: using AM saliency features a of moving objects D,k (x, y), k 1,2, n for an infrared video sequence I in the framework of particle filtering k (x, y), k 2, 3.., n, and outputs a target center coordinate (x) k ,y k ) N and target width w k N, h, k 2,3 k N is 2,3, asTracking a result;
the posterior probability of the moving target required in the infrared target tracking process is calculated by using the method shown in the formula (3), thereby realizing the stable tracking of the infrared target,
in the formula (3), the first and second groups,
is the weight of the ith particle in the kth frame,. alpha.denotes proportional to;
is the posterior probability of the moving object obtained in the tracking process,
is the incremental state vector for the ith particle in the kth frame,
is a moving object observation; exp denotes an exponential operation, λ is an adjustable scale factor,
is the degree of similarity of the ith particle in the kth frame to the target template in that frame.
The similarity degree of the ith particle in the kth frame and the target template of the frame
The method comprises the following steps:
comparing the AM significance characteristics of the current frame target with the AM significance characteristics of the current template by using a formula (4) through a normalized cross-correlation algorithm (NCC), and solving the similarity degree of the ith particle in the kth frame and the target template of the frame
In formula (4), (M, N) is the position coordinates of the pixel points, and M, N is the width, height, T of the current frame k (m, n) is the target template in the kth frame,
is the target template in frame k-1,
is the observed value of the particle in the k-th frame,
is the observed value of the particle in the k-1 frame, Σ denotes the sum, and e denotes the belonging.
Step six: judging whether the currently used target template needs to be updated, and if the target template meets the template updating condition, turning to the step seven; otherwise, go to step five;
the template updating conditions are as follows: n is continuous D Of frames
Is greater than a set threshold th.
Step seven: and updating the target template and turning to the step five. The specific target template updating method comprises the following steps:
input image I k (x, y); calculating the similarity degree of the ith particle in the kth frame and the target template of the frame
At the same time in the particle collection
In search for the greatest degree of similarity
Particles of (2)
In that
Using optimal particles in the frame image
Calculating the degree of similarity
Selecting the frame k corresponding to the maximum similarity coefficient * Obtaining a new target template using equation (5)
Compared with the prior art, the invention has the remarkable advantages that:
1) AM salient features in the modulation domain can completely separate an infrared moving object from a complex background;
2) the state vector increment of the target expands the original state vector of the target, so that the judgment of the moving target state in the historical frame is increased, and the problem of particle diversity loss caused by particle resampling in the particle filtering process can be effectively solved;
3) the target template updating method can reduce tracking result drift caused by accumulated errors in the updating process;
the method can overcome the defect that the infrared target tracking in the traditional pixel domain is easy to lose, stably tracks the infrared target under the complex background, and has feasibility.
Drawings
FIG. 1 is a flow chart of a modulation domain infrared target tracking method based on state vector increment according to the invention.
Fig. 2 is a schematic diagram of extraction of salient features of an infrared target AM.
Fig. 3 is a schematic diagram of modulated domain infrared target tracking based on state increments.
Fig. 4 is a target template update flow diagram.
Detailed Description
Concept of first, infrared target particle filter tracking
Defining a six-element state vector x k Representing each frame image I k Moving object state in (x, y):
in the formula (1), x k ,y k The coordinates of the center position of the target are represented,
representing velocity information, δ, in the horizontal and vertical directions of the target k ,γ k Representing the size of the target morphological model.
The standard linear target state update equation can be expressed as:
x k+1 ＝Ax k +v k (2)
in the formula (2), the first and second groups,
in the formula (3), the first and second groups,
in the formula (4), | δ | < 1 represents the scale change of the target morphological model in each frame image.
In the formula (2), the first and second groups of the compound,
the standard target observation model can be expressed as:
z k ＝h(x k ,n k ) (6)
in the formula (6), n k Representing observation noise, z k Represents the observed value, and h (-) represents the observed function.
For a set of N particles
The weight of each particle is
The standard posterior probability can be expressed as:
in equation (7), the likelihood function (weight) of each particle
Can be expressed as:
in the formula (8), oc represents proportional to, exp represents exponential operation,
is the similarity degree of the ith particle in the kth frame and the target template of the frame, and can be solved by a normalized cross-correlation algorithm (NCC):
in formula (9), (M, N) is the coordinates of the pixel position, and M, N is the width and height of the current frame，T k (m, n) is the target template in the k-th frame,
is the observed value of the particle in the kth frame, Σ represents the sum, and e represents the belonging.
