CN105184740A - Non-uniform stripe correction method of infrared focal plane image - Google Patents
Non-uniform stripe correction method of infrared focal plane image Download PDFInfo
- Publication number
- CN105184740A CN105184740A CN201510261452.XA CN201510261452A CN105184740A CN 105184740 A CN105184740 A CN 105184740A CN 201510261452 A CN201510261452 A CN 201510261452A CN 105184740 A CN105184740 A CN 105184740A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- msubsup
- mfrac
- msup
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000008569 process Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 2
- JXASPPWQHFOWPL-UHFFFAOYSA-N Tamarixin Natural products C1=C(O)C(OC)=CC=C1C1=C(OC2C(C(O)C(O)C(CO)O2)O)C(=O)C2=C(O)C=C(O)C=C2O1 JXASPPWQHFOWPL-UHFFFAOYSA-N 0.000 claims 1
- 238000009795 derivation Methods 0.000 claims 1
- 238000001914 filtration Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000004800 variational method Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Landscapes
- Transforming Light Signals Into Electric Signals (AREA)
- Image Processing (AREA)
- Studio Devices (AREA)
Abstract
The invention provides a non-uniform stripe correction method of an infrared focal plane image. The method comprises that 1) positions of pixel points of the infrared focal plane image with stripe noise are collected; 2) unidirectional total-variation de-striping energy functional model is established for the infrared focal plane image with stripe noise; 3) an auxiliary variable is introduced to replace a non-differentiable item in the energy functional model to obtain a new energy function; and 4) a problem to be solved is divided into three sub problems according to the new energy function, the three sub problems are solved in an alternative iteration manner, and a corrected image f is output. The method utilizes the characteristic that the stripe influences the gradient vertical to the stripe but not other gradients, the unidirectional stripe noise can be rapidly and effectively removed, advantages of stripe removing and image detail storing are combined, the adaptability is high, and the computation complexity is low.
Description
Technical Field
The invention belongs to the field of infrared image processing, and particularly relates to an infrared focal plane image non-uniformity strip correction method which is suitable for quickly removing unidirectional strip noise in multi-sensor imaging data.
Background
Due to the influence of infrared sensor materials, process level limitation and external environment, the response of each detecting element in the infrared focal plane array is inconsistent, so that non-uniform stripe noise appears in image data, and the subsequent data processing is seriously influenced. The difficulty of the stripe noise correction is to effectively remove various different types of stripe noise and simultaneously more completely preserve the original structural information of the image. Banding correction algorithms can be broadly divided into three broad categories:
the first major category of statistical matching-based methods is represented by histogram matching and moment matching methods. The statistical matching method is simple and quick and is easy to realize. The defects are that the statistical characteristics of the rows or the columns of the image are assumed to be consistent, the positions of the strips are required to be known in advance, the situation is not practical, and the strip removing effect is poor.
The second category is based on image filtering technology, represented by low-pass filtering and power filters, and has the advantages of performing filtering processing only on the stripes, not processing non-stripe contents, and simplicity and easiness in implementation. However, depending on the accuracy of the detection of the bands, the bands that are missed are ignored and not processed, and the wrong detection will filter out the image content, resulting in information loss. It is difficult to accurately detect the bands in practical situations.
The third category is based on a variation regularization method, a stripe correction problem is regarded as an inverse problem of estimating a non-stripe-contained image, an energy function containing a data item and a regularization item is established, and the non-stripe-contained image is obtained through iterative solution of a minimized energy function. An article "Towardoptallustrating of MODISdatausing angular imaging model" by Bouali et al proposes a variational method for correcting remote sensing satellite image strips, and an energy functional of the variational method only comprises a gradient domain data fidelity item TV in an x directionx(f) And data constraint term α TV in y-directionyTotal variation energy functional of (f-g):
wherein TVxAnd TVyThe total variation is shown for the x and y directions of the image, respectively. The second term represents that we look for an image which changes almost as much as the stripe image in the vertical direction because the stripe noise almost only affects the image gradient in the horizontal direction, while the first term represents that only the stripe image is smoothed in the horizontal direction, the model effectively reflects the unidirectional property of the stripe, but the model result image has a large deviation in gray value with the original image as a whole.
