CN106529473B - Mixed pixel adaptive decomposition method based on multiple dimensioned window - Google Patents
Mixed pixel adaptive decomposition method based on multiple dimensioned window Download PDFInfo
- Publication number
- CN106529473B CN106529473B CN201610986431.9A CN201610986431A CN106529473B CN 106529473 B CN106529473 B CN 106529473B CN 201610986431 A CN201610986431 A CN 201610986431A CN 106529473 B CN106529473 B CN 106529473B
- Authority
- CN
- China
- Prior art keywords
- pixel
- window
- mixed pixel
- group
- mixed
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/04—Interpretation of pictures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/89—Lidar systems specially adapted for specific applications for mapping or imaging
Abstract
The mixed pixel adaptive decomposition method based on multiple dimensioned window that the present invention relates to a kind of, a count each component and its abundance in mixed pixel spatial dimension, establish initial window size；B is constructed using pixel value, abundance as datum, and group score value is the system of linear equations of unknown number；The linear dependence of separate equation in c group of equations, judges whether equation group belongs to the underdetermined system of equations；If the equation group of d building belongs to the underdetermined system of equations, the scale of window is increased into two picture element units, as new window size, repeats b, step c；E presses calculated each group score value, constructs the high spatial resolution group score value image in the mixed pixel spatial dimension.By calculating mixed pixel number of components to be decomposed on image, multiple dimensioned window is obtained, the resolving equation group of the mixed pixel is adaptively constructed, can avoid underdetermined problem, successfully mixed pixel each on image can be decomposed, obtain the higher component image of spatial resolution.
Description
Technical field
The present invention relates to the process in remote sensing digital image processing technologies in a kind of photogrammetry and remote sensing section, more particularly, to one
Kind image mixed pixel decomposition method.
Background technique
The spatial resolution of remote sensing image is one of the important feature for determining its application value.Part remote sensor due to
The restriction of technical aspect, compares earth's surface research object, and the Pixel domain resolution ratio of imaging is often lower.This makes certain pixels
Record value by a variety of atural object classifications mixing form, reduce the application value of image.Using this compared with low spatial resolution
Image research in application, for obtain mixed pixel in the other reflected value of various regions species, need to carry out the decomposition of mixed pixel.
During Decomposition of Mixed Pixels, the atural object classification in mixed pixel spatial dimension is known as component, by ground species
Occupied area ratio is known as abundance (such as attached drawing 1) not in the spatial dimension, and component is generally by existing high spatial resolution
Object classification chart or land-use map extract, and correspondingly, the abundance of each component is calculated by the spatial dimension of component combination mixed pixel
It obtains.The basic principle of Decomposition of Mixed Pixels is, according to spectral mixing model, the relationship of value and each abundance to mixed pixel into
Row calculates, and solves the value of component.Component in mixed pixel is often more than one, unknown for decomposing single mixed pixel
The number of number (component) is greater than equation (pixel) number, belong to underdetermined problem (such as attached drawing 1, shown in mixed pixel including 3 kinds of groups
Point, 1 equation can be listed).It, need to (such as attached drawing 2 be by other pixels on low spatial resolution image for resolving group score value
The group score value of mixed pixel 1 is calculated, 3 equations need to be listed by surrounding mixed pixel 2 and mixed pixel 3, be equal to mixing
The number of components 3 of pixel 1), so that not linearly related equation number is more than or equal to unknown number number, constructs suitable fixed or over-determined systems,
It is acquired by the methods of least square.
