CN112927161B - Method and device for enhancing multispectral remote sensing image and storage medium - Google Patents
Method and device for enhancing multispectral remote sensing image and storage medium Download PDFInfo
- Publication number
- CN112927161B CN112927161B CN202110272277.XA CN202110272277A CN112927161B CN 112927161 B CN112927161 B CN 112927161B CN 202110272277 A CN202110272277 A CN 202110272277A CN 112927161 B CN112927161 B CN 112927161B
- Authority
- CN
- China
- Prior art keywords
- image
- principal component
- multispectral
- enhancing
- weighted
- 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 41
- 230000002708 enhancing effect Effects 0.000 title claims abstract description 33
- 238000000513 principal component analysis Methods 0.000 claims abstract description 48
- 230000009466 transformation Effects 0.000 claims abstract description 37
- 230000004048 modification Effects 0.000 claims abstract description 11
- 238000012986 modification Methods 0.000 claims abstract description 11
- 238000000605 extraction Methods 0.000 claims abstract description 8
- 230000009467 reduction Effects 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 39
- 239000011159 matrix material Substances 0.000 claims description 18
- 239000013598 vector Substances 0.000 claims description 11
- 230000003247 decreasing effect Effects 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 9
- 230000001186 cumulative effect Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 239000003795 chemical substances by application Substances 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000003491 array Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 239000003623 enhancer Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- 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
- G06T2207/10036—Multispectral image; Hyperspectral image
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method, a device and a storage medium for enhancing a multispectral remote sensing image, wherein the method comprises the following steps: performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting main component components which are not related to each other; modifying the template coefficient by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficient into the first principal component after image weighted PCA transformation for enhancing the texture and edge details of the image; and after the first principal component is subjected to fractional order differential enhancement, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function. The device comprises: the device comprises a compensation and extraction module, a modification and enhancement module and an adjustment module. The invention effectively enhances the definition and contrast of the image.
Description
Technical Field
The invention relates to the field of multispectral remote sensing images, in particular to a method and a device for enhancing a multispectral remote sensing image and a storage medium.
Background
Multispectral remote sensing images are widely used in military, geoscience, agriculture and astronomy, as well as many other fields, and the demand for higher quality multispectral remote sensing images is rapidly increasing. In the imaging process, due to the influence of factors such as a sensor, environment and the like, some interference and noise are introduced, so that the problems of visual quality reduction, contrast reduction, detail loss and the like of an image occur.
Thus, image enhancement techniques are used as a fundamental step in remote sensing image processing and analysis applications to improve human perception and for subsequent machine analysis. The multispectral remote sensing image is obtained by shooting a plurality of single-waveband remote sensing images in surface feature radiation, and obtained image data comprises spectral information of a plurality of wavebands. Although a large amount of research results are available in the existing single-waveband remote sensing image enhancement method, multispectral remote sensing image enhancement still needs to be solved. This is because multispectral image data combines information of image spatial dimension with information of spectral band, so that information enhancement of spatial dimension and feature extraction of spectral dimension need to be considered at the same time, which brings difficulty to multispectral enhancement research.
Disclosure of Invention
Aiming at the problem that the remote sensing image enhancement algorithm of a single waveband cannot be combined with effective information of images of ground objects on other wavebands, the invention provides a method, a device and a storage medium for enhancing a multispectral remote sensing image, the multispectral image processed by the method has obviously enhanced edge details and can retain more texture characteristics compared with the original image; the invention effectively enhances the definition and the contrast of the image, and solves the problem that the single-waveband remote sensing image enhancement algorithm cannot fully reflect the detail characteristics of ground objects in different wavebands, and the details are described as follows:
in a first aspect, a method for enhancing a multispectral remote sensing image, the method comprising:
performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting main component components which are not related to each other;
modifying the template coefficient by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficient into the first principal component after image weighted PCA transformation for enhancing the texture and edge details of the image;
and after the first principal component is subjected to fractional order differential enhancement, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function.
In one implementation, the modifying the template coefficient by using the logarithmic function to form the fractional order mask for maintaining the correlation between the adjacent pixels specifically includes:
endowing different weights to adjacent pixels according to the difference of the actual distance, fitting the weights by adopting a decreasing logarithmic function, and obtaining an improved template in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, the eight pixels around the central pixel are weighted by the weight of the logarithmic function fitting to obtain a template in eight directions, the template is expanded by taking the central pixel as an origin and added to obtain an improved template, and the normalized improved template is added to one half of the mask to obtain a final differential template.