Disclosure of Invention
The invention aims to provide a method for correcting non-uniform strips of an infrared focal plane image, which is used for expressing a strip correction problem as an inverse problem of estimating a real image from a strip image, utilizing the characteristic knowledge that a strip only has influence on a gradient perpendicular to the strip without influencing the gradient along the direction of the strip, quickly and effectively removing unidirectional strip noise, and has the advantages of strip removal and image detail storage, strong adaptability and low calculation complexity.
Compared with the prior art, the invention has the following advantages:
first, we consider both stripe removal and image detail preservation. The invention represents the banding correction problem as the inverse problem of estimating the real image from the banding image, and utilizes the characteristic knowledge that the banding only has influence on the gradient perpendicular to the banding image and does not influence the gradient along the direction of the banding image. The method can effectively remove the stripe noise and better save the image details.
Second, the computational complexity is low. The method utilizes the splitting Bregman method to carry out numerical optimization solution, and effectively avoids the infinitesimal terms. Meanwhile, Guass _ Sidel iteration is adopted for solving, and the convergence rate is higher.
Thirdly, the adaptability is strong. The invention provides an algorithm parameter adjusting interface, which is suitable for various strip noises.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a real infrared image with a slight banding removal effect, and FIG. 2a is an original noise image; FIG. 2b is an image after non-uniformity correction; FIG. 2c is the difference between the original noise image and the non-uniformity corrected image;
FIG. 3 is the result of removing the median band in the real infrared image, and FIG. 3a is the original noise image; FIG. 3b is an image after non-uniformity correction; FIG. 3c is the difference between the original noise image and the non-uniformity corrected image;
FIG. 4 is a diagram of a true infrared image heavy banding removal effect, and FIG. 4a is an original noise image; FIG. 4b is an image after non-uniformity correction; fig. 4c shows the original noise image and the image difference after non-uniformity correction.
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.
The invention proposes the data fidelity item including the gray scale domain according to the characteristic of the unidirectional stripe noisey-direction gradient domain data fidelity term beta TVy(f-g) and data constraint term α TV in the x-directionx(f) The energy functional of (2) is realized by solving and discretizing by a splitting Bregman method.
The method for realizing the invention is as follows: firstly, a reasonable energy functional model is provided by utilizing the property of unidirectional stripe noise
g (x, y) is a strip image output by the detector, f (x, y) is an original non-strip image, and E (f) represents an energy functional about an image f to be estimated; the energy functional includes a grayscale domain data fidelity termy-direction gradient data fidelity term beta TVy(f-g) and x-direction gradient penalty term α TVx(f) (ii) a Wherein TVxAnd TVyRespectively representing the total variation of the x and y directions of the image; α, β are regularization parameters used to adjust the regularization strength. (x, y) represents the position of a pixel point in the image, and Ω represents the set of image pixel coordinates. The second action specifically expands the expression.
Method for solving introduced auxiliary variable d by splitting Bregman methodx,dy,bx,byEstablishing a new energy functional:
dx,dyis an alternative term, dxAlternative to alpha TVx(f),dySurrogate beta TVy(f-g),bx,byThen it is a data retrieval item, such that <math><mrow>
<mfrac>
<msub>
<mi>λ</mi>
<mn>1</mn>
</msub>
<mn>2</mn>
</mfrac>
<msubsup>
<mrow>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>d</mi>
<mi>x</mi>
</msub>
<mo>-</mo>
<msub>
<mo>▿</mo>
<mi>x</mi>
</msub>
<mi>f</mi>
<mo>-</mo>
<msub>
<mi>b</mi>
<mi>x</mi>
</msub>
<mo>|</mo>
<mo>|</mo>
</mrow>
<mn>2</mn>
<mn>2</mn>
</msubsup>
</mrow>
<mrow>
<mfrac>
<msub>
<mi>λ</mi>
<mn>2</mn>
</msub>
<mn>2</mn>
</mfrac>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>d</mi>
<mi>y</mi>
</msub>
<mo>-</mo>
<msub>
<mo>▿</mo>
<mi>y</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>f</mi>
<mo>-</mo>
<mi>g</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>b</mi>
<mi>y</mi>
</msub>
<mo>|</mo>
<mo>|</mo>
</mrow></math> And the speed is faster and close to 0, and the iterative convergence speed is accelerated. And alpha and beta are regularization parameters, and the regularization constraint strength is adjusted. Lambda [ alpha ]1,λ2Is a penalty parameter for constraining the introduced auxiliary variables.
