CN112419194B - Short wave infrared band assisted remote sensing image thin cloud and fog correction method - Google Patents
Short wave infrared band assisted remote sensing image thin cloud and fog correction method Download PDFInfo
- Publication number
- CN112419194B CN112419194B CN202011339970.6A CN202011339970A CN112419194B CN 112419194 B CN112419194 B CN 112419194B CN 202011339970 A CN202011339970 A CN 202011339970A CN 112419194 B CN112419194 B CN 112419194B
- Authority
- CN
- China
- Prior art keywords
- cloud
- pixel
- pixels
- similar
- wave infrared
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000012937 correction Methods 0.000 title claims abstract description 33
- 238000001228 spectrum Methods 0.000 claims abstract description 15
- 230000003595 spectral effect Effects 0.000 claims abstract description 11
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 7
- 238000012163 sequencing technique Methods 0.000 claims abstract description 4
- 230000005855 radiation Effects 0.000 claims description 40
- 238000006073 displacement reaction Methods 0.000 claims description 20
- 238000009499 grossing Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 230000000452 restraining effect Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 239000003595 mist Substances 0.000 abstract description 7
- 230000003287 optical effect Effects 0.000 abstract description 3
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000002329 infrared spectrum Methods 0.000 abstract 1
- 230000015556 catabolic process Effects 0.000 description 3
- 238000006731 degradation reaction Methods 0.000 description 3
- 238000003702 image correction Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Images
Classifications
-
- G06T5/80—
-
- 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
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- Image Processing (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
Abstract
Aiming at the problem of cloud and mist coverage widely existing in visible light and near infrared spectrum bands of optical remote sensing images, a short wave infrared band assisted remote sensing image cloud and mist correction method is disclosed. Firstly, the short wave infrared band is used as a reference image to search a plurality of similar pixels of cloud area pixels in a non-cloud area one by utilizing the characteristics that the short wave infrared band is hardly influenced by cloud and fog and can provide complete earth surface information; secondly, selecting similar pixels by taking scattering rules among different wave bands as quantitative constraints, and sequencing the selected pixels by comprehensively considering spectral similarity; and finally, constructing a space spectrum Markov random field model with boundary conditions, solving by using a multi-label graph cut optimization algorithm, determining a global optimal similar pixel filling scheme, and realizing high-fidelity correction of the mist in the image. The invention has convenient operation, relatively easy satisfaction of data requirements, and stronger expandability and practical value.
Description
Technical Field
The invention belongs to the technical field of remote sensing image processing, and relates to a short wave infrared band assisted optical remote sensing image haze correction method.
Background
When the satellite (aircraft) sensor is used for observing the ground, the acquired optical image is frequently interfered by cloud and fog due to the influence of factors such as atmosphere and the like, and the effectiveness of data is reduced. Therefore, how to correct the remote sensing image acquired under the unfavorable imaging condition by using technical means is also a very important issue. The existing thin cloud and fog correction methods can be divided into two categories based on single-frame and multi-frame images. The single image-based correction method mainly utilizes the space and spectral characteristics of the image to realize the correction of cloud and mist, and comprises a spectral transformation method, a frequency domain method, dark target subtraction and the like; the correction method based on multiple images is usually to search effective auxiliary information of different wave bands, different time phases or different sensor images to correct the cloud and fog images, and commonly used methods include wavelet transformation, regression analysis, pixel replacement and the like. Methods based on multiple images often produce visually pleasing results due to the introduction of ancillary data. However, the strict requirement for auxiliary data causes the method to be difficult to implement and have poor universality. Compared with the prior knowledge expansion method, the single image-based method has wider application scenes by using different prior knowledge expansion. However, the potential application of the existing single image correction method to the data spectral band information is insufficient, and the fundamental reason of cloud degradation, namely the scattering rule, is not considered, so that the spectrum distortion occurs in the correction result. Therefore, it is necessary to further fully mine the potential information in the multispectral data and consider the scattering mechanism to perform high fidelity correction on the image.
Disclosure of Invention
The invention aims to provide a short-wave infrared band-assisted remote sensing image haze correction method, which aims at solving the problems that the existing single remote sensing image haze correction method is insufficient in consideration of physical degradation mechanism and insufficient in mining of potential auxiliary information contained in data.