With respect to the space variance of pixel, the prior art is to be adopted centered on mixed pixel to be decomposed using window
Collecting the pixel of its neighborhood, (such as attached drawing 3, window is using 3 pixels as scale, centered on mixed pixel 1, acquires the picture of its neighborhood
Member), to construct resolving equation group.Mathematical relationship between window size and number of components are as follows:S is window size, and L is
Number of components in mixed pixel.One of its problem is that using the scale of window (is picture shared by cross/longitudinal direction on image
First number) be it is fixed, pixel is often acquired using the window of some scale on a width image, is not avoided that equation group owes fixed
The generation of problem.For example, using scale for 3 window, the pixel number that can be acquired is up to 9, when mixing picture to be decomposed
When number of components is greater than 9 in member, i.e., (such as attached drawing 4 acquires pixel, mixing to be decomposed with the window that scale is 3 to generation underdetermined problem
Pixel number of components is 16, i.e., equation number is 9, and unknown number is 16 in equation, and equation group belongs to underdetermined problem)；And when using ruler
When the window that degree is 5, the pixel that can be acquired is up to 25, when number of components is greater than 25 in mixed pixel to be decomposed, that is, sends out
Raw underdetermined problem；And so on.The two of its problem are, using the window of fixed size, even if the pixel number of its acquisition is greater than
Number of components in mixed pixel to be decomposed, still may cause underdetermined problem.For example, mixed pixel number of components is 5, window size
It is 3, the pixel number of window acquisition is 9, there are the identical pixel of 6 abundance in the pixel set of acquisition, therefore pixel equation group
Order is 4, is less than number of components, causes underdetermined problem (as shown in Fig. 5).Causing the basic reason of the above problem is, the prior art
The fixed size window form of use lacks flexibility when constructing the resolving equation group of mixed pixel, it is more not to adapt to complexity
The earth's surface situation of change.
Summary of the invention
It is an object of the invention in view of the above shortcomings of the prior art, provide a kind of mixing picture based on multiple dimensioned window
First adaptive decomposition method.
The technical scheme is that being calculated just when being decomposed to mixed pixel according to the number of components of pixel to be decomposed
Beginning window size (press formula:S takes the odd number upper limit of the condition of satisfaction), window, the window are constructed by the initial gauges
Its neighborhood pixel is acquired centered on mixed pixel, if the equation group as constructed by these pixels and the mixed pixel at center belongs to
Suitable fixed or overdetermined problem then completes the building for resolving equation group；If belonging to underdetermined problem, parent window scale increases by two pixels
Pixel between the window of new scale and archeus window is added in former resolving equation group, judgement by unit as new scale
Whether the resolving equation group newly constructed belongs to suitable fixed or overdetermined problem, if being still underdetermined problem, window size is continued growing
Two picture element units, until the resolving equation group of its building belongs to suitable fixed or overdetermined problem；Finally, utilizing least square method solution
The equation group for calculating building, calculates each group score value.
The purpose of the present invention is what is be achieved through the following technical solutions:
A kind of mixed pixel adaptive decomposition method based on multiple dimensioned window, comprising the following steps:
A, on the remote sensing image of low spatial resolution, each pixel is traversed, to each pixel referring in the mixed pixel
Component number, establish initial window size, using the mixed pixel as the center of window, acquire the mixing picture by home window
Pixel in first neighborhood；
B, to pixel collected and mixed pixel, according to spectral mixing model fitting mixed pixel value and its component, rich
The mathematical relationship of degree is constructed using group score value as the equation group of unknown number；
C, judge whether the equation group of building belongs to the underdetermined system of equations, handled according to judging result；
If d, the equation group constructed belongs to the underdetermined system of equations, the scale of window is increased into two picture element units, as new
Window size acquires the pixel between new window and parent window, is added in original equation group, repeats step c processing；If building
Equation group is not belonging to the underdetermined system of equations, then uses least square method, resolves equation group；
E, calculated each group score value is pressed, according to spatial distribution of each component in mixed pixel, constructs the mixing picture
High spatial resolution component image in first spatial dimension.
Establish initial window size described in step a, the initial gauges of window be by number of components in mixed pixel come
It is calculated, S, which should take, meets formulaMinimum odd number
In formula: S is window size, and L is the number of components in mixed pixel.
Mixed pixel described in step b, by the mixed pixel of home window collected neighborhood pixel and its center, foundation
Spectral mixing model is constructed using group score value as the equation group of unknown number.