In one implementation, the fractional order mask is a 5 x 5 mask.
In a second aspect, an apparatus for enhancing a multispectral remote sensing image, the apparatus comprising:
the compensation and extraction module is used for reducing the dimension of the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness and extracting main component components which are not related to each other;
the modification and enhancement module is used for modifying the template coefficients by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficients into the first principal component after the image weighted PCA transformation for enhancing the texture and edge details of the image;
and the adjusting module is used for performing fractional order differential enhancement on the first principal component, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function.
In one implementation, the modification and enhancement module includes:
the modification submodule is used for endowing different weights to adjacent pixels according to the difference of actual distances, fitting the weights by adopting a decreasing logarithmic function, and obtaining an improved template in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, the eight pixels around the central pixel are weighted by the weight of the logarithmic function fitting to obtain a template in eight directions, the template is expanded by taking the central pixel as an origin and added to obtain an improved template, and the normalized improved template is added to one half of the mask to obtain a final differential template.
And the enhancement sub-module is used for introducing the improved template coefficient into the first principal component after the image weighted PCA transformation, and enhancing the texture and edge details of the image.
In a third aspect, an apparatus for enhancing a multispectral remote sensing image, the apparatus further comprising:
a processor and a memory, the memory having stored therein program instructions, the processor calling the program instructions stored in the memory to cause the apparatus to perform the method steps of any of the first aspects.
In a fourth aspect, a computer-readable storage medium, storing a computer program comprising program instructions which, when executed by a processor, cause the processor to carry out the method steps of any of the first aspects.
The technical scheme provided by the invention has the beneficial effects that:
1. the method can quickly and effectively enhance the multispectral image, and combines the information enhancement of the spatial dimension with the feature extraction of the spectral dimension, so that all wave bands of the multispectral image can be fully utilized;
2. the image processed by the method not only can well enhance the edge texture details of the multispectral image, but also can effectively improve the definition and contrast of the image.
Drawings
FIG. 1 is a flow chart of a method of enhancing a multi-spectral remote sensing image;
FIG. 2 is a schematic diagram of seven bands of a multi-spectral image;
FIG. 3 is a schematic diagram of the target image after enhancement processing of FIG. 2;
FIG. 4 is a schematic diagram of seven bands of another multispectral image;
FIG. 5 is a schematic diagram of the target image after enhancement processing of FIG. 4;
FIG. 6 is a schematic diagram of seven bands of another multispectral image;
FIG. 7 is a schematic diagram of the target image after enhancement processing of FIG. 6;
FIG. 8 is a schematic diagram of a device for enhancing a multispectral remote sensing image;
FIG. 9 is a schematic structural diagram of a modification and enhancement module;
fig. 10 is another structural diagram of the enhancement device for the multispectral remote sensing image.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention are described in further detail below.
A method of enhancing a multispectral image, see fig. 1, the method comprising the steps of:
step 101: performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting each independent principal component which can more express image detail characteristics;
the variance is divided into a first principal component, a second principal component, … … and an nth principal component in sequence from large to small. Since the variance is reduced in sequence, the information contained in each principal component is also reduced in sequence, and usually, most of the information is contained in the first 95% of the principal components, so that the first 95% of the principal components are retained, the last 5% of the principal components are discarded, and meanwhile, in order to further enhance the details of the image, fractional order differential enhancement is separately performed on the first principal component with the largest information amount.
Step 102: in consideration of the difference of actual distances between surrounding pixels and a central pixel, modifying the template coefficient by using a logarithmic function to form a fractional mask capable of better keeping the correlation of adjacent pixels, and introducing the improved mask with preset values into the first principal component after image weighted PCA (principal component analysis) conversion for enhancing the texture and edge details of the image;
in specific implementation, the embodiment of the present invention takes the 5 × 5 size as an example for description, and may be set to other sizes according to needs in practical applications, which is not limited in the embodiment of the present invention.
Step 103: after fractional order differential enhancement is carried out on the first principal component improved in the step 102, weighted PCA inverse transformation is carried out on the first 95% principal component components, and brightness adjustment is carried out on the multispectral image after weighted PCA inverse transformation by adopting a brightness adjustment function;
firstly, the global mean value of each wave band image in the multispectral image is adjusted by utilizing linear transformation to prevent the whole image from being too dark or too bright, and the local mean value of each wave band image in the multispectral image is adjusted by utilizing a gamma correction algorithm to prevent the local details of the image from being lost, so that the contrast of the multispectral image is improved.
