CN109215025B - Infrared weak and small target detection method based on non-convex rank approach minimization - Google Patents
Infrared weak and small target detection method based on non-convex rank approach minimization Download PDFInfo
- Publication number
- CN109215025B CN109215025B CN201811116093.9A CN201811116093A CN109215025B CN 109215025 B CN109215025 B CN 109215025B CN 201811116093 A CN201811116093 A CN 201811116093A CN 109215025 B CN109215025 B CN 109215025B
- Authority
- CN
- China
- Prior art keywords
- target
- image
- infrared
- block image
- norm
- 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
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an infrared weak and small target detection method based on non-convex rank approach minimization, belonging to the field of infrared image processing and target detection; which comprises the following steps of 1: traversing the original image by adopting a sliding window to construct an infrared block image; step 2: constructing a target function by utilizing non-convex rank approximation minimization, inputting the infrared block image into the target function, and solving the target function by utilizing an augmented Lagrange multiplier method and a differential convex programming method to obtain a background block image and a target block image; and step 3: reconstructing a background image and a target image according to the background block image and the target block image; and 4, step 4: performing threshold segmentation on the target image to determine the position of the target, and outputting a target detection result; the method solves the problem that the detection accuracy of the infrared weak and small target is low due to the fact that factors such as strong edges, partial noise, a false alarm source and the like have sparse characteristics in the existing IPI method, and achieves the effect of inhibiting the influence of sparse characteristics of other factors on the detection accuracy.
Description
Technical Field
The invention belongs to the field of infrared image processing and target detection, and particularly relates to an infrared weak and small target detection method based on non-convex rank approximation minimization.
Background
The infrared imaging technology has the characteristics of non-contact property, strong capability of capturing details and the like, and realizes the detection of continuous long-distance targets day and night without being influenced by obstacles such as smoke, fog and the like; the infrared search and tracking IRST (infrared search and tracking) system is widely applied to the fields of military, civil use and the like, wherein the infrared small and weak target detection technology is used as a basic function of the IRST system and has important significance in infrared search, infrared early warning and long-distance target detection. However, due to the lack of texture and structural information of the target in the infrared band, the influence of long distance, complex background and various clutter, the infrared target is often spotted or spotted and even submerged in the background, which makes the detection of the infrared weak and small target extremely difficult.
The infrared weak and small target detection technology is divided into two main categories: the technology for detecting the weak and small targets based on a single frame and the technology for detecting the weak and small targets based on multiple frames have the advantages that due to the fact that the technology for detecting the weak and small targets based on the multiple frames needs to be combined with the multiple frames to capture the motion tracks of the targets and eliminate noise interference, extremely large calculation amount and storage amount are needed, requirements on hardware are high, and application in practical engineering is few. Currently, the commonly used detection methods based on a single frame are classified into the following three categories:
(1) background suppression: the background suppression method is based on the assumption of background consistency in the infrared image, a filter is adopted to predict the background of the infrared image, then the background is subtracted from the original image, and finally threshold segmentation is carried out to detect the dim target. Maximum median filtering, maximum mean filtering, top-hat transformation, two-dimensional least mean square filtering, etc. all belong to the category of background suppression. Although this type of method is simple to implement, the background suppression method is very susceptible to noise clutter due to the assumption that noise does not conform to consistency, resulting in poor suppression of most infrared images with low signal-to-noise ratio.
(2) Visual saliency: the human visual system hvs (human visual system) involves three mechanisms, contrast, visual attention and eye movement, the most of which is the contrast mechanism, i.e. in the assumed infrared image, the most prominent object is targeted. For example, a gaussian difference filter calculates a saliency map using two different gaussian filters, and detects and identifies a target; the method based on local contrast utilizes the characteristics that the local contrast of a small neighborhood containing a target is high, but the local contrast of a background area of the target which is not contained is low, and achieves the aim of detection by computing a local contrast map, highlighting the target and restraining the background. When the infrared image conforms to the assumption of visual saliency, the method can obtain excellent effect, but in practical application scenes, the assumption is difficult to meet, for example, when a salient false alarm source exists, the false detection problem is difficult to overcome, and the accuracy is low.