Step c, judge whether the equation group of building belongs to the underdetermined system of equations described in d, after building resolves equation group, sentence
Breaking, whether it belongs to the underdetermined system of equations, when belonging to the underdetermined system of equations, by increasing by two pixel lists on the scale of parent window
Position, expands the spatial dimension of window, to form new window, the equation of the pixel between parent window and new window is added to original
Resolve equation group；The acquisition mode of mixed pixel neighborhood pixel is to pass through the mixed pixel and the number of components of peripheral neighborhood pixel
Situation constructs the window of one or more scale, resolves in equation group to acquire pixel and constantly be added to.
The technical characteristic that the present invention is shared with the prior art is:
It is the form using window, the peripheral neighborhood pixel centered on mixed pixel to be decomposed is acquired,
The mathematical relationship of each component in mixed pixel and its Spatial ambience, abundance is fitted according to spectral mixing model, by collected
Pixel and the building of the mixed pixel at center resolve equation group and calculate the value of each component by means such as least square methods, pass through
High-resolution group of score value image is constructed in distribution of the component in mixed pixel.
Difference with the prior art of the present invention is characterized in:
Initial window size is calculated by number of components in mixed pixel, on whole picture image, each mixed pixel is because of it
Number of components is different, and the initial gauges of window are different；
When constructing resolving equation group by the collected pixel of window and center mixed pixel, it need to judge that equation group is
It is no to belong to suitable fixed or over-determined systems；
When the resolving equation group as constructed by window collected pixel is not belonging to suitable fixed or overdetermined problem, by former window
Increase by two picture element units on the scale of mouth, expands the spatial dimension of window, to form new window, by parent window and new window
Between pixel equation be added to it is former resolve in equation group, judge whether new equation group belongs to suitable fixed or over-determined systems；
Acquisition to the neighborhood pixel of mixed pixel is the component situation and peripheral neighborhood pixel by the mixed pixel
Situation constructs the window of one or more scale, to acquire pixel and constantly be added in initial resolving equation group.
The utility model has the advantages that multiple dimensioned window is obtained by calculating mixed pixel number of components to be decomposed on image, it is adaptive
It constructs the resolving equation group of the mixed pixel with answering, can avoid the generation of underdetermined problem, it can be successfully to mixing each on image
Pixel is decomposed, and the higher component image of spatial resolution is obtained.
Detailed description of the invention
Mixed pixel and its component in 1 remote sensing image of attached drawing
Three mixed pixels that component is identical, abundance is different in 2 remote sensing image of attached drawing
Attached drawing 3 acquires pixel with window form on image
Attached drawing 4 acquires pixel by window center of the mixed pixel of various ingredients
The pixel of four kinds of different components of 5 window of attached drawing acquisition
The multiple dimensioned window of attached drawing 6 acquires pixel schematic diagram
A kind of mixed pixel adaptive decomposition method flow diagram based on multiple dimensioned window of attached drawing 7
Attached drawing 8 is using multiple dimensioned window to 1 adaptive decomposition instance graph of mixed pixel
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with drawings and examples pair
The present invention is described in further detail.