And (4) carrying out iterative solution on the unidirectional total variation strip energy-removed functional model, and outputting to obtain a corrected image f. Decomposing the solved problem into three sub-problems according to the new energy flood, carrying out alternate iterative solution on the three sub-problems, and outputting to obtain a corrected image f; wherein,
the first sub-problem is: d, b is fixed, f is solved to obtain
The second sub-problem is: f and b are fixed, d is solved to obtain
The third sub-problem is: f and d are fixed, and b is solved to obtain
In order to facilitate hardware implementation, discretization implementation is adopted. As shown in fig. 1, the specific flow of the iterative solution of the present invention is as follows:
1) inputting an infrared focal plane image g containing stripe noise, wherein the size of the image is MxN; number of initialization iterations k equals 1, image fk-1G, auxiliary variableIs a zero matrix of size M × N.
2) Will assist variableSolving by substituting the energy functional iterative objective function to obtain a corrected image fk(ii) a The specific solving process is as follows:
fixing the value realization of b, d solving f
3) The corrected image fkSubstituting the iteration objective functions corresponding to the second sub-problem solution d and the third sub-problem solution b to obtain auxiliary variables
Fixing the value realization of b, f solving d
Numerical implementation of fixed d, f solving b
4) If | fk+1-fk‖2/‖fk+1‖2If the iteration number is less than or greater than the required iteration number and is a preset threshold value, f is outputk+1As a final processed image, end; otherwise, go to step 5).
5) Update k to k +1, return to step 3).
Regularization parameters alpha and beta are introduced into the method, and penalty is introduced into the splitting Bregman optimization solving processParameter lambda1And λ2. The present invention lists sets of parameters corresponding to different intensity stripe examples, which are merely exemplary and not intended to limit the present invention. The number of iterations is fixed at 40, and α is 20, β is 350, λ under the light band1=20,λ290; in the medium band, α is 80, β is 800, λ1=30,λ2110; α -200, β -800, λ under severe banding1=15,λ2=130。
Fig. 2 is a band removal result of a real infrared noise image (light band noise). FIG. 2a is an original noisy image, with slight banding noise; FIG. 2b is an image after non-uniformity correction; fig. 2c shows the difference between the original noise image and the non-uniformity corrected image (for describing the case of removing the stripe noise), where α is 20, β is 350, and λ is1=20,λ290. The group of images clearly shows that the method has good removal effect on the stripes of the light noise image, namely the details of the image are kept, and the noise is removed to the maximum extent.
Fig. 3 shows the band elimination result of a real infrared noise image (moderate band noise). FIG. 3a is an original noisy image with moderate banding noise; FIG. 3b is an image after non-uniformity correction; fig. 3c shows the original noise image and the non-uniformity corrected image difference (for describing the band noise removal), α is 80, β is 800, lambda1 is 30, and lambda2 is 110. The group of images clearly shows that the method has good removal effect on the strips of the moderate noise image, namely the details of the image are kept (the Chinese characters in the image are still clear), and the noise is removed to the maximum extent.
Fig. 4 is a band removal result of a real infrared noise image (heavy band noise). FIG. 4a is an original noisy image, containing severe banding noise; FIG. 4b is an image after non-uniformity correction; fig. 4c shows the original noise image and the non-uniformity corrected image difference (for describing the band noise removal), where α is 200, β is 800, lambda1 is 15, and lambda2 is 130. The group of images clearly shows that the invention has good removal effect on the strips of the heavy noise image, and simultaneously retains the details of the image (the edges of leaves and buildings are still clear), and the strip noise is well suppressed.
Compared with the images in fig. 2, 3 and 4, the method disclosed by the invention has a good strip removing effect on the strip noise infrared focal plane images with different intensities, and is also good for saving details.