(3) Separating a target background: the method utilizes the non-local autocorrelation of the infrared image background and the sparsity of the target to convert the target detection problem into an optimization problem; the method can be further divided into a method based on an ultra-complete dictionary and low-rank representation and a method based on low-rank background and sparse target restoration. The first method needs to construct an ultra-complete dictionary with different target sizes and shapes in advance by a Gaussian intensity model, the process of constructing the target dictionary is complicated, the detection result is greatly influenced by the dictionary, and the Gaussian intensity model is not applicable any more if the target sizes and shapes are changed greatly; in the second method, a low-rank original block Image can be obtained by means of an IPI (acquired Path-Image) model, and then a background and a target Image are recovered by optimizing a target function by means of the characteristic of target sparsity, so that a detection result is finally obtained; the second method is excellent in effect, but has the following two problems: firstly, because strong edges, partial noise and false alarm sources are also sparse, the detection accuracy can be reduced; secondly, due to the fact that the process of objective function optimization needs iteration, real-time performance is difficult to achieve. Therefore, there is a need for an infrared small target detection method that overcomes the above problems.
Disclosure of Invention
The invention aims to: the invention provides an infrared small and weak target detection method based on non-convex rank approach minimization, and solves the problem that the existing IPI method has the sparse characteristic due to factors such as strong edges, partial noise, false alarm sources and the like, so that the infrared small and weak target detection accuracy is low.
The technical scheme adopted by the invention is as follows:
an infrared weak and small target detection method based on non-convex rank approximation minimization comprises the following steps:
step 1: traversing the original image by adopting a sliding window to construct an infrared block image;
step 2: based on non-convex rank approximation minimization thought, gamma norm combined with weighted l1Norm and l2,1The norm jointly constructs a target function, and after the infrared block images are input into the target function, the target function is solved by utilizing an augmented Lagrange multiplier method and a differential convex programming method to obtain a background block image and a target block image;
and step 3: reconstructing a background image and a target image according to the background block image and the target block image;
and 4, step 4: and performing threshold segmentation on the target image to determine the position of the target, and outputting a target detection result.
Preferably, the step 1 comprises the steps of:
step 1.1: obtaining the infrared image D epsilon R to be processedm×n;
Step 1.2: traversing an original image D according to the step length s by adopting a sliding window W with the size of p multiplied by p, and converting a matrix vector with the size of p multiplied by p in the sliding window W each time into p2A column vector of x 1;
step 1.3: repeating the step 1.2 according to the window sliding times q until the traversal is completed, and forming a new matrix, namely the infrared block image, by all the column vectors
Preferably, the step 2 comprises the steps of:
Step 2.2: weighted l in combination with rank minimization metric1Norm and l2,1Norm, constructing an objective function;
step 2.3: infrared block imageAfter the target function is input, the method of augmented Lagrange multiplier is adopted to solve the target function output background block imageAnd target block image
Preferably, said step 2.2 comprises the steps of:
step 2.2.1: suppose that an image X belongs to Rm×nThe method comprises the following steps of constructing an objective function to separate a low-rank component A and a sparse component E, wherein the objective function comprises the following formula:
min||A||γ+λ||E||w,1+β||N||2,1
s.t.X=A+E+N
wherein λ and β represent balance coefficients, | | · | | non-calculationγNamely, it isRepresenting a pseudo-norm, | · | | non-conducting phosphorw,1Namely, it isRepresenting weighted l1Norm, | · | luminance2,1Namely, it isRepresents l2,1A norm;
step 2.2.2: and optimizing the objective function by adopting an augmented Lagrange equation, wherein the augmented Lagrange equation is as follows:
wherein Y represents a Lagrange multiplier, mu represents a nonnegative penalty factor, w represents a weight coefficient matrix, and w ═ 1 ∈ Rm ×n,<·>Represents inner product operation, | · | non-conducting phosphorFNamely, it isRepresenting the Frobenius norm.