Firstly, collecting the remote sensing image of low spatial resolution and the high spatial resolution earth's surface of areal, same range
Classification chart, using this two images as given data；
Then, to each wave band of the remote sensing image of low spatial resolution, traversal each pixel thereon, to each pixel into
The following processing of row:
The first step, count mixed pixel spatial dimension in each component and its abundance (component is in Surface classification figure
Classification, abundance are the area ratio that component accounts for the pixel), referring to the component number in the mixed pixel, establish initial window
Mouthful scale (formula such as:S is window size, and L is the number of components in mixed pixel, and the minimum for making S meet condition is odd
Number), and using the mixed pixel as the center of window, the pixel in the mixed pixel neighborhood is acquired by home window；Such as 8 institute of attached drawing
Show, to decompose mixed pixel 1, statistics obtains, the component in spatial dimension has: five kinds of a, b, c, d, e, each abundance is successively are as follows:
1/3,2/9,2/9,1/9,1/9；By formula it is found that its home window is 3, and centered on mixed pixel 1, establish initial window
Mouthful, (2,2,3,4,4,4,4,4 pixel in figure, wherein the same symbol indicates that component, abundance are all identical for pixel in acquisition window
Pixel)；
Second step, using linear spectral mixture model, is fitted on the wave band and mixes to pixel collected and mixed pixel
The equation of the value of pixel and wherein each component, abundance is constructed using mixed pixel value, abundance as datum, and group score value is unknown number
System of linear equations；The equation group constructed with the home window of mixed pixel 1 is following, and (V indicates value, and f indicates abundance, such as VaIt indicates
The value of component a, faIndicate the abundance of component a, V1Indicate the value of mixed pixel 1):
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V1
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V2
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V2
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V3
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Third step, the linear dependence of the separate equation in group of equations remove linear correlation side in group of equations
Equation number after journey belongs to underdetermined problem if party's number of passes is less than the number of components of mixed pixel to be decomposed；It can by the first step
Know, 2,4 pixels have that component is identical as abundance in the pixel of acquisition, therefore there are linear dependences for its equation, remove line
Property relevant equation after, equation number 4 is less than number of components 5, is underdetermined problem；
The scale of window is increased by two picture element units, as new if the equation group of building belongs to underdetermined problem by the 4th step
Window size, acquire the pixel between new window and parent window, be added in original equation group, repeat the processing of third step；If
The equation group of building is not belonging to underdetermined problem, then uses least square method, resolves equation group；As shown in Fig. 8, in former window
On the basis of mouth scale 3, increase by two picture element units, construct the window that scale is 5, and centered on mixed pixel 1, acquisition is new
Pixel (totally 16) between window and parent window, after removing linearly related equation, equation group is not belonging to underdetermined problem,
Equation group is following (equation for the same composition abundance pixel that the point number expression of null is omitted in equation group):
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V1
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V2
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V3
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V4
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V5
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V6
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V7
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V8
Va·fa+Vb·fb+Vc·fc+Vd·fd+Ve·fe=V9
Using least square method, equation group is calculated, can obtain a group score value Va、Vb、Vc、Vd、Ve。
It is mixed to construct this according to spatial distribution of each component in mixed pixel by calculated each group score value for 5th step
Close the high spatial resolution group score value image within the scope of Pixel domain.In attached drawing 8, in conjunction with a, b, c, d, e inside mixed pixel 1
The spatial distribution of component, using each group score value calculated as the high spatial resolution value within the scope of correlation space.
Finally, being organized integral group subrane by all high spatial resolution group score values being calculated, being made of wave band
Image.
Claims (1)
1. a kind of mixed pixel adaptive decomposition method based on multiple dimensioned window, which comprises the following steps:
A, on the remote sensing image of low spatial resolution, each pixel is traversed, to each mixed pixel referring in the mixed pixel
Component number, establish initial window size, using the mixed pixel as the center of window, acquire the mixing picture by home window
Pixel in first neighborhood；
B, to pixel collected and mixed pixel, according to spectral mixing model fitting mixed pixel value and its component, abundance
Mathematical relationship is constructed using group score value as the equation group of unknown number；