Claims (5)
1. A method for correcting non-uniformity strips of an infrared focal plane image is characterized by comprising the following steps:
1) acquiring an infrared focal plane image containing stripe noise;
2) establishing a one-way total variation de-banding energy functional model for the infrared focal plane image containing the banding noise:
wherein E (f) represents an energy functional about the corrected image f, and g is an infrared focal plane image containing stripe noise; the energy functional includes a grayscale domain data fidelity termTVx(f) And TVy(f) Respectively representing the total variation of the corrected image f in the x direction and the y direction; alpha, beta are the regularization parameters,representing the derivation, (x, y) representing the position of a pixel point in the image, and omega representing the coordinate set of the image pixel;
3) introducing an auxiliary variable dx,dy,bx,byReplacing non-differentiable items in the energy functional model to further obtain a new energy functional:
penalty parameter lambda1、λ2Auxiliary variables introduced for constraints;
4) decomposing the solved problem into three sub-problems according to the new energy flood, carrying out alternate iterative solution on the three sub-problems, and outputting to obtain a corrected image f; wherein,
the first sub-problem is: d and b are fixed, and f is solved to obtain:
the second sub-problem is: fixing f and b, solving d to obtain:
the third sub-problem is: fixing f and d, and solving b to obtain:
2. the method for correcting the non-uniformity strip of the infrared focal plane image according to claim 1, wherein the step 3) is implemented by:
1) initializing auxiliary variablesA zero matrix and the iteration number k is 1;
2) will assist variableSubstituting the first subproblem to solve the iterative objective function corresponding to the f, and solving to obtain a corrected image fk;
3) The corrected image fkSubstituting the iteration objective functions corresponding to the second sub-problem solution d and the third sub-problem solution b to obtain auxiliary variables
4) If it isOr k reaches the maximum value of the preset iteration times and is a preset threshold value, and the corrected image fkThe final corrected image f is obtained, and the process is finished; otherwise, entering step 4);
5) update k to k +1, return to step 3).
3. The method of claim 2, wherein the iterative objective function to which the first sub-problem corresponds is
fk(i, j) is the pixel value at the (i, j) position of the kth iteration;represents the (k-1) th iterationAn auxiliary variable for the epoch;respectively represent Bregman variables at the k-1 iteration; g represents the original image.
4. The method of correcting non-uniformity strips in an infrared focal plane image of claim 3, wherein the iterative objective function to which the second sub-problem corresponds is:
wherein,is fromToOf linear operator, whereinRepresents N2The real number of dimensions, the dimensional space, similarly,is also fromToThe linear operator of (2). shrink is a soft threshold function.
5. The method of correcting non-uniformity strips in an infrared focal plane image of claim 4, wherein said third sub-problem corresponds to an iterative objective function of:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510261452.XA CN105184740A (en) | 2015-05-20 | 2015-05-20 | Non-uniform stripe correction method of infrared focal plane image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510261452.XA CN105184740A (en) | 2015-05-20 | 2015-05-20 | Non-uniform stripe correction method of infrared focal plane image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105184740A true CN105184740A (en) | 2015-12-23 |
Family
ID=54906795
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510261452.XA Pending CN105184740A (en) | 2015-05-20 | 2015-05-20 | Non-uniform stripe correction method of infrared focal plane image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105184740A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106462957A (en) * | 2016-05-19 | 2017-02-22 | 深圳大学 | Method and system for removing stripe noise in infrared image |
CN106952254A (en) * | 2017-03-17 | 2017-07-14 | 北京思比科微电子技术股份有限公司 | The detection method of the bad row of image bad row |
CN109767403A (en) * | 2019-01-23 | 2019-05-17 | 西安科技大学 | A kind of infrared focal plane array strip noise cancellation method based on scene |
CN112150377A (en) * | 2020-08-27 | 2020-12-29 | 北京空间机电研究所 | Infrared image non-uniformity correction coefficient analytical solution alternative iteration optimization method |
CN116402725A (en) * | 2023-06-08 | 2023-07-07 | 四川工程职业技术学院 | Oblique strip removing method, device, equipment and medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102183307A (en) * | 2011-03-17 | 2011-09-14 | 北京航空航天大学 | Method for acquiring thermal radiation noise of gradient temperature gas medium in infrared imaging guidance |