Finally, the center position of the moving object can be expressed as:
second, concept of AM (amplitude modulation) significant feature extraction of infrared target
Firstly, analyzing an input infrared video through an 18-channel Gabor filter bank, and establishing an AM-FM characteristic model of a moving target; and then, extracting AM salient features of the moving object by adopting a salient component analysis (DCA) algorithm. Fig. 2 is a schematic diagram of extraction of significant features of an infrared target AM, and the specific steps are as follows:
the method comprises the following steps: and establishing an AM-FM characteristic model based on the Gabor filter bank.
Obtaining image I of each frame by adopting partial Hilbert transform k Complex extended image s of (x, y) k (x,y)：
In the formula (11), j represents an imaginary number,
representing part of the Hilbert transform.
Plural extended images s in formula (11) k (x, y) can be expressed by AM-FM function rentals, namely:
in equation (11), M ═ 18 denotes the number of channels of the filter in the Gabor filter bank, and a k,m (x, y) represents the AM function, i.e. AM characteristic, at the point (x, y) obtained by the mth Gabor filter in the kth frame image,
the FM function at the point (x, y) obtained by the mth Gabor filter in the kth frame image, i.e., the FM feature, exp represents the exponential operation, and j represents the imaginary number.
Channel response y of AM-FM feature through Gabor Filter Bank m (x, y), (m ═ 1, 2.., 18) calculated as:
in the formula (13), Re [ ·]The real part of the representation function is,
the gradient is indicated.
In formula (14), G m (. cndot.) denotes the frequency response of the mth Gabor filter.
The partial Hilbert transform and AM-FM feature model building are described in detail in the literature (Havlick, Joseph P., et al, "Skewed 2D Hilbert transforms and computed AM-FM models," Image Processing,1998.ICIP 98.proceedings.1998International Conference on. Vol.1.IEEE, 1998.).
The formula (13-14) is the AM-FM characteristic model established by the invention.
Step two: and (5) extracting the AM significant features.
Defining a channel selection criterion Γ (x, y):
in the formula (15), max represents the maximum value.
Defining AM significanceCharacteristic feature a D And (x, y) is the AM eigenvalue corresponding to the channel with the largest channel response in 18 channels of the Gabor filter bank.
Modulation domain infrared target tracking based on state vector increment
The method comprises the steps of stably tracking a moving target in an infrared video under a particle filter framework by utilizing the AM (amplitude modulation) significant characteristics of the moving target, and eliminating the interference of a complex background on the target by utilizing the state vector increment of the target in the tracking process; and judging whether the currently used target template needs to be updated or not, and if the target template meets the template updating condition, updating the current target template so as to realize the stable tracking of the infrared target. Fig. 3 is a schematic diagram of modulated domain infrared target tracking based on state increment, and the specific process is as follows:
1. representation of state vector increments
Expanding the original state vector of the target, and defining the increment vector of the state vector as follows:
in equation (16), the subscript k denotes the current frame, k-1 denotes the previous frame, x k ,y k ,x k-1 ,y k-1 The coordinates of the center position of the target are represented,
representing velocity information, δ, in the horizontal and vertical directions of the target k ,γ k ,δ k-1 ,γ k-1 Representing the size of the target morphology model.
The target state transition equation based on the state vector increment can be expressed as:
in formula (17), A is the same as formula (3), v k The same as in equation (5).
The state vector increment based target observation model can be expressed as:
2. modulation domain infrared target tracking based on state vector increment
And for the input video, extracting the position information of the target in the first frame image by adopting a manual selection mode, and simultaneously extracting a target template. And tracking the target by adopting a modulation domain infrared target tracking method based on state vector increment from the second frame of image. The specific tracking process is as follows:
for each frame image I k (x, y) calculating posterior probability of the moving target required in the infrared target tracking process using the method shown in formula (19) to realize stable tracking of the infrared target,
in the formula (19), the first and second groups,
is the weight of the ith particle in the kth frame,. alpha.represents a value proportional to,
is the posterior probability of the moving object obtained in the tracking process,
is the incremental state vector for the ith particle in the kth frame,
is the corresponding observed value, exp represents the exponential operation, λ is the adjustable scale factor,
is the degree of similarity of the ith particle in the kth frame to the target template in that frame.