The technical scheme adopted by the invention is as follows: a short wave infrared band assisted remote sensing image haze correction method comprises the following steps:
step 1: distinguishing a cloud area from a non-cloud area through a cloud detection algorithm or a manual drawing mode, and searching a plurality of similar pixels of cloud area pixels in the non-cloud area one by one;
step 2: optimizing the similar pixels by taking a scattering model as a criterion, judging whether the radiation difference between the cloud region pixels and the non-cloud region similar pixels in each band quantitatively meets the scattering model, comprehensively considering the radiation difference on the short wave infrared band, sequencing, and selecting the first m pixels with the highest similarity; the method specifically comprises the following substeps:
step 2.1: optimizing the similar pixels screened in the step 1, calculating the radiation difference value of each cloud area pixel and the non-cloud area similar pixels, and regarding the radiation difference value as the cloud and fog increment of each wave band;
step 2.2: the absolute radiation difference of the cloud area image element and the non-cloud area similar image element in a short wave infrared band and the fitting degree with the scattering model are comprehensively considered, namely when the cloud area image element and the non-cloud area similar image element have extremely high similarity or are completely consistent, the cloud area image element and the non-cloud area similar image element quantitatively meet the scattering model, otherwise, the cloud area image element and the non-cloud area similar image element deviate from the scattering model; sorting the similarity of the preferred pixels, and selecting the first m pixels with the highest similarity;
and step 3: a space spectrum Markov random field model with boundary conditions is constructed to determine a global optimal similar pixel replacement scheme, a spectrum smoothing item, namely the relative position of a similar pixel on a short wave infrared auxiliary image and an image to be repaired is kept unchanged, and a space smoothing item, namely the spatial continuity and consistency between four neighborhoods of the pixel to be corrected and four neighborhoods of the similar pixel are kept are comprehensively considered by the space spectrum Markov random field model, the space spectrum Markov random field model is solved by a multi-label graph cutting algorithm, the global optimal similar pixel replacement scheme is determined, and the haze correction is realized.
Further, in step 1, with the short wave infrared band as a reference, searching a plurality of similar pixels of cloud area pixels in non-cloud area one by one, wherein the screening rule is as follows: the difference of the absolute radiation value of a local window with a cloud area pixel as a center and a non-cloud area pixel as a center in a short wave infrared band is less than a threshold t1Expressed as:
|P(X′+L(X′))-P(X′)|=Δt≤t1 (1)
in the formula, X' represents a pixel at the same position of a cloud region pixel X in a visible light and near infrared band image on a short wave infrared band; l (X ') represents the displacement of the pixel X' with its similar pixels in the short wave infrared channel, and is represented by(s)x’,sy’) (ii) a P is a local area of n multiplied by n on the short wave infrared channel, delta t is an absolute radiation error, t1Is a threshold value.
Further, in step 2.1, the radiation difference between the cloud area pixel and the similar pixel in the visible light and the near infrared band is calculated as follows,
where ρ isc,b、ρc,g、ρc,r、ρc,nirRespectively representing the radiation differences of similar pixels of the cloud pixel and the non-cloud area in blue, green, red and near infrared bands, namely the radiation values of the cloud in corresponding bands; rhoiSubscripts nir, r, g, b represent the near infrared, red, green, and blue bands, for the pixel apparent radiation value of band i; x represents a cloud region picture element, and X + L (X') represents a similar picture element of a non-cloud region.
Further, when the cloud area pixel and the non-cloud area similar pixel in step 2.2 have extremely high similarity or are completely consistent, the radiation difference value of each band conforms to the scattering law constraint, taking the variable in formula (2) as an example, the constraint can be expressed as follows:
max[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]-min[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]<t2 (3)
wherein max-]And min-]The maximum and minimum operations are respectively shown, and the subscripts nir, r, g, b represent the near infrared, red, green and blue bands; t is t2For empirical thresholds, a smaller value indicates a stronger constraint on the applied scattering law, γi,jIs represented by rhoc,i,ρc,jThe scattering parameters obtained by calculation satisfy the following relation:
wherein, γi,jFor the atmospheric scattering parameter estimated by the bands i and j, λ is the center wavelength, ρc,iAnd ρc,jRespectively representing the cloud radiation values of the wave bands i and j, wherein the value ranges of i and j are sets [ nir, r, g, b ]]. Through the operation, the similar pixels in the step 1 can be optimized, and pixels with higher similarity and corresponding displacement graphs L (X') are obtained.