In a specific implementation, the PCA generally takes the first K principal components with variance cumulative contribution rate exceeding 85%, and performs reconstruction, that is, reduces to K dimension, and the remaining principal components mainly contain noise and redundancy, and thus can be discarded. The embodiment of the invention takes the principal component with the variance cumulative contribution rate of more than 95 percent to reserve most information.
In summary, in the embodiment of the present invention, through the steps 101 to 103, the definition (i.e., the texture and edge details of the enhanced image) and the contrast of the image are effectively enhanced, and the problem that the single-band remote sensing image enhancement algorithm cannot sufficiently represent the detail features of the ground object in different bands is solved.
In the following, with reference to fig. 2 and a specific calculation formula, a method for enhancing a multispectral image in the above embodiment is detailed and expanded, and the method includes the following steps:
the embodiment of the invention adopts an experimental object which is a multispectral remote sensing image obtained by a TM (thematic mapper) carried by a United states land satellite Landsat-5, and provides a method for enhancing the multispectral remote sensing image aiming at the problem that a single-waveband remote sensing image enhancement algorithm cannot be combined with effective information of other wavebands.
The technical scheme of the invention is further specifically described below with reference to the accompanying drawings. Each step is as follows:
step 201: processing the multispectral remote sensing image by adopting weighted PCA transformation, and compensating the original image of each wave band by utilizing average gradient and texture roughness so as to extract characteristic components with higher definition and contrast;
(1) data set X ═ X with n bands1,x2,…,xn}∈RmWherein the wave band xk(k-1, 2, … n) is a m-dimensional column vector, each sample is given a weight value according to the role of the sample in the enhancement process, and the weight vector is defined as γ - γ ═ γ1,γ2,…,γnAnd i.e.:
where α is the adjustment parameter, AG is the mean gradient, which means the mean value of the rate of change of the grey level, used to indicate the image sharpness, expressed as:
wherein M is the number of rows of the pixel matrix of the image, N is the number of columns of the pixel matrix of the image, Ik(i, j) is the image pixel coordinates, C is the coarseness, a physical quantity used to describe the granularity size and distribution in the texture, the larger the texel size, the further the distance between the elements, and the coarser the texture. Smooth regions in the image have less roughness, while regions with rich texture have relatively greater roughness, expressed as:
wherein σk 2Is the local variance of the image, taking a 3 x 3 size neighborhood.
(2) Averaging the data from the original data set to
Thus, a weighted data set X can be obtainedWComprises the following steps:
thus, the covariance matrix ΣXWComprises the following steps:
(3) for matrix ΣXWPerforming characteristic decomposition to obtain the characteristic value lambda of the covariance matrixkAnd the corresponding feature vector phikSorting the eigenvalues in descending orderk。
The covariance matrix represents a group of bases under which the data has the maximum variance in the direction of the eigenvector, and the eigenvalue is the variance of the data in the direction of the eigenvector, and the magnitude of the variance represents the magnitude of the information quantity. The feature vectors are principal component components, the feature values give the information content carried by each principal component, and the feature values are arranged in a descending order, namely the feature vectors are arranged in a descending order, namely the principal component components are arranged in a descending order according to importance. And then calculating the cumulative variance contribution rate of the principal components, namely the proportion of the sum of the characteristic values of the first d principal components in the sum of all the characteristic values.
The larger the ratio is, the more comprehensively the first d principal components represent the information of the original data. Principal component cumulative variance contribution rate MdThe calculation formula is as follows:
(4) selecting the first d characteristic values of lambda ═ diag [ lambda ]1,λ2,…λd]And the corresponding eigenvector phi ═ phi [ ]1,φ2,…,φd]As a basis for subspace, to contribute cumulative varianceThe contribution rate can meet more than 95 percent. Finally, the extracted principal component is used for replacing the original n variables to generate a projection matrix Y phiTXW。
Step 202: an improved 5 x 5 fractional order differential template is used to enhance the first principal component of the weighted PCA extraction.
(1) Considering that the actual distances of the eight surrounding pixels from the central pixel are different, the correlations between the pixels and the central pixel are different, and the closer the pixel is to the central pixel, the greater the similarity between the pixel and the central pixel is, so that the adjacent pixels are given different weights according to the difference of the actual distances, i.e. the smaller the distance is given a higher weight, and the larger the distance is given a lower weight. Therefore, the embodiment of the invention adopts the decreasing logarithmic function fitting weight, the distances x are respectively 1 from small to large,
2，
substituting x into the function to obtain the weight w as follows:
(2) after normalization, improved templates in x, y negative direction and upper right diagonal direction can be obtained:
TABLE 1 improved template for negative (i.e. 180) orientation
Table 2y negative (i.e. 270) modified template
Table 3 improved template for the upper right diagonal (i.e., 135 °) of
Wherein, a1，a0The original coefficient of the fractional order differential, a, defined for Grunwald-Letnikov (G-L, well known to those skilled in the art)0＝1，a1V is the fractional order.