Equation (19)) In the k frame, the similarity degree between the ith particle and the target template in the frame
The method comprises the following steps:
comparing the AM significance characteristics of the current frame target with the AM significance characteristics of the current template by a normalized cross-correlation algorithm (NCC) by using a formula (20) to solve the similarity degree of the ith particle in the kth frame and the target template of the frame
In formula (2), (M, N) is the position coordinates of the pixel points, and M, N is the width, height, T of the current frame k (m, n) is the target template in the kth frame,
is the target template in frame k-1,
is the observed value of the particle in the k-th frame,
is the observed value of the particle in frame k-1, Σ represents the sum, and e represents the belonging.
3. Target template update
Fig. 4 is a target template update flow diagram. The specific target template updating steps are as follows:
the method comprises the following steps: input image I k (x,y)；
Step two: calculating the similarity degree of the ith particle in the kth frame and the target template of the frame
At the same time in the particle collection
In search for the greatest degree of similarity
Particles of (2)
Step three: comparing in the current frame
And updating the threshold th with the set target template if n is continuous D Of frames
If the values are larger than the set threshold value, the target template needs to be updated, and the step IV is carried out; otherwise, entering the next frame and turning to the fifth step;
step four: in that
Using optimal particles in the frame image
Calculating the degree of similarity
Selecting the frame k corresponding to the maximum similarity coefficient * Obtaining a new target template using equation (21)
Step five: and finishing the tracking of the current frame and entering the next frame.
Fourthly, a process for executing the method of the invention
The method comprises the following steps: using redAn external camera shoots an infrared video sequence I containing a moving object k (x, y), wherein k is 1,2, a.
Step two: input infrared video sequence I through 18-channel Gabor filter group k (x, y), k is 1,2, n, and an AM-FM characteristic model s of the moving object is established k (x,y),k＝1,2,...,n；
Step three, adopting a significant component analysis (DCA) algorithm to obtain an AM-FM characteristic model s of the moving target k Extracting AM significant features a of the moving target from (x, y), k 1,2 D,k (x, y), where k is 1, 2.., n, which is the AM characteristic value corresponding to the channel with the largest channel response among the 18 channels of the Gabor filter bank;
step four: inputting the first frame I in the infrared video sequence image 1 (x, y) based on the target center coordinates (x) 1 ,y 1 ) And target width w 1 High h, h 1 True value initialization target template T 1 (m,n)；
Step five: using AM saliency features a of moving objects D,k (x, y), k 1,2, n for an infrared video sequence I under the framework of particle filtering k (x, y), k 2, 3.., n, and outputs a target center coordinate (x) k ,y k ) K 2,3, n and a target width w k K 2,3, n, high h k N is used as a tracking result;
step six: judging whether the currently used target template needs to be updated, and if the target template meets the template updating condition, turning to the step seven; otherwise, go to step five;
step seven: and updating the target template and turning to the step five.
The beneficial effects of the present invention can be further illustrated by the following experiments:
in the experiment, four algorithms of a traditional pixel domain particle filter tracking algorithm (PDT), a traditional modulation domain particle filter tracking algorithm (MDT), a modulation domain infrared target tracking method based on state vector increment (no target template updating, MDVT) and a modulation domain infrared target tracking method based on state vector increment (target template updating, MDVT _ DU) are adopted for comparison.
The experimental results show that target loss finally occurs in PDT and MDT algorithms along with the distance of the target, and the MDVT and MDVT _ DU algorithms increase judgment of the moving target state in the historical frame due to the addition of the target state vector increment, so that the problem of particle diversity loss caused by particle resampling in the particle filtering process can be effectively solved, and the stable tracking of the infrared target can be realized even if the target is far away from a visual field and the size is changed.