Further, in step 3, the position relationship between the cloud area pixel and the optimal replacement pixel is represented by a displacement graph, and the following model is constructed to determine the optimal displacement graph:
in the formula, omega represents a cloud area to be corrected, N is four neighborhoods on an image space, X' represents four neighborhood points of X, and the coordinates of a cloud area pixel X are (X, y); l represents a displacement label; "l (x) ═(s)x,sy) "denotes by (x + s)x,y+sy) The pixel of the location replaces pixel (x, y). Since X and X 'are coordinated in the image, the shift amounts denoted by L (X) and L (X') are theoretically the same;
first term E in formula (5)dIs a data item, when the tag is active, i.e. similar picture elements are located in non-cloud areas, Ed0; otherwise EdInfinity, +,; second item EspectaclIs a spectral smoothing term expressed as:
Espectral(L(X),L(X′))=||ρ(X+L(X))-ρ(X+L(X′))||2 (6)
the term is used for restraining the relative position of the cloud area pixel and the non-cloud area similar pixel in the short wave infrared band from being unchanged in the positions of the visible light and the near infrared band, so that the similar pixel searched in the short wave infrared is used for replacement during correction, wherein L (X') is obtained in step 2;
the third term is a spatial smoothing term, expressed as:
Espatial(L(X),L(X″))=||ρ(X+L(X))-ρ(X+L(X″))||2+||ρ(X″+L(X))-ρ(X″+L(X″))||2 (7)
this term is mainly used to constrain the calculated displacements of two adjacent pixels X and X ″ to be the same, i.e., L (X) L (X ″), so as to maintain spatial continuity and consistency between the corrected cloud region and the non-cloud region. Wherein, the calculation mode of L (X ') can refer to the calculation mode of L (X') in the steps 1 and 2. Alpha and beta are weight parameters used for measuring the influence of the spectral band information and the spatial information on the correction process.
The invention has the advantages that:
(1) the cloud interference of the image is effectively removed, a cloud-free image is obtained, and data support is provided for subsequent remote sensing monitoring application;
(2) the method can obtain more accurate similar pixel searching results only by utilizing different spectral band information of the image without the help of other time phase auxiliary information;
(3) introducing scattering rule constraint, optimizing similar pixels, and quantitatively improving image correction precision and spectral fidelity;
(4) aiming at the problem of thin cloud mist correction, a space spectrum Markov random field model is constructed to determine an optimal cloud area correction strategy.
Drawings
FIG. 1: a flow chart of an embodiment of the invention.
Detailed Description
For the purpose of facilitating the understanding and practice of the present invention, as will be described in further detail below with reference to the accompanying drawings and examples, it is to be understood that the examples described herein are for purposes of illustration and explanation, and are not intended to limit the invention.
Remote sensing imaging is susceptible to turbid atmosphere, resulting in cloud and fog occlusion frequently occurring in the acquired data. According to the characteristic that the cloud fog intensity is sharply weakened along with the increase of the wavelength, clear earth surface reference can be provided by using a high transmission waveband in multispectral data, similar pixels are screened in consideration of a scattering process, a space spectrum Markov random field is further constructed to realize a cloud area global optimal correction scheme, and high-fidelity image restoration is obtained.
Referring to fig. 1, the short-wave infrared band assisted remote sensing image haze correction method provided by the invention comprises the following steps:
step 1: cloud areas and non-cloud areas in the image are distinguished through a cloud detection algorithm or manual drawing and the like, and a plurality of similar pixels of cloud area pixels in the non-cloud areas are searched one by taking a high-transmission short wave infrared band as a clear ground reference. The screening rule is as follows: the difference of the absolute radiation value of a local window with a cloud area pixel as a center and a non-cloud area pixel as a center in a short wave infrared band is less than a threshold t1. Can be expressed as:
|P(X′+L(X′))-P(X′)|=Δt≤t1 (1)
in the formula, X' represents a pixel at the same position of a cloud region pixel X in a visible light and near infrared band image on a short wave infrared band; l (X ') represents the displacement of the pixel X' with its similar pixels in the short wave infrared channel, and is represented by(s)x′,sy′) (ii) a P is a local area of n multiplied by n on the short wave infrared channel, delta t is an absolute radiation error, t1As the threshold, 0.1 × n × n is empirically set.