Preferably, said step 2.3 comprises the steps of:
step 2.3.2: iterative solution of target function based on augmented Lagrange equation and difference convex programming method to obtain low-rank matrix, namely background block imageAnd sparse matrices, i.e. target block images
Preferably, said step 2.3.2 comprises the steps of:
step 2.3.2.1: initializing parameters of an augmented Lagrange equation, and enabling the iteration number k to be 0 and the maximum iteration number to be maxk;
step 2.3.2.2: fix A, N, Y, update Ek+1The calculation formula is as follows:
wherein S isτ(. represents a soft threshold shrink operator, Sτ(·)=sgn(x)max(|x|-τ,0);
Step 2.3.2.3: fixed E, N, Y, update A with the differential convex programming methodk+1The calculation formula is as follows:
let M be X-Ek+1-Nk+Yk/μkThe following can be obtained by using the differential convex programming method:
Ak+1=Udiag{σ*}VT,
wherein U and V are left and right singular matrices of M, diag denotes a diagonal matrix, denotes that f (-) is at σkGradient of (a)MRepresenting the singular value of M;
step 2.3.2.4: fix A, E, Y, update Nk+1The following were used:
let Q be X-Ak+1-Ek+1+Yk/μkThen, there are:
wherein [ N ]k+1]:,iRepresents Nk+1The ith column;
step 2.3.2.5: fix A, E, N, update Yk+1The following were used:
Yk+1=Yk+μ(X-Ak+1-Ek+1-Nk+1);
step 2.3.2.6: update the weight wk+1:
Wherein C and εTRepresents the update constant, C ≧ 1, εT>0;
Step 2.3.2.7: updating muk+1=ρμk
Wherein rho represents a growth coefficient, and rho is more than 1;
step 2.3.2.8: the number of iterations k is k + 1;
step 2.3.2.9: judging whether k is larger than maxk, if so, stopping iteration, and turning to step 2.3.2.10; if not, judgingIf yes, stopping iteration, and going to step 2.3.2.10, and if not, going to step 2.3.2.2, wherein epsilon represents a loop termination threshold;
step 2.3.2.10: obtaining an optimal solution A*,E*,N*Outputting the final background block imageAnd target block image
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. the invention utilizes a non-convex rank approximation minimization method, restrains the low-rank component of the infrared block image by introducing a non-convex gamma norm, and introduces weighted l1Norm improvement for sparse component approximation2,1The norm is a structured sparse norm, so that the suppression of sparse high-frequency components (such as strong edges) is enhanced, and the optimal value of an objective function is solved by using an augmented Lagrange multiplier method and a differential convex programming method; solves the problem of the prior IPI method thatFactors such as strong edges, partial noise, false alarm sources and the like have sparse characteristics, so that the problem of low detection accuracy of the infrared weak and small target is solved, and the effect of inhibiting the influence of sparse characteristics of other factors on the detection accuracy is achieved;
2. the invention converts the infrared weak and small target detection problem into the solving problem of the target function, can self-adaptively separate the target and the background without calculating any characteristics, can efficiently and accurately detect the weak and small target, and simultaneously has the gamma norm, l1Norm sum l2,1The combination of the norm and the norm improves the anti-noise capability of the algorithm, and even if certain noise exists, the algorithm can still accurately detect the target, thereby further improving the accuracy of infrared weak and small target detection;
3. the invention reduces the times of singular value decomposition, has faster convergence speed, reduces the operation time of the algorithm and improves the real-time property.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
FIG. 1 is a flow chart of the present invention;
FIG. 2 is an infrared image of the present invention containing a small target;
FIG. 3 is a block image constructed from FIG. 2 according to the present invention;
FIG. 4 is a diagram of the background block image and the target block image separated from FIG. 3 according to the present invention;
FIG. 5 is a diagram of the target image and the background image recovered from FIG. 4 according to the present invention;
FIG. 6 is a gray scale three-dimensional distribution diagram of the target image of FIGS. 2 and 5 according to the present invention;
FIG. 7 is a diagram illustrating the adaptive threshold segmentation of the target image of FIG. 5 to obtain a detection result according to the present invention;
FIG. 8 is a graph of the result of the IPI method versus the FIG. 2 test and a three-dimensional gray scale;
FIG. 9 is a graph of the NIPPS method versus the test results of FIG. 2 and a three-dimensional gray scale;
FIG. 10 is a graph of the detection result of FIG. 2 by the Top-Hat method and a three-dimensional gray scale;
FIG. 11 is a diagram of the result of the detection of FIG. 2 by the MPCM method and a three-dimensional gray scale map;
FIG. 12 is a graph comparing the effects of the present invention;
fig. 13 is a sequence diagram of a comparative use of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
It is noted that relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The technical problem is as follows: the problem that the existing IPI method has low accuracy of infrared weak and small target detection due to the fact that factors such as strong edges, partial noise, false alarm sources and the like have sparse characteristics is solved;
the technical means is as follows:
an infrared weak and small target detection method based on non-convex rank approximation minimization comprises the following steps:
step 1: traversing the original image by adopting a sliding window to construct an infrared block image;
step 2: based on non-convex rank approximation minimization thought, gamma norm combined with weighted l1Norm and l2,1The norm jointly constructs a target function, and after the infrared block images are input into the target function, the target function is solved by utilizing an augmented Lagrange multiplier method and a differential convex programming method to obtain a background block image and a target block image;
and step 3: reconstructing a background image and a target image according to the background block image and the target block image;
and 4, step 4: and performing threshold segmentation on the target image to determine the position of the target, and outputting a target detection result.