(3) To simplify the expression, let:
wherein, c1……c9To adjust the parameters.
Eight pixels around the center pixel are weighted by the weight of the logarithmic function fitting to find 3 × 3 masks in 8 directions (0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, 315 °) such as the x, y negative direction, they are extended to 5 × 5 masks with the center pixel as the origin, all 5 × 5 masks are superimposed to obtain a new 5 × 5 mask, and the new 5 × 5 mask is expressed by the parameters in equation (8), as shown in table 3.
TABLE 3 modified 5X 5 templates
| c5a2 | (c4+c9)a2 | (2c3+c8)a2 | (c4+c9)a2 | c5a2 |
| (c4+c9)a2 | (c2+2c7)a1 | (2c1+3c6)a1 | (c2+2c7)a1 | (c4+c9)a2 |
| (2c3+c8)a2 | (2c1+3c6)a1 | 8a0 | (2c1+3c6)a1 | (2c3+c8)a2 |
| (c4+c9)a2 | (c2+2c7)a1 | (2c1+3c6)a1 | (c2+2c7)a1 | (c4+c9)a2 |
| c5a2 | (c4+c9)a2 | (2c3+c8)a2 | (c4+c9)a2 | c5a2 |
And adding half of the normalized 5 × 5 mask to obtain the final 5 × 5 differential template in table 3, so as to ensure that the detail enhancement is performed on the basis of the overall enhancement of the image.
In a specific implementation, the enhancement is performed by using a 5 × 5 mask, which is equivalent to performing spatial filtering on the image by using the 5 × 5 mask, and the spatial filtering is to move the mask point by point in the image to be processed. That is, the center point of the 5 × 5 mask is overlapped with the pixel point (x, y) of the image, and then the mask coefficient is multiplied by the pixel values included in the 5 × 5 neighborhood around the point (x, y) respectively to sum up, so as to obtain a value replacing the pixel value of the point (x, y). That is, the template is extended outward by the center point, so that there are only 3 × 3, 5 × 5, 7 × 7, 9 × 9, etc., of which 3 × 3, 5 × 5 are more common.
Since the 5 × 5 template is derived from the 3 × 3 template, the 7 × 7 template needs to be derived from the 5 × 5 template, and the result obtained by assigning the coefficients of the 3 × 3 template according to the weights is different from the result obtained by assigning the coefficients of the 5 × 5 template according to the weights. The 7 × 7 template is not effective in use, and therefore the embodiment of the present invention will be described by taking a 5 × 5 template as an example.
Step 203: and enhancing the global contrast and the local details of all wave bands of the multispectral image after the weighted PCA inverse transformation by adopting a brightness adjusting function, and superposing and synthesizing all wave band images to obtain a finally enhanced gray level image.
The formula of the brightness adjustment function is as follows:
wherein x iskIs the image of the k wave band in the multispectral image, (x)k)max，(xk)minAre respectively xkMaximum and minimum gray values of xk' As an enhanced image, ElIs a single band local mean, EgIs a sheetGlobal mean of band, EsGamma is the gamma coefficient for the global mean of all bands.
From the local details of a single band, the local mean E of a bandlLess than its global mean EgPerforming gamma compression on the wave band to perform brightness compensation on local details of the image, and otherwise performing gamma expansion; globally, when the global mean E of a bandlGlobal mean value E of all bands less than p timessThen, the whole enhancement is carried out on the wave band, if the global mean value E of the single wave bandlGlobal mean E of all bands greater than q timessThe brightness value of the wave band is reduced, and the loss of details of the dark part of the wave band is avoided.
Fig. 2, 4, and 6 show raw multispectral image data acquired by three sets of Landsat-5TM sensors. Each set of data contains 7 bands of 500 × 500 pixels, TM1 (blue band), TM2 (red band), TM3 (green band), TM4 (near infrared band), TM5 (mid infrared band), TM6 (thermal infrared band), and TM7 (far infrared band). Fig. 3, 5 and 7 show the enhanced images of fig. 2, 4 and 6, respectively. As can be seen by comparing the enhanced images 2, 4, 6 with the original images 3, 5, 7, the enhanced images have better visual effect, the global features of the original images are preserved, and the edge and texture details of the images are highlighted.