Step 2: and optimizing the similar pixels on the basis of the scattering model. And judging whether the radiation difference between the bands of the cloud area pixels and the non-cloud area similar pixels quantitatively meets a scattering model, comprehensively considering the radiation difference on the short wave infrared band for sequencing, and selecting the first m pixels with the highest similarity, wherein the value of m is generally selected to be 8. The method specifically comprises the following substeps:
step 2.1: and (3) optimizing the similar pixels screened in the step (1), calculating the radiation difference value of each cloud area pixel and the non-cloud area similar pixels, and regarding the radiation difference value as the cloud and fog increment of each wave band. The radiation difference of the cloud area pixel and the similar pixel in the visible light and the near infrared band is calculated as follows:
where ρ isc,b、ρc,g、ρc,r、ρc,nirRespectively representing the radiation differences of the cloud pixel and the similar non-cloud pixel in blue, green, red and near infrared bands, namely the radiation values of the cloud in the corresponding bands; rhoiThe subscripts nir, r, g, b represent the near infrared, red, green and blue bands for the pixel apparent radiance value of band i. X represents a cloud region pixel, and X + L (X') represents a similar pixel of a visible light and near-infrared band non-cloud region; qualitatively, as the wavelength increases, its radiance difference should decrease.
Step 2.2: when the cloud area image element and the non-cloud area similar image element have extremely high similarity or are completely consistent, the scattering model is quantitatively satisfied, otherwise, the scattering model is deviated. Specifically, if the cloud area pixel and the non-cloud area similar pixel have extremely high similarity or completely coincide, the radiation difference value of each band should conform to the scattering law constraint, and taking the variable in equation (2) as an example, the constraint can be expressed as follows:
max[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]-min[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]<t2 (3)
wherein max [ alpha ], [ beta ], [ alpha ]]And min 2]The maximum and minimum operations are respectively shown, and the subscripts nir, r, g, b represent the near infrared, red, green and blue bands; t is t2An empirical threshold, the smaller the value, the stronger the applied scattering law constraint; γ is an atmospheric scattering parameter, which can be calculated from the cloud radiation relationship between different bands, for example, the cloud radiation relationship between bands i and j can be expressed as:
wherein, γi,jFor the atmospheric scattering parameter estimated by the bands i and j, λ is the center wavelength, ρc,iAnd ρc,jRespectively representing the cloud radiation values of the wave bands i and j, wherein the value ranges of i and j are sets [ nir, r, g, b ]]. Through the operation, the similar pixels in the step 1 can be optimized, and pixels with higher similarity and corresponding displacement graphs L (X') are obtained.
And step 3: the pixel replacement scheme for determining the global optimum by constructing a space spectrum Markov random field model aiming at the thin cloud and fog correction problem is stable and universal. The position relation between the cloud area pixel and the optimal replacement pixel can be represented by a displacement graph, and the optimal displacement graph can be determined through the following model:
in the formula, Ω represents a cloud area to be corrected, that is, the cloud area obtained in step 1 by a cloud detection method or a manual drawing method, N is a four-neighborhood region in an image space, and X ″ represents a four-neighborhood region point of X; l denotes a displacement label.
First term E in formula (5)dIs a data item. When the tag is active (i.e., similar picture elements are located in non-cloud regions), Ed0; otherwise EdInfinity. Second item EspectaclFor the spectral smoothing term, it can be expressed as:
Espectral(L(X),L(X′))=||ρ(X+L(X))-ρ(X+L(X′))||2 (6)
wherein rho is the apparent radiation value of the pixel. The method is used for restraining the relative position of the cloud area pixel and the non-cloud area similar pixel in the short wave infrared band from being unchanged in the positions of visible light and near infrared band, so that the similar pixel is used for replacement during correction. Wherein L (X') is calculated in advance by step 2.