The step 1 comprises the following steps:
step 1.1: obtaining the infrared image D epsilon R to be processedm×n;
Step 1.2: traversing an original image D according to the step length s by adopting a sliding window W with the size of p multiplied by p, and converting a matrix vector with the size of p multiplied by p in the sliding window W each time into p2A column vector of x 1;
step 1.3: repeating the step 1.2 according to the window sliding times q until the traversal is completed, and forming a new matrix, namely the infrared block image, by all the column vectors
The step 2 comprises the following steps:
Step 2.2: weighted l in combination with rank minimization metric1Norm and l2,1Norm, constructing an objective function;
step 2.3: infrared block imageAfter the target function is input, the method of augmented Lagrange multiplier is adopted to solve the target function output background block imageAnd target block image
Step 2.2 comprises the following steps:
step 2.2.1: suppose that an image X belongs to Rm×nThe method comprises the following steps of constructing an objective function to separate a low-rank component A and a sparse component E, wherein the objective function comprises the following formula:
min||A||γ+λ||E||w,1+β||N||2,1
s.t.X=A+E+N
wherein λ and β represent balance coefficients, | | · | | non-calculationγNamely, it isRepresenting a pseudo-norm, | · | | non-conducting phosphorw,1Namely, it isRepresenting weighted l1Norm, | · | luminance2,1Namely, it isRepresents l2,1A norm;
step 2.2.2: and optimizing the objective function by adopting an augmented Lagrange equation, wherein the augmented Lagrange equation is as follows:
wherein Y represents a Lagrange multiplier, mu represents a nonnegative penalty factor, w represents a weight coefficient matrix, and w ═ 1 ∈ Rm ×n,<·>Represents inner product operation, | · | non-conducting phosphorFNamely, it isRepresenting the Frobenius norm.
Step 2.3 comprises the following steps:
step 2.3.2: iterative solution of target function based on augmented Lagrange equation and difference convex programming method to obtain low-rank matrix, namely background block imageAnd sparse matrices, i.e. target block images
Step 2.3.2 comprises the following steps:
step 2.3.2.1: initializing parameters of an augmented Lagrange equation, and enabling the iteration number k to be 0 and the maximum iteration number to be maxk;
step 2.3.2.2: fix A, N, Y, update Ek+1The calculation formula is as follows:
wherein S isτ(. represents a soft threshold shrink operator, Sτ(·)=sgn(x)max(|x|-τ,0);
Step 2.3.2.3: fixed E, N, Y, update A with the differential convex programming methodk+1The calculation formula is as follows:
let M be X-Ek+1-Nk+Yk/μkThe following can be obtained by using the differential convex programming method:
Ak+1=Udiag{σ*}VT,
wherein U and V are left and right singular matrices of M, diag denotes a diagonal matrix, denotes that f (-) is at σkGradient of (a)MRepresenting the singular value of M;
step 2.3.2.4: fix A, E, Y, update Nk+1The following were used:
let Q be X-Ak+1-Ek+1+Yk/μkThen, there are:
wherein [ N ]k+1]:,iRepresents Nk+1The ith column;
step 2.3.2.5: fix A, E, N, update Yk+1The following were used:
Yk+1=Yk+μ(X-Ak+1-Ek+1-Nk+1);
step 2.3.2.6: update the weight wk+1:
Wherein C and εTRepresents the update constant, C ≧ 1, εT>0;
Step 2.3.2.7: updating muk+1=ρμk
Wherein rho represents a growth coefficient, and rho is more than 1;
step 2.3.2.8: the number of iterations k is k + 1;
step 2.3.2.9: judging whether k is larger than maxk, if so, stopping iteration, and turning to step 2.3.2.10; if not, judging | D0-Ak+1-Ek+1-Nk+1||2/||D0||2If yes, stopping iteration, and going to step 2.3.2.10, if not, going to step 2.3.2.2, wherein epsilon represents a loop termination threshold value;
step 2.3.2.10: obtaining an optimal solution A*,E*,N*Outputting the final background block imageAnd target block image
The technical effects are as follows: the invention utilizes a non-convex rank approximation minimization method, restrains the low-rank component of the infrared block image by introducing a non-convex gamma norm, and introduces weighted l1Norm improvement for sparse component approximation2,1The norm is a structured sparse norm, so that the suppression of sparse high-frequency components (such as strong edges) is enhanced, and the optimal value of an objective function is solved by using an augmented Lagrange multiplier method and a differential convex programming method; the problem that the detection accuracy of the infrared weak and small target is low due to the fact that factors such as strong edges, partial noise and a virtual alarm source are sparse in the existing IPI method is solved, and the effect of inhibiting the influence of the sparse characteristics of other factors on the detection accuracy is achieved.