Based on the same inventive concept, as an implementation of the above method, referring to fig. 8, an embodiment of the present invention further provides an apparatus for enhancing a multispectral remote sensing image, where the apparatus includes:
the compensation and extraction module 1 is used for performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting main component components which are not related to each other;
the modification and enhancement module 2 is used for modifying the template coefficients by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficients into the first principal component after the image weighted PCA transformation for enhancing the texture and edge details of the image;
and the adjusting module 3 is used for performing fractional order differential enhancement on the first principal component, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function.
In one implementation, referring to fig. 9, the modification and enhancement module 2 includes:
the modification submodule 21 is used for endowing adjacent pixels with different weights according to the difference of actual distances, fitting the weights by adopting a decreasing logarithmic function, and obtaining an improved template in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, the eight pixels around the central pixel are weighted by the weight of the logarithmic function fitting to obtain a template in eight directions, the template is expanded by taking the central pixel as an origin and added to obtain an improved template, and the normalized improved template is added to one half of the mask to obtain a final differential template.
An enhancer module 22 for introducing improved template coefficients into the first principal component after image weighted PCA transformation to enhance texture and edge details of the image.
It should be noted that the device description in the above embodiments corresponds to the description of the method embodiments, and the details of the embodiments of the present invention are not repeated herein.
The execution main bodies of the modules and units can be devices with calculation functions, such as a computer, a single chip microcomputer and a microcontroller, and in the specific implementation, the execution main bodies are not limited in the embodiment of the invention and are selected according to the requirements in practical application.
Based on the same inventive concept, an embodiment of the present invention further provides an apparatus for enhancing a multispectral remote sensing image, with reference to fig. 10, where the apparatus includes: a processor 4 and a memory 5, the memory 5 having stored therein program instructions, the processor 4 calling the program instructions stored in the memory 5 to cause the apparatus to perform the following method steps in an embodiment:
performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting main component components which are not related to each other;
modifying the template coefficient by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficient into the first principal component after image weighted PCA transformation for enhancing the texture and edge details of the image;
and after the first principal component is subjected to fractional order differential enhancement, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function.
In one implementation, modifying the template coefficients by using a logarithmic function to form a fractional mask that maintains the correlation of adjacent pixels specifically includes:
endowing different weights to adjacent pixels according to the difference of the actual distance, fitting the weights by adopting a decreasing logarithmic function, and obtaining an improved template in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, the eight pixels around the central pixel are weighted by the weight of the logarithmic function fitting to obtain a template in eight directions, the template is expanded by taking the central pixel as an origin and added to obtain an improved template, and the normalized improved template is added to one half of the mask to obtain a final differential template.
In one implementation, the fractional order mask is a 5 x 5 mask.
It should be noted that the device description in the above embodiments corresponds to the method description in the embodiments, and the embodiments of the present invention are not described herein again.
The execution main bodies of the processor 4 and the memory 5 may be devices having a calculation function, such as a computer, a single chip, a microcontroller, and the like, and in the specific implementation, the execution main bodies are not limited in the embodiment of the present invention, and are selected according to the needs in the practical application.
The memory 5 and the processor 4 transmit data signals through the bus 6, which is not described in detail in the embodiment of the present invention.
Based on the same inventive concept, embodiments of the present invention further provide a computer-readable storage medium, where the storage medium includes a stored program, and when the program runs, the apparatus on which the storage medium is located is controlled to execute the method steps in the foregoing embodiments.
The computer readable storage medium includes, but is not limited to, flash memory, hard disk, solid state disk, and the like.
It should be noted that the descriptions of the readable storage medium in the above embodiments correspond to the descriptions of the method in the embodiments, and the descriptions of the embodiments of the present invention are not repeated here.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. The procedures or functions according to the embodiments of the invention may be carried out in whole or in part when the computer program instructions are loaded and executed on a computer.
The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored on or transmitted over a computer-readable storage medium. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium or a semiconductor medium, etc.
In the embodiment of the present invention, except for the specific description of the model of each device, the model of other devices is not limited, as long as the device can perform the above functions.
Those skilled in the art will appreciate that the drawings are only schematic illustrations of preferred embodiments, and the above-mentioned serial numbers of the embodiments of the present invention are only for description and do not represent the merits of the embodiments.