WO2014198625A1 (en) * | 2013-06-13 | 2014-12-18 | Basf Se | Optical detector and method for manufacturing the same |
-
2015
- 2015-05-20 CN CN201510261452.XA patent/CN105184740A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102183307A (en) * | 2011-03-17 | 2011-09-14 | 北京航空航天大学 | Method for acquiring thermal radiation noise of gradient temperature gas medium in infrared imaging guidance |
WO2014198625A1 (en) * | 2013-06-13 | 2014-12-18 | Basf Se | Optical detector and method for manufacturing the same |
Non-Patent Citations (2)
Title |
---|
YI CHANG等: "Simultaneous Destriping and Denoising for Remote Sensing Images With Unidirectional Total Variation and Sparse Representation", 《IEEE GEOSCIENCE AND REMOTE SENSING LETTERS》 * |
彭小军等: "基于DSP+FPGA+ASIC 的实时红外图像处理系统", 《微计算机信息》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106462957A (en) * | 2016-05-19 | 2017-02-22 | 深圳大学 | Method and system for removing stripe noise in infrared image |
CN106462957B (en) * | 2016-05-19 | 2019-03-08 | 深圳大学 | The minimizing technology and system of fringes noise in a kind of infrared image |
CN106952254A (en) * | 2017-03-17 | 2017-07-14 | 北京思比科微电子技术股份有限公司 | The detection method of the bad row of image bad row |
CN109767403A (en) * | 2019-01-23 | 2019-05-17 | 西安科技大学 | A kind of infrared focal plane array strip noise cancellation method based on scene |
CN109767403B (en) * | 2019-01-23 | 2022-12-16 | 西安科技大学 | Infrared focal plane array strip noise elimination method based on scene |
CN112150377A (en) * | 2020-08-27 | 2020-12-29 | 北京空间机电研究所 | Infrared image non-uniformity correction coefficient analytical solution alternative iteration optimization method |
CN112150377B (en) * | 2020-08-27 | 2024-03-29 | 北京空间机电研究所 | Infrared image non-uniformity correction coefficient analysis solution alternation iteration optimization method |
CN116402725A (en) * | 2023-06-08 | 2023-07-07 | 四川工程职业技术学院 | Oblique strip removing method, device, equipment and medium |
CN116402725B (en) * | 2023-06-08 | 2023-08-22 | 四川工程职业技术学院 | Oblique strip removing method, device, equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108388896B (en) | License plate identification method based on dynamic time sequence convolution neural network | |
CN105184740A (en) | Non-uniform stripe correction method of infrared focal plane image | |
CN109214349B (en) | Object detection method based on semantic segmentation enhancement | |
CN106920220B (en) | The turbulent flow method for blindly restoring image optimized based on dark primary and alternating direction multipliers method | |
CN109509164B (en) | Multi-sensor image fusion method and system based on GDGF | |
CN109993091A (en) | A kind of monitor video object detection method eliminated based on background | |
CN110796616B (en) | Turbulence degradation image recovery method based on norm constraint and self-adaptive weighted gradient | |
Gilboa | Nonlinear eigenproblems in image processing and computer vision | |
CN105608679A (en) | Image denoising method integrated with structure tensor and non-local total variation | |
CN104680498A (en) | Medical image segmentation method based on improved gradient vector flow model | |
CN106097256A (en) | A kind of video image fuzziness detection method based on Image Blind deblurring | |
CN105427259A (en) | Multi-directional weighted TV and non local self-similarity regularization image deblurring method | |
CN104458766A (en) | Cloth surface blemish detection method based on structure texture method | |
CN103236052B (en) | Automatic cell localization method based on minimized model L1 | |
CN103268482A (en) | Low-complexity gesture extracting and gesture depth acquiring method | |
CN116258877A (en) | Land utilization scene similarity change detection method, device, medium and equipment | |
CN110021002A (en) | A kind of image interfusion method and device | |
CN113421210B (en) | Surface point Yun Chong construction method based on binocular stereoscopic vision | |
CN102184529B (en) | Empirical-mode-decomposition-based edge detecting method | |
Hou et al. | A variational formulation for fingerprint orientation modeling | |
CN113034371A (en) | Infrared and visible light image fusion method based on feature embedding | |
CN110175972B (en) | Infrared image enhancement method based on transmission map fusion | |
CN102314687B (en) | Method for detecting small targets in infrared sequence images | |
CN114140400B (en) | Method for detecting cigarette packet label defect based on RANSAC and CNN algorithm | |
CN116883447A (en) | Infrared image target and background separation method and system based on space-time tensor decomposition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20151223 |