The features and properties of the present invention are described in further detail below with reference to examples.
Example 1
As shown in fig. 1 to 13, a method for detecting an infrared weak and small target based on non-convex rank approach minimization includes the following steps:
step 1: traversing the original image by adopting a sliding window to construct an infrared block image;
step 2: based on non-convex rank approximation minimization thought, gamma norm combined with weighted l1Norm and l2,1The norm jointly constructs a target function, and after the infrared block images are input into the target function, the target function is solved by utilizing an augmented Lagrange multiplier method and a differential convex programming method to obtain a background block image and a target block image;
and step 3: reconstructing a background image and a target image according to the background block image and the target block image;
and 4, step 4: and performing self-adaptive threshold segmentation on the target image to determine the position of the target and outputting a target detection result.
The step 1 comprises the following steps:
step 1.1: obtaining the infrared image D epsilon R to be processedm×nSize 240 × 320;
step 1.2: traversing an original image D according to the step length of 10 by adopting a sliding window W with the size of 50 multiplied by 50, and converting a matrix vector with the size of 50 multiplied by 50 in the sliding window W each time into a column vector with the size of 2500 multiplied by 1;
step 1.3: repeating the step 1.2 according to the window sliding times 560 until the traversal is completed, and forming a new matrix, namely 2500 x 560 infrared block images by all the column vectors
The step 2 comprises the following steps:
Step (ii) of2.2: weighted l in combination with rank minimization metric1Norm and l2,1Norm, constructing an objective function;
step 2.3: infrared block imageAfter the target function is input, the method of augmented Lagrange multiplier is adopted to solve the target function output background block imageAnd target block image
Step 2.2 comprises the following steps:
step 2.2.1: suppose that an image X belongs to Rm×nThe method comprises the following steps of constructing an objective function to separate a low-rank component A and a sparse component E, wherein the objective function comprises the following formula:
min||A||γ+λ||E||w,1+β||N||2,1
s.t.X=A+E+N
wherein, λ and β represent equilibrium coefficients,β=1.8,γ=0.002,||·||γnamely, it isRepresenting a pseudo-norm, | · | | non-conducting phosphorw,1Namely, it isRepresenting weighted l1Norm, | · | luminance2,1Namely, it isRepresents l2,1A norm;
step 2.2.2: and optimizing the objective function by adopting an augmented Lagrange equation, wherein the augmented Lagrange equation is as follows:
wherein Y represents a lagrange multiplier, μ represents a non-negative penalty factor, μ is 90, w represents a weight coefficient matrix, w is 1 ∈ Rm×n,<·>Represents inner product operation, | · | non-conducting phosphorFNamely, it isRepresenting the Frobenius norm.
Step 2.3 comprises the following steps:
step 2.3.2: iterative solution of target function based on augmented Lagrange equation and difference convex programming method to obtain low-rank matrix, namely background block imageAnd sparse matrices, i.e. target block images
Step 2.3.2 comprises the following steps:
step 2.3.2.1: initializing parameters of the augmented Lagrange equation, setting the iteration number k to be 0, setting the maximum iteration number to be maxk, setting A, E, N and Y to be 0,w=1,μ=90;
step 2.3.2.2: fix A, N, Y, update Ek+1The calculation formula is as follows:
wherein S isτ(. represents a soft threshold shrink operator, Sτ(·)=sgn(x)max(|x|-τ,0);
Step 2.3.2.3: fixed E, N, Y, update A with the differential convex programming methodk+1The calculation formula is as follows:
let M be X-Ek+1-Nk+Yk/μkThe following can be obtained by using the differential convex programming method:
Ak+1=Udiag{σ*}VT,
wherein U and V are left and right singular matrices of M, diag denotes a diagonal matrix, denotes that f (-) is at σkGradient of (a)MRepresenting the singular value of M;
step 2.3.2.4: fix A, E, Y, update Nk+1The following were used:
let Q be X-Ak+1-Ek+1+Yk/μkThen, there are:
wherein [ N ]k+1]:,iRepresents Nk+1The ith column;