The third term is a spatial smoothing term, which can be expressed as:
Espatial(L(X),L(X″))=||ρ(X+L(X))-ρ(X+L(X″))||2+||ρ(X″+L(X))-ρ(X″+L(X″))||2 (7)
this term is mainly used to constrain the calculated displacements of two adjacent pixels X and X ″ to be the same, i.e., L (X) L (X ″), so as to maintain spatial continuity and consistency between the corrected cloud region and the non-cloud region. Wherein, L (X) and L (X ') represent the same displacement, and the calculation manner of L (X ') can refer to the calculation manner of L (X ') in steps 1 and 2. Alpha and beta are weight parameters used for measuring the influence of the spectral band information and the spatial information on the correction process. Because the method disclosed by the patent mainly depends on short-wave infrared auxiliary information and the dependence on spatial information is less, alpha is 1, and beta is 0.5 in the experiment. The method adopts a multi-label graph cut algorithm to carry out optimization solution on the model, determines a global optimal replacement scheme and realizes image correction.
The method fully utilizes the difference of the influence of the mist on different wave bands, takes the short wave infrared wave band as clear earth surface reference, and considers the constraint of a scattering mechanism at the same time, thereby realizing the accurate searching mode of similar pixels; based on the method, a space spectrum Markov random field model is constructed to realize high-fidelity correction of the mist under different scenes. The method can accurately recover the degradation information caused by the cloud and fog, and has the advantages of low data requirement, easy realization, strong expandability and high practical value.
It should be understood that parts of the specification not set forth in detail are well within the prior art.
It should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (5)
1. A short wave infrared band assisted remote sensing image thin cloud and fog correction method is characterized by comprising the following steps:
step 1: distinguishing a cloud area from a non-cloud area through a cloud detection algorithm or a manual drawing mode, and searching a plurality of similar pixels of cloud area pixels in the non-cloud area one by one;
step 2: optimizing the similar pixels by taking a scattering model as a criterion, judging whether the radiation difference between the cloud region pixels and the non-cloud region similar pixels in each band quantitatively meets the scattering model, comprehensively considering the radiation difference on the short wave infrared band, sequencing, and selecting the first m pixels with the highest similarity; the method specifically comprises the following substeps:
step 2.1: optimizing the similar pixels screened in the step 1, calculating the radiation difference value of each cloud area pixel and the non-cloud area similar pixels, and regarding the radiation difference value as the cloud and fog increment of each wave band;
step 2.2: the absolute radiation difference of the cloud area image element and the non-cloud area similar image element in a short wave infrared band and the fitting degree with the scattering model are comprehensively considered, namely when the cloud area image element and the non-cloud area similar image element have extremely high similarity or are completely consistent, the cloud area image element and the non-cloud area similar image element quantitatively meet the scattering model, otherwise, the cloud area image element and the non-cloud area similar image element deviate from the scattering model; sorting the similarity of the preferred pixels, and selecting the first m pixels with the highest similarity;
and step 3: a space spectrum Markov random field model with boundary conditions is constructed to determine a global optimal similar pixel replacement scheme, a spectrum smoothing item, namely the relative position of a similar pixel on a short wave infrared auxiliary image and an image to be repaired is kept unchanged, and a space smoothing item, namely the spatial continuity and consistency between four neighborhoods of the pixel to be corrected and four neighborhoods of the similar pixel are kept are comprehensively considered by the space spectrum Markov random field model, the space spectrum Markov random field model is solved by a multi-label graph cutting algorithm, the global optimal similar pixel replacement scheme is determined, and the haze correction is realized.
2. The short-wave infrared band-assisted remote sensing image haze correction method according to claim 1, characterized in that: in the step 1, a short wave infrared band is used as a reference, a plurality of similar pixels of cloud area pixels in non-cloud areas are searched one by one, and the screening rule is as follows: the difference of the absolute radiation value of a local window with a cloud area pixel as a center and a non-cloud area pixel as a center in a short wave infrared band is less than a threshold t1Expressed as:
|P(X′+L(X′))-P(X′)|=Δt≤t1 (1)
in the formula, X' represents a pixel at the same position of a cloud region pixel X in a visible light and near infrared band image on a short wave infrared band; l (X ') represents the displacement of the pixel X' with its similar pixels in the short wave infrared channel, and is represented by(s)x’,sy’) (ii) a P is a local area of n multiplied by n on the short wave infrared channel, delta t is an absolute radiation error, t1Is a threshold value.