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, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (5)
1. A method for enhancing a multispectral remote sensing image, the method comprising:
1) performing dimensionality reduction on the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness, and extracting main component components which are not related to each other;
2) modifying the template coefficient by utilizing a logarithmic function to form a fractional order mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficient into the first principal component after image weighted PCA transformation for enhancing the texture and edge details of the image;
3) after fractional order differential enhancement is carried out on the first principal component, weighted PCA inverse transformation is carried out on the principal component of the previous preset value, and the multispectral image after weighted PCA inverse transformation is adjusted by adopting a brightness adjusting function;
wherein, the step 1) is specifically as follows:
(1) data set X ═ X with n bands1,x2,…,xn}∈RmWave band xkThe method is characterized in that the method is an m-dimensional column vector, wherein k is 1,2 and … n, each sample is endowed with a weight according to different roles of each sample in the enhancing process, and the weight vector is defined as gamma, wherein gamma is { gamma ═ gamma1,γ2,…,γnAnd i.e.:
where α is the adjustment parameter, AG is the mean gradient, which means the mean value of the rate of change of the grey level, used to indicate the image sharpness, expressed as:
where M is the number of rows of the pixel matrix of the image and N is the pixel moment of the imageNumber of arrays, Ik(i, j) are image pixel coordinates, and C is roughness, a physical quantity used to describe grain size and distribution in texture;
wherein σk 2Is the local variance of the image;
(2) averaging the data from the original data set to
Obtaining a weighted data set XWComprises the following steps:
covariance matrix
Comprises the following steps:
(3) for matrix
Performing characteristic decomposition to obtain the characteristic value lambda of the covariance matrixkAnd the corresponding feature vector phikSorting the eigenvalues in descending orderk；
Principal component cumulative variance contribution rate MdThe calculation formula is as follows:
(4) selecting the first d characteristic values of lambda ═ diag [ lambda ]1,λ2,…λd]And the corresponding eigenvector phi ═ phi [ ]1,φ2,…,φd]As a basis of the subspace, the extracted principal component is used to replace the original n variables, and a projection matrix Y ═ Φ is generatedTXW；
Wherein, the step 2) is as follows:
according to the difference of the actual distance, different weights are given to the adjacent pixels, and the weights are fitted by adopting a decreasing logarithmic function, so that the obtained weights w are as follows in sequence:
obtaining improved templates in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, weighting eight pixels around the central pixel by using the weight of logarithmic function fitting to obtain templates in eight directions, expanding and adding the templates by taking the central pixel as an origin to obtain an improved template, and adding one half of the mask of the improved template after normalization to obtain a final differential template;
wherein, the brightness adjusting function in the step 3) is specifically:
wherein x iskIs the image of the k wave band in the multispectral image, (x)k)max，(xk)minAre each xkMaximum and minimum gray values of xk' As an enhanced image, ElIs a single band local mean, EgIs a single band global mean, EsIs the global mean value of all bands, gamma is the gamma coefficient, from the global, when the global mean value E of a bandlGlobal mean value E of all bands less than p timessThen, the whole wave band is enhanced, if the global mean value E of the single wave bandlGlobal mean E of all bands greater than q timessThe band brightness value is decreased.
2. The method for enhancing the multispectral remote sensing image according to claim 1, wherein the fractional mask is a 5 x 5 mask.