step 2.3.2.5: fix A, E, N, update Yk+1The following were used:
Yk+1=Yk+μ(X-Ak+1-Ek+1-Nk+1);
step 2.3.2.6: update the weight wk+1:
Wherein C and εTDenotes the update constant, C ≧ 1, C ═ 1.2, εT>0,εT=0.4;
Step 2.3.2.7: updating muk+1=ρμk,
Wherein ρ represents a growth coefficient, ρ > 1, and ρ is 1.1;
step 2.3.2.8: the number of iterations k is k + 1;
step 2.3.2.9: judging whether k is larger than maxk, if so, stopping iteration, and turning to step 2.3.2.10; if not, judging | D0-Ak+1-Ek+1-Nk+1||2/||D0||2If ≦ ε is true, stop iteration, go to step 2.3.2.10, if false, go to step 2.3.2.2, where ε represents the threshold for loop termination, ε is 10-7;
Step 2.3.2.10: obtaining an optimal solution A*,E*,N*Outputting the final background block imageAnd target block image
The specific steps of the step 3 are as follows: for input background block imageTake out B0Each column in the image is reconstructed into a small matrix with the size of 50 multiplied by 50, and then a background image B epsilon R of 240 multiplied by 320 is sequentially formed according to the sequencem×nFor the position contained by a plurality of small blocks, the gray value of the position is determined by adopting a median filtering mode, and the target image T is formed by adopting the same modeT0Reconstructing;
the specific steps of the step 4 are as follows: and performing adaptive threshold segmentation on the target image T, wherein a threshold Th is m + c sigma, m represents the mean value of all gray scales in the target image T, sigma represents the standard deviation of all gray scales in the target image T, c represents a constant between 1 and 10, and the segmentation is completed to obtain a target detection result.
Effect analysis was performed according to the attached figures: FIG. 2 shows an infrared image with a complex background, with a very bright white false alarm source in addition to a small target; FIG. 3 is a block image D constructed from an original image through step 10(ii) a FIG. 4 shows the result of step 2 from step D0Recovered B0And T0(ii) a Fig. 5 is the background B and the target map T reconstructed through step 3; fig. 6 is a three-dimensional gray scale diagram of the original image D and the target image T, and it can be seen that the separated target image suppresses the background well, and the gray scales of the background at the positions other than the small target are 0; FIG. 7 is the final test result; 8-11 are several other methods (IPI, NIPPS, Top-Hat, MPCM in sequence) for detecting the small target in FIG. 2 (for ease of illustration, the result has been binarized), and corresponding three-dimensional distribution of gray scale, it can be seen that the rest four methods do not completely suppress the background, and have different degrees of noise, which will affect the subsequent positioning detection, FIG. 13 is an original diagram of one of the sequences selected for comparison, which has been gray-scale processed, and thus is gray-green; fig. 12 is a comparison of the performance of the present invention with that of the other four methods, and it is obvious that the signal-to-noise ratio gain SCRG and the background suppression factor BSF are the largest, and the target detection accuracy of the present invention is far better than that of the other methods.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (6)
1. An infrared weak and small target detection method based on non-convex rank approximation minimization is characterized in that: the method comprises the following steps:
step 1: traversing the original image by adopting a sliding window to construct an infrared block image;
step 2: based on non-convex rank approximation minimization thought, gamma norm combined with weighted l1Norm and l2,1The norm jointly constructs a target function, and after the infrared block images are input into the target function, the target function is solved by utilizing an augmented Lagrange multiplier method and a differential convex programming method to obtain a background block image and a target block image;
and step 3: reconstructing a background image and a target image according to the background block image and the target block image;
and 4, step 4: and performing threshold segmentation on the target image to determine the position of the target, and outputting a target detection result.