3. The short-wave infrared band-assisted remote sensing image haze correction method according to claim 1, characterized in that: in step 2.1, the radiation difference of the cloud area pixel and the similar pixel in the visible light and the near infrared band is calculated as follows:
where ρ isc,b、ρc,g、ρc,r、ρc,nirRespectively representing the radiation differences of similar pixels of the cloud pixel and the non-cloud area in blue, green, red and near infrared bands, namely the radiation values of the thin cloud in corresponding bands; rhoiSubscripts nir, r, g, b represent the near infrared, red, green, and blue bands, for the pixel apparent radiation value of band i; x represents a cloud area pixel, L (X ') represents the displacement amount between the pixel X ' and a similar pixel on a short-wave infrared channel, and X + L (X ') represents a similar pixel in a non-cloud area.
4. The short-wave infrared band-assisted remote sensing image haze correction method according to claim 3, characterized in that: in step 2.2, when the cloud area pixel and the non-cloud area similar pixel have extremely high similarity or are completely consistent, the radiation difference value of each wave band conforms to the scattering law constraint, taking the variable in the formula (2) as an example, the constraint can be expressed as follows:
max[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]-min[γnir,r,γnir,g,γnir,b,γr,g,γr,b,γg,b]<t2 (3)
wherein max-]And min-]The maximum and minimum operations are respectively shown, and the subscripts nir, r, g, b represent the near infrared, red, green and blue bands; t is t2For empirical thresholds, a smaller value indicates a stronger constraint on the applied scattering law, γi,jIs represented by rhoc,i,ρc,jThe scattering parameters obtained by calculation satisfy the following relation:
wherein, γi,jFor the atmospheric scattering parameter estimated by the bands i and j, λ is the center wavelength, ρc,iAnd ρc,jRespectively representing the cloud radiation values of the wave bands i and j, wherein the value ranges of i and j are sets [ nir, r, g, b ]]。
5. The short-wave infrared band-assisted remote sensing image haze correction method according to claim 1, characterized in that: in step 3, the position relation between the cloud area pixels and the optimal replacement pixels is represented by a displacement graph, and the following model is constructed to determine the optimal displacement graph:
in the formula, omega represents a cloud area to be corrected, N is four neighborhoods on an image space, X' represents four neighborhood points of X, and the coordinates of a cloud area pixel X are (X, y); l represents a displacement label; "l (x) ═(s)x,sy) "denotes by (x + s)x,y+sy) The pixel at the position replaces the pixel (X, y), and since X and X 'are consistent in coordinates in the image, the offset amounts denoted by L (X) and L (X') are theoretically the same;
first term E in formula (5)dIs a data item, when the tag is active, i.e. similar picture elements are located in non-cloud areas, Ed0; otherwise Ed=+∞;
Second item EspectaclIs a spectral smoothing term expressed as:
Espectral(L(X),L(X′))=||ρ(X+L(X))-ρ(X+L(X′))||2 (6)
the method is used for restraining the relative position of cloud region pixels and non-cloud region similar pixels in a short wave infrared band from being unchanged in the positions of visible light and near infrared bands, so that the similar pixels searched in the short wave infrared band are used for replacement during correction, wherein L (X ') represents the displacement between pixel X' and the similar pixels thereof on a short wave infrared channel;
the third term is a spatial smoothing term, expressed as:
Espatial(L(X),L(X″))=||ρ(X+L(X))-ρ(X+L(X″))||2+||ρ(X″+L(X))-ρ(X″+L(X″))||2(7)
this term is mainly used to constrain the calculated displacements of two adjacent pixels X and X ″ to be the same, i.e., L (X) ═ L (X ″), and is used to maintain spatial continuity and consistency of the corrected cloud and non-cloud regions; wherein, the calculation mode of L (X ') is the same as that of L (X'), and alpha and beta are weight parameters for measuring the influence of the spectrum information and the space information on the correction process.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011339970.6A CN112419194B (en) | 2020-11-25 | 2020-11-25 | Short wave infrared band assisted remote sensing image thin cloud and fog correction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011339970.6A CN112419194B (en) | 2020-11-25 | 2020-11-25 | Short wave infrared band assisted remote sensing image thin cloud and fog correction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112419194A CN112419194A (en) | 2021-02-26 |