3. An apparatus for enhancing a multispectral remote sensing image, the apparatus comprising:
the compensation and extraction module is used for reducing the dimension of the multispectral image by adopting weighted PCA (principal component analysis) transformation, performing weighted compensation on the original multispectral image of each wave band by utilizing average gradient and texture roughness and extracting main component components which are not related to each other;
the modifying and enhancing module is used for modifying the template coefficient by utilizing a logarithmic function to form a fractional mask for keeping the correlation of adjacent pixels, and introducing the improved template coefficient into the first principal component after the image weighted PCA transformation for enhancing the texture and edge details of the image;
the adjusting module is used for performing fractional order differential enhancement on the first principal component, performing weighted PCA inverse transformation on the principal component of the previous preset value, and adjusting the multispectral image subjected to weighted PCA inverse transformation by adopting a brightness adjusting function;
wherein the compensation and extraction module comprises:
(1) data set X ═ X with n bands1,x2,…,xn}∈RmBand xkThe method is characterized in that the method is an m-dimensional column vector, wherein k is 1,2 and … n, each sample is endowed with a weight according to different roles of each sample in the enhancing process, and the weight vector is defined as gamma, wherein gamma is { gamma ═ gamma1,γ2,…,γnAnd i.e.:
where α is the adjustment parameter, AG is the mean gradient, which means the mean value of the rate of change of the grey level, used to indicate the image sharpness, expressed as:
wherein M is the number of rows of the pixel matrix of the image, N is the number of columns of the pixel matrix of the image, Ik(i, j) are image pixel coordinates, and C is roughness, a physical quantity used to describe grain size and distribution in texture;
wherein σk 2Is the local variance of the image;
(2) averaging the data from the original data set to
Obtaining a weighted data set XWComprises the following steps:
covariance matrix
Comprises the following steps:
(3) for matrix
Performing characteristic decomposition to obtain the characteristic value lambda of the covariance matrixkAnd the corresponding feature vector phikSorting the eigenvalues in descending orderk；
Principal component cumulative variance contribution rate MdThe calculation formula is as follows:
(4) selecting the first d characteristic values lambda ═ diag [ lambda ]1,λ2,…λd]And the corresponding eigenvector phi ═ phi [ ]1,φ2,…,φd]As a basis of the subspace, the extracted principal component is used to replace the original n variables, and a projection matrix Y ═ Φ is generatedTXW；
Wherein the modification and enhancement module comprises:
and the modification submodule is used for endowing different weights to adjacent pixels according to the difference of the actual distances, fitting the weights by adopting a decreasing logarithmic function, and obtaining the weights w which are sequentially as follows:
obtaining improved templates in the x, y negative direction and the upper right diagonal direction after normalization;
similarly, weighting eight pixels around the central pixel by using the weight of logarithmic function fitting to obtain templates in eight directions, expanding and adding the templates by taking the central pixel as an origin to obtain an improved template, and adding one half of the mask of the improved template after normalization to obtain a final differential template;
the enhancement submodule is used for introducing the improved template coefficient into the first principal component after the image weighted PCA transformation, and enhancing the texture and edge details of the image;
the brightness adjustment function is specifically:
wherein x iskIs the image of the k wave band in the multispectral image, (x)k)max，(xk)minAre respectively xkMaximum and minimum gray values of xk' As an enhanced image, ElIs a single band local mean, EgIs a single band global mean, EsIs the global mean of all bands, gamma is the gamma coefficient, from allLocal, global mean value E of a bandlGlobal mean value E of all bands less than p timessThen, the whole enhancement is carried out on the wave band, if the global mean value E of the single wave bandlGlobal mean E of all bands greater than q timessThe band brightness value is decreased.
4. An apparatus for enhancing a multispectral remote sensing image, the apparatus comprising:
a processor and a memory, the memory having stored therein program instructions, the processor calling upon the program instructions stored in the memory to cause the apparatus to perform the method steps of any of claims 1-2.
5. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program comprising program instructions which, when executed by a processor, cause the processor to carry out the method steps of any of claims 1-2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110272277.XA CN112927161B (en) | 2021-03-12 | 2021-03-12 | Method and device for enhancing multispectral remote sensing image and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110272277.XA CN112927161B (en) | 2021-03-12 | 2021-03-12 | Method and device for enhancing multispectral remote sensing image and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112927161A CN112927161A (en) | 2021-06-08 |
| CN112927161B true CN112927161B (en) | 2022-07-01 |
Family
ID=76172881