2. The infrared weak and small target detection method based on non-convex rank approach minimization as claimed in claim 1, characterized in that: the step 1 comprises the following steps:
step 1.1: obtaining an original image to be processed, namely an infrared image D epsilon Rm×n;
Step 1.2: traversing the infrared image D according to the step length s by adopting a sliding window W with the size of p multiplied by p, and converting a matrix vector with the size of p multiplied by p in the sliding window W each time into p2A column vector of x 1;
3. The method for detecting the infrared weak and small target based on the non-convex rank approximation minimization according to claim 1 or 2, characterized in that: the step 2 comprises the following steps:
Step 2.2: weighted l in combination with rank minimization metric1Norm and l2,1Norm, constructing an objective function;
4. The infrared weak and small target detection method based on non-convex rank approximation minimization according to claim 3, characterized in that: the step 2.2 comprises the following steps:
step 2.2.1: suppose that an image X belongs to Rm×nThe method comprises the following steps of constructing an objective function to separate a low-rank component A and a sparse component E, wherein the objective function comprises the following formula:
min||A||γ+λ||E||w,1+β||N||2,1
s.t.X=A+E+N
wherein λ and β represent balance coefficients, | | · | | non-calculationγNamely, it isRepresenting a pseudo-norm, | · | | non-conducting phosphorw,1Namely, it isRepresenting weighted l1Norm, | · | luminance2,1Namely, it isRepresents l2,1A norm;
step 2.2.2: and optimizing the objective function by adopting an augmented Lagrange equation, wherein the augmented Lagrange equation is as follows:
5. The infrared weak and small target detection method based on non-convex rank approximation minimization according to claim 4, characterized in that: the step 2.3 comprises the following steps:
6. The infrared weak and small target detection method based on non-convex rank approximation minimization according to claim 5, characterized in that: the step 2.3.2 comprises the following steps:
step 2.3.2.1: initializing parameters of an augmented Lagrange equation, and enabling the iteration number k to be 0 and the maximum iteration number to be maxk;
step 2.3.2.2: fix A, N, Y, update Ek+1The calculation formula is as follows:
wherein S isτ(. represents a soft threshold shrink operator, Sτ(·)=sgn(x)max(|x|-τ,0);
Step 2.3.2.3: fixed E, N, Y, update A with the differential convex programming methodk+1The calculation formula is as follows:
let M be X-Ek+1-Nk+Yk/μkThe following can be obtained by using the differential convex programming method:
Ak+1=Udiag{σ*}VT,
wherein U and V are left and right singular matrices of M, diag denotes a diagonal matrix, denotes that f (-) is at σkGradient of (a)MRepresenting the singular value of M;
step 2.3.2.4: fix A, E, Y, update Nk+1The following were used:
let Q be X-Ak+1-Ek+1+Yk/μkThen, there are:
wherein [ N ]k+1]:,iRepresents Nk+1The ith column;
step 2.3.2.5: fix A, E, N, update Yk+1The following were used:
Yk+1=Yk+μ(X-Ak+1-Ek+1-Nk+1);
step 2.3.2.6: update the weight wk+1:
Wherein C and εTRepresents the update constant, C ≧ 1, εT>0;
Step 2.3.2.7: updating muk+1=ρμk
Wherein rho represents a growth coefficient, and rho is more than 1;
step 2.3.2.8: the number of iterations k is k + 1;
step 2.3.2.9: judging whether k is larger than maxk, if so, stopping iteration, and turning to step 2.3.2.10; if not, judging | D0-Ak+1-Ek+1-Nk+1||2/||D0||2If yes, stopping iteration, and going to step 2.3.2.10, if not, going to step 2.3.2.2, wherein epsilon represents a loop termination threshold value;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811116093.9A CN109215025B (en) | 2018-09-25 | 2018-09-25 | Infrared weak and small target detection method based on non-convex rank approach minimization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811116093.9A CN109215025B (en) | 2018-09-25 | 2018-09-25 | Infrared weak and small target detection method based on non-convex rank approach minimization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109215025A CN109215025A (en) | 2019-01-15 |
CN109215025B true CN109215025B (en) | 2021-08-10 |
Family
ID=64984888
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811116093.9A Active CN109215025B (en) | 2018-09-25 | 2018-09-25 | Infrared weak and small target detection method based on non-convex rank approach minimization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109215025B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109934815B (en) * | 2019-03-18 | 2023-04-14 | 电子科技大学 | Tensor recovery infrared small target detection method combined with ATV constraint |
CN109839362B (en) * | 2019-03-20 | 2021-08-17 | 北京中科知草科技有限公司 | Infrared spectrum quantitative analysis method based on progressive denoising technology |
CN110443209B (en) * | 2019-08-08 | 2022-07-29 | 南京邮电大学 | Infrared small target detection method based on matrix recovery |
CN111160181B (en) * | 2019-12-20 | 2022-07-05 | 西北工业大学 | Small target detection method based on infrared video image |
CN111967466B (en) * | 2020-07-08 | 2022-08-16 | 北京航空航天大学 | Infrared weak and small target detection method utilizing multidirectional sparse low-rank decomposition |