CN112419194B true CN112419194B (en) | 2022-04-12 |
Family
ID=74842369
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011339970.6A Active CN112419194B (en) | 2020-11-25 | 2020-11-25 | Short wave infrared band assisted remote sensing image thin cloud and fog correction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112419194B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105138994A (en) * | 2015-08-31 | 2015-12-09 | 中国科学院遥感与数字地球研究所 | Water bloom identification method and device based on hyperspectral remote sensing image |
CN110335208A (en) * | 2019-06-10 | 2019-10-15 | 武汉大学 | It is a kind of based on the spissatus minimizing technology of high-resolution remote sensing image gradually corrected |
CN111353959A (en) * | 2020-03-02 | 2020-06-30 | 莫登奎 | Efficient method suitable for removing haze of large-scale optical remote sensing image |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090272888A1 (en) * | 2008-01-25 | 2009-11-05 | Nugent Paul W | Thermal infrared imaging system and associated methods for radiometric calibration |
-
2020
- 2020-11-25 CN CN202011339970.6A patent/CN112419194B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105138994A (en) * | 2015-08-31 | 2015-12-09 | 中国科学院遥感与数字地球研究所 | Water bloom identification method and device based on hyperspectral remote sensing image |
CN110335208A (en) * | 2019-06-10 | 2019-10-15 | 武汉大学 | It is a kind of based on the spissatus minimizing technology of high-resolution remote sensing image gradually corrected |
CN111353959A (en) * | 2020-03-02 | 2020-06-30 | 莫登奎 | Efficient method suitable for removing haze of large-scale optical remote sensing image |
Non-Patent Citations (2)
Title |
---|
《Removal of Thin Cirrus Scattering Effects in Landsat8 OLI Images Using the Cirrus Detecting Channel》;Bo-Cai Gao等;《Remote Sensing》;20170812;全文 * |
《联合统计信息与散射模型的GF-5 AHSI可见光影像薄云校正》;张弛 等;《遥感学报》;20200430;第24卷(第4期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112419194A (en) | 2021-02-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10325151B1 (en) | Method of extracting image of port wharf through multispectral interpretation | |
Li et al. | Multi-feature combined cloud and cloud shadow detection in GaoFen-1 wide field of view imagery | |
Makarau et al. | Haze detection and removal in remotely sensed multispectral imagery | |
CN103020913B (en) | Remote-sensing image stripe noise removing method based on segmented correction | |
US11227367B2 (en) | Image processing device, image processing method and storage medium | |
US20230094679A1 (en) | Satellite Image Sequence Cloud Region Repairing Method and Apparatus | |
US20100142842A1 (en) | Image processing device for determining cut lines and related methods | |
CN109446962B (en) | Land cover annual change detection method and system | |
Cui et al. | Multiscale Intensity Propagation to Remove Multiplicative Stripe Noise From Remote Sensing Images. | |
CN108830814A (en) | A kind of relative radiometric correction method of remote sensing image | |
CN111489299B (en) | Defogging method for multispectral remote sensing satellite image | |
US20010048758A1 (en) | Image position matching method and apparatus | |
US11386649B2 (en) | Automated concrete/asphalt detection based on sensor time delay | |
CN112419194B (en) | Short wave infrared band assisted remote sensing image thin cloud and fog correction method | |
US20050283310A1 (en) | Algorithm for line tracking using a circular sector search window | |
CN105405112A (en) | Multispectral satellite image distance deviation index defogging method | |
Vijayaraj et al. | Concepts of image fusion in remote sensing applications | |
CN102629369B (en) | Single color image shadow removal method based on illumination surface modeling | |
Champion | Automatic detection of clouds and shadows using high resolution satellite image time series | |
CN102915524B (en) | Method for eliminating shadow based on match of inside and outside check lines of shadow area | |
CN113436123B (en) | High-resolution SAR and low-resolution multispectral image fusion method based on cloud removal-resolution improvement cooperation | |
Shedlovska et al. | Shadow removal algorithm for remote sensing imagery | |
Sreekar et al. | Enhanced Efficient Image Dehazing for Onboard Satellite Imagery Processing | |
CN116503742B (en) | Remote sensing image cloud replacement method based on super-pixel and graph cut algorithm | |
Yang et al. | Optical Remote Sensing Image Optimized Dehazing Algorithm Based On Hot |
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 |