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110272277.XA Active CN112927161B (en) | 2021-03-12 | 2021-03-12 | Method and device for enhancing multispectral remote sensing image and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112927161B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114283548A (en) * | 2021-12-27 | 2022-04-05 | 北京科技大学天津学院 | Fire continuous monitoring method and system for unmanned aerial vehicle |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1588447A (en) * | 2004-08-19 | 2005-03-02 | 复旦大学 | Remote sensitive image fusing method based on residual error |
| CN102521815A (en) * | 2011-11-02 | 2012-06-27 | 薛笑荣 | Fast fusion system and fast fusion method for images |
| EP3229207A2 (en) * | 2016-04-08 | 2017-10-11 | Goodrich Corporation | Warp models for registering multi-spectral imagery |
| CN109800655A (en) * | 2018-12-25 | 2019-05-24 | 核工业北京地质研究院 | The recognition methods in ETM remote sensing image SANDSTONE URANIUM DEPOSITS supply-excretion-runoff area |
| CN110163830A (en) * | 2019-04-19 | 2019-08-23 | 三峡大学 | Image interfusion method based on Riesz-Lap transformation and PCNN |
| CN110211044A (en) * | 2019-05-24 | 2019-09-06 | 北京理工大学 | Multispectral imaging method and device based on demosaicing algorithms and principal component analysis |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8866920B2 (en) * | 2008-05-20 | 2014-10-21 | Pelican Imaging Corporation | Capturing and processing of images using monolithic camera array with heterogeneous imagers |
-
2021
- 2021-03-12 CN CN202110272277.XA patent/CN112927161B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1588447A (en) * | 2004-08-19 | 2005-03-02 | 复旦大学 | Remote sensitive image fusing method based on residual error |
| CN102521815A (en) * | 2011-11-02 | 2012-06-27 | 薛笑荣 | Fast fusion system and fast fusion method for images |
| EP3229207A2 (en) * | 2016-04-08 | 2017-10-11 | Goodrich Corporation | Warp models for registering multi-spectral imagery |
| CN109800655A (en) * | 2018-12-25 | 2019-05-24 | 核工业北京地质研究院 | The recognition methods in ETM remote sensing image SANDSTONE URANIUM DEPOSITS supply-excretion-runoff area |
| CN110163830A (en) * | 2019-04-19 | 2019-08-23 | 三峡大学 | Image interfusion method based on Riesz-Lap transformation and PCNN |
| CN110211044A (en) * | 2019-05-24 | 2019-09-06 | 北京理工大学 | Multispectral imaging method and device based on demosaicing algorithms and principal component analysis |
Non-Patent Citations (5)
| Title |
|---|
| Remote Sensing Image Fusion Based on Discrete Fractional Random Transform and PCA;Qing Guo等;《34th Canadian Symposium on Remote Sensing》;20130831;1-3 * |
| 利用双差异图和PCA的SAR图像变化检测;刘陆洋 等;《计算机工程与设计》;20190731;第40卷(第7期);2002-2006 * |
| 基于PCA变换的多元遥感影像融合方法研究;黄伦春 等;《工程地球物理学报》;20140330;266-270 * |
| 基于低概率检测的高光谱异常目标检测算法研究;王玉磊等;《黑龙江大学自然科学学报》;20100625(第03期);135-140 * |
| 多光谱图像融合系统的设计与实现;白家铭;《中国优秀硕士学位论文全文数据库 (信息科技辑)》;20200715;I138-1253 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112927161A (en) | 2021-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shahdoosti et al. | Pansharpening of clustered MS and Pan images considering mixed pixels | |
| Yang et al. | An adaptive method for image dynamic range adjustment | |
| JP6012408B2 (en) | Pan-sharpening panchromatic and multispectral images using dictionaries | |
| CN111583123A (en) | Wavelet transform-based image enhancement algorithm for fusing high-frequency and low-frequency information | |
| Zhao et al. | Detail-preserving image denoising via adaptive clustering and progressive PCA thresholding | |
| CN109919859B (en) | Outdoor scene image defogging enhancement method, computing device and storage medium thereof | |
| US6049623A (en) | Directional adaptive noise reduction | |
| CN110232668B (en) | Multi-scale image enhancement method | |
| WO2016165076A1 (en) | Method and system for image enhancement | |
| JP6943251B2 (en) | Image processing equipment, image processing methods and computer-readable recording media | |
| CN106934398B (en) | Image de-noising method based on super-pixel cluster and rarefaction representation | |
| CN112927161B (en) | Method and device for enhancing multispectral remote sensing image and storage medium | |
| CN117575953B (en) | Detail enhancement method for high-resolution forestry remote sensing image | |
| CN115035011B (en) | Low-illumination image enhancement method of self-adaption RetinexNet under fusion strategy | |
| Parihar et al. | A comprehensive analysis of fusion-based image enhancement techniques | |
| Su et al. | Bayesian depth estimation from monocular natural images | |
| Kim et al. | A multidimensional histogram equalization by fitting an isotropic Gaussian mixture to a uniform distribution | |
| Wang et al. | A self-supervised deep denoiser for hyperspectral and multispectral image fusion | |
| CN106504208B (en) | High-spectrum image width destriping method based on orderly minimum value and wavelet filtering | |
| Chen et al. | Retinex low-light image enhancement network based on attention mechanism | |
| CN112784747B (en) | Multi-scale eigen decomposition method for hyperspectral remote sensing image | |
| CN112801853B (en) | Parallel acceleration method for extracting hyperspectral image features based on NAPC (non-uniform resource locator) | |
| Park et al. | Multiple transformation function estimation for image enhancement | |
| KR101287508B1 (en) | Apparatus and method for contrast enhancement using dominant brightness level and adaptive intensity transformation | |
| CN116228595A (en) | Low-illumination image enhancement method based on Retinex-Net and wavelet transformation fusion |
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 |