CN113421279B (en) * | 2021-07-07 | 2023-03-31 | 中国科学院光电技术研究所 | Infrared weak and small target detection method based on weighted nuclear norm minimization |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103268607A (en) * | 2013-05-15 | 2013-08-28 | 电子科技大学 | Common object detection method on weak supervision condition |
CN104933683A (en) * | 2015-06-09 | 2015-09-23 | 南昌大学 | Non-convex low-rank reconstruction method for rapid magnetic resonance (MR) imaging |
CN107133930A (en) * | 2017-04-30 | 2017-09-05 | 天津大学 | Ranks missing image fill method with rarefaction representation is rebuild based on low-rank matrix |
CN107680120A (en) * | 2017-09-05 | 2018-02-09 | 南京理工大学 | Tracking Method of IR Small Target based on rarefaction representation and transfer confined-particle filtering |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100565244C (en) * | 2007-10-31 | 2009-12-02 | 北京航空航天大学 | Multimode multi-target accurate tracking apparatus and method |
US8311720B2 (en) * | 2009-01-09 | 2012-11-13 | Robert Bosch Gmbh | Lost target function for adaptive cruise control |
CN102789642B (en) * | 2011-05-16 | 2017-08-25 | 索尼公司 | Direction of extinction determines method and apparatus, camera self-calibration method and device |
CN103456015B (en) * | 2013-09-06 | 2016-10-05 | 电子科技大学 | A kind of SAR target detection method based on optimum score field Gabor spectrum signature |
-
2018
- 2018-09-25 CN CN201811116093.9A patent/CN109215025B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103268607A (en) * | 2013-05-15 | 2013-08-28 | 电子科技大学 | Common object detection method on weak supervision condition |
CN104933683A (en) * | 2015-06-09 | 2015-09-23 | 南昌大学 | Non-convex low-rank reconstruction method for rapid magnetic resonance (MR) imaging |
CN107133930A (en) * | 2017-04-30 | 2017-09-05 | 天津大学 | Ranks missing image fill method with rarefaction representation is rebuild based on low-rank matrix |
CN107680120A (en) * | 2017-09-05 | 2018-02-09 | 南京理工大学 | Tracking Method of IR Small Target based on rarefaction representation and transfer confined-particle filtering |
Non-Patent Citations (2)
Title |
---|
基于通用对象估计的目标检测与模糊车牌识别算法研究;赵丹凤;《中国优秀硕士学位论文全文数据库 信息科技辑》;20170215(第02期);第I138-3702页 * |
基于非凸近似的矩阵秩极小化算法研究及应用;陈勇勇;《中国优秀硕士学位论文全文数据库 信息科技辑》;20180315(第03期);第I138-1897页 * |
Also Published As
Publication number | Publication date |
---|---|
CN109215025A (en) | 2019-01-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109215025B (en) | Infrared weak and small target detection method based on non-convex rank approach minimization | |
CN109934815B (en) | Tensor recovery infrared small target detection method combined with ATV constraint | |
CN109325446B (en) | Infrared weak and small target detection method based on weighted truncation nuclear norm | |
CN109447073B (en) | Infrared weak and small target detection method based on tensor robust principal component analysis | |
Wang et al. | Infrared dim and small target detection based on stable multisubspace learning in heterogeneous scene | |
CN107563433B (en) | Infrared small target detection method based on convolutional neural network | |
CN111080675B (en) | Target tracking method based on space-time constraint correlation filtering | |
CN110135344B (en) | Infrared dim target detection method based on weighted fixed rank representation | |
CN110163818A (en) | A kind of low illumination level video image enhancement for maritime affairs unmanned plane | |
Qian et al. | Infrared dim-small target tracking via singular value decomposition and improved Kernelized correlation filter | |
Gao et al. | Improving the performance of infrared and visible image fusion based on latent low-rank representation nested with rolling guided image filtering | |
CN110706208A (en) | Infrared dim target detection method based on tensor mean square minimum error | |
CN109345563A (en) | The moving target detecting method decomposed based on low-rank sparse | |
CN113822352A (en) | Infrared dim target detection method based on multi-feature fusion | |
CN111967466A (en) | Infrared weak and small target detection method utilizing multidirectional sparse low-rank decomposition | |
Li et al. | Image enhancement algorithm based on depth difference and illumination adjustment | |
CN112766223A (en) | Hyperspectral image target detection method based on sample mining and background reconstruction | |
CN107301631B (en) | SAR image speckle reduction method based on non-convex weighted sparse constraint | |
CN110992342B (en) | SPCP infrared small target detection method based on 3DATV constraint | |
CN109285148B (en) | Infrared weak and small target detection method based on heavily weighted low rank and enhanced sparsity | |
Yufeng et al. | Research on SAR image change detection algorithm based on hybrid genetic FCM and image registration | |
CN116051444A (en) | Effective infrared and visible light image self-adaptive fusion method | |
Wang et al. | Infrared small target detection method combined with bilateral filter and local entropy | |
Liao | Optimization and Application of Image Defogging Algorithm Based on Deep Learning Network | |
Liu et al. | Research on lane line segmentation algorithm based on deeplabv3 |
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 |