Claims (6)
1. A modulation domain infrared target tracking method based on state vector increment is characterized by comprising the following steps:
the method comprises the following steps: shooting an infrared video sequence I containing a moving object by using an infrared camera k (x, y), wherein k is 1,2, a, n, and a moving target in the image is to be used as a target to be tracked for subsequent processing;
step two: infrared video sequence I input by 18-channel Gabor filter group k (x, y), k is 1,2, n, and an AM-FM characteristic model s of the moving object is established k (x，y)，k＝1，2，...，n；
Thirdly, an AM-FM characteristic model s of the moving target is analyzed from the moving target by adopting a significant component analysis DCA algorithm k Extracting AM significant features a of the moving target from (x, y), k 1,2 D，k (x, y), where k is 1, 2.., n, which is the AM characteristic value corresponding to the channel with the largest channel response among the 18 channels of the Gabor filter bank;
step four: inputting the first frame I in the infrared video sequence image 1 (x, y) based on the target center coordinates (x) 1 ，y 1 ) And target width w 1 High h, h 1 True value initialization target template T 1 (m，n)；
Step five: using AM saliency features a of moving objects D，k (x, y), k 1,2, n for an infrared video sequence I under the framework of particle filtering k (x, y), k 2,3, n, and outputting a target center coordinateLabel (x) k ，y k ) K 2,3, …, n and target width w k K is 2,3, …, n, h k K is 2,3, …, n is used as the tracking result; the posterior probability of the moving target required in the infrared target tracking process is calculated by using the method shown in the formula (3), thereby realizing the stable tracking of the infrared target,
in the formula (3), the first and second groups,
is the weight of the ith particle in the kth frame,. alpha.denotes proportional to;
is the posterior probability of the moving object obtained in the tracking process,
is the incremental state vector for the ith particle in the kth frame,
is a moving object observation; exp denotes an exponential operation, λ is an adjustable scale factor,
is the similarity degree of the ith particle in the kth frame and the target template of the frame;
step six: judging whether the currently used target template needs to be updated or not, and executing the step seven if the target template meets the template updating condition; otherwise, go to step five;
step seven: and updating the target template and turning to the step five.
2. The modulated domain infrared target tracking method of claim 1, characterized by: in the second step: analyzing the input infrared video by adopting an 18-channel Gabor filter bank, and establishing an AM-FM characteristic model of the moving target, wherein the model is shown as a formula (1):
in the formula (1), s k (x, y) denotes an input image I k Complex extended image of (x, y), where M ═ 18 denotes the number of channels of the filter in the Gabor filter bank, a k，m (x, y) represents the AM function, i.e. AM characteristic, at the point (x, y) obtained by the mth Gabor filter in the kth frame image,
the FM function at the point (x, y) obtained by the mth Gabor filter in the kth frame image, i.e., the FM feature, exp represents the exponential operation, and j represents the imaginary number.
3. The modulated domain infrared target tracking method of claim 1, characterized by: in the third step: extracting AM (amplitude modulation) significant characteristics of the moving target by adopting a significant component analysis (DCA) algorithm, and defining AM significant characteristics a D (x, y) is the AM eigenvalue corresponding to the channel with the largest channel response among 18 channels of the Gabor filter bank, and the channel selection criterion is defined as:
in formula (2), Γ (x, y) represents a channel selection criterion, y m (x, y), m 1,2, 18 denotes the channel response of the Gabor filterbank, G m (. cndot.) denotes the frequency response of the mth Gabor filter, max denotes the maximum value, and | cndot | denotes the absolute value.
4. The modulated domain infrared target tracking method of claim 1, characterized by: phase of ith particle in the kth frame and target template of the frameDegree of similarity
The method comprises the following steps:
comparing the AM significance characteristics of the current frame target with the AM significance characteristics of the current template by using a formula (4) through a normalized cross-correlation algorithm NCC, and solving the similarity degree of the ith particle in the kth frame and the frame target template
In formula (4), (M, N) is the position coordinates of the pixel points, and M, N is the width, height, T of the current frame k (m, n) is the target template in the k-th frame,
is the target template in frame k-1,
is the observed value of the particle in the k-th frame,
is the observed value of the particles in the (k-1) th frame.
5. The modulated domain infrared target tracking method of claim 1, characterized by: in the sixth step, the target template updating conditions are as follows: n is continuous D Of frames
Is greater than a set threshold th.
6. The modulated domain infrared target tracking method of claim 1, characterized by: the target template updating method in the step seven comprises the following steps:
the method comprises the following steps: input image I k (x，y)；
Step two: calculating the similarity degree of the ith particle in the kth frame and the target template of the frame
At the same time in the particle collection
In search for the maximum degree of similarity
Particles of (2)
Step three: in that
Using optimal particles in the frame image
Calculating the degree of similarity
Selecting the frame k corresponding to the maximum similarity coefficient * Obtaining a new target template using equation (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711480518.XA CN109993771B (en) | 2017-12-29 | 2017-12-29 | Modulation domain infrared target tracking method based on state vector increment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711480518.XA CN109993771B (en) | 2017-12-29 | 2017-12-29 | Modulation domain infrared target tracking method based on state vector increment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109993771A CN109993771A (en) | 2019-07-09 |
| CN109993771B true CN109993771B (en) | 2022-09-13 |
Family
ID=67109111
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711480518.XA Active CN109993771B (en) | 2017-12-29 | 2017-12-29 | Modulation domain infrared target tracking method based on state vector increment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109993771B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115205510B (en) * | 2022-07-12 | 2023-04-07 | 中国科学院长春光学精密机械与物理研究所 | Complex scene infrared point target identification method based on saliency characteristic information discrimination |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105389807A (en) * | 2015-10-26 | 2016-03-09 | 南京理工大学 | Particle filter infrared tracking method with fusion of gradient feature and adaptive template |
| CN107146240A (en) * | 2017-05-05 | 2017-09-08 | 西北工业大学 | The video target tracking method of taking photo by plane detected based on correlation filtering and conspicuousness |
-
2017
- 2017-12-29 CN CN201711480518.XA patent/CN109993771B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105389807A (en) * | 2015-10-26 | 2016-03-09 | 南京理工大学 | Particle filter infrared tracking method with fusion of gradient feature and adaptive template |
| CN107146240A (en) * | 2017-05-05 | 2017-09-08 | 西北工业大学 | The video target tracking method of taking photo by plane detected based on correlation filtering and conspicuousness |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109993771A (en) | 2019-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Infrared dim target detection based on total variation regularization and principal component pursuit | |
| CN108665481B (en) | Self-adaptive anti-blocking infrared target tracking method based on multi-layer depth feature fusion | |
| CN103871076B (en) | Extracting of Moving Object based on optical flow method and super-pixel segmentation | |
| CN108446634B (en) | Aircraft continuous tracking method based on combination of video analysis and positioning information | |
| CN111028292B (en) | Sub-pixel level image matching navigation positioning method | |
| CN112184752A (en) | Video target tracking method based on pyramid convolution | |
| CN106709472A (en) | Video target detecting and tracking method based on optical flow features | |
| CN110490907B (en) | Moving target tracking method based on multi-target feature and improved correlation filter | |
| CN103268480A (en) | System and method for visual tracking | |
| CN110738690A (en) | unmanned aerial vehicle video middle vehicle speed correction method based on multi-target tracking framework | |
| CN108257155B (en) | Extended target stable tracking point extraction method based on local and global coupling | |
| CN109448023B (en) | Satellite video small target real-time tracking method | |
| CN110827262B (en) | Weak and small target detection method based on continuous limited frame infrared image | |
| CN104484890A (en) | Video target tracking method based on compound sparse model | |
| CN109410248B (en) | Flotation froth motion characteristic extraction method based on r-K algorithm | |
| CN105869174A (en) | Sky scene image segmentation method | |
| CN111402303A (en) | Target tracking architecture based on KFSTRCF | |
| CN112785626A (en) | Twin network small target tracking method based on multi-scale feature fusion | |
| CN109993771B (en) | Modulation domain infrared target tracking method based on state vector increment | |
| CN112991394B (en) | KCF target tracking method based on cubic spline interpolation and Markov chain | |
| Li et al. | Moving object detection in dynamic scenes based on optical flow and superpixels | |
| Angelo | A novel approach on object detection and tracking using adaptive background subtraction method | |
| CN105741317B (en) | Infrared motion target detection method based on time-space domain significance analysis and rarefaction representation | |
| CN110111368B (en) | Human body posture recognition-based similar moving target detection and tracking method | |
| CN107564029B (en) | Moving target detection method based on Gaussian extreme value filtering and group sparse RPCA |
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 |