C, judge whether the equation group of building belongs to the underdetermined system of equations, handled according to judging result；
If d, the equation group constructed belongs to the underdetermined system of equations, the scale of window is increased into two picture element units, as new window
Scale acquires the pixel between new window and parent window, is added in original equation group, repeats step c processing；If the equation of building
Group is not belonging to the underdetermined system of equations, then uses least square method, resolves equation group；
E, calculated each group score value is pressed, according to spatial distribution of each component in mixed pixel, constructs mixed pixel sky
Between high spatial resolution component image in range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610986431.9A CN106529473B (en) | 2016-11-10 | 2016-11-10 | Mixed pixel adaptive decomposition method based on multiple dimensioned window |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610986431.9A CN106529473B (en) | 2016-11-10 | 2016-11-10 | Mixed pixel adaptive decomposition method based on multiple dimensioned window |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106529473A CN106529473A (en) | 2017-03-22 |
| CN106529473B true CN106529473B (en) | 2019-12-03 |
Family
ID=58350560
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610986431.9A Expired - Fee Related CN106529473B (en) | 2016-11-10 | 2016-11-10 | Mixed pixel adaptive decomposition method based on multiple dimensioned window |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106529473B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2282544A2 (en) * | 2009-08-03 | 2011-02-09 | AIT Austrian Institute of Technology GmbH | Method and device for compressing recorded image data |
| CN104809691A (en) * | 2015-05-05 | 2015-07-29 | 李云梅 | Image fusion method based on sliding window mixed pixel decomposition |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102608592B (en) * | 2012-04-05 | 2013-06-12 | 吉林大学 | Snow passive microwave mixed pixel decomposition method based on classified information of five types of ground features |
| CN102982208B (en) * | 2012-11-30 | 2015-02-11 | 电子科技大学 | Dynamic reliability model updating method based on Bayes factor optimization |
| CN105279523B (en) * | 2015-10-22 | 2018-07-17 | 中国科学院遥感与数字地球研究所 | A kind of semisupervised classification method of combination Decomposition of Mixed Pixels and Active Learning |
| CN105488805B (en) * | 2015-12-15 | 2017-11-17 | 吉林大学 | Multifrequency dual polarization forest land accumulated snow passive microwave mixed pixel decomposition method |
-
2016
- 2016-11-10 CN CN201610986431.9A patent/CN106529473B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2282544A2 (en) * | 2009-08-03 | 2011-02-09 | AIT Austrian Institute of Technology GmbH | Method and device for compressing recorded image data |
| CN104809691A (en) * | 2015-05-05 | 2015-07-29 | 李云梅 | Image fusion method based on sliding window mixed pixel decomposition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106529473A (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Long et al. | A box-counting method with adaptable box height for measuring the fractal feature of images | |
| CN101562675B (en) | No-reference image quality evaluation method based on Contourlet transform | |
| CN104123705B (en) | A kind of super-resolution rebuilding picture quality Contourlet territory evaluation methodology | |
| CN103177431A (en) | Method of spatial-temporal fusion for multi-source remote sensing data | |
| CN102036098B (en) | Full-reference type image quality evaluation method based on visual information amount difference | |
| CN103354617B (en) | Boundary strength compressing image quality objective evaluation method based on DCT domain | |
| CN102855616B (en) | Image fusion method based on multi-scale dictionary learning | |
| CN111008651B (en) | Image reproduction detection method based on multi-feature fusion | |
| CN103871058A (en) | Compressed sampling matrix decomposition-based infrared small target detection method | |
| CN103247047A (en) | Image edge detection method based on fractional order partial differential | |
| CN100417191C (en) | Method of reducing noise for combined images | |
| CN103337073B (en) | A kind of two dimensional image threshold segmentation method based on three-dimensional entropy | |
| CN108846800A (en) | A kind of non-reference picture quality appraisement method of image super-resolution rebuilding | |
| CN1251145C (en) | Pyramid image merging method being integrated with edge and texture information | |
| CN102609914A (en) | Signal-correlated noise estimating method for image sensor | |
| CN101183460A (en) | Color picture background clutter quantizing method | |
| CN106127261A (en) | A kind of fast multiresolution gray level image template matching method | |
| CN106529473B (en) | Mixed pixel adaptive decomposition method based on multiple dimensioned window | |
| CN104463822A (en) | Multi-focus image fusing method and device based on multi-scale overall filtering | |
| CN110060220A (en) | Based on the image de-noising method and system for improving BM3D algorithm | |
| CN102547363A (en) | No-reference image quality evaluating method on basis of contourlet transform domain image energy features | |
| CN103914807A (en) | Non-locality image super-resolution method and system for zoom scale compensation | |
| Jee et al. | Sensitivity and resolution improvement in RGBW color filter array sensor | |
| CN109447952B (en) | Semi-reference image quality evaluation method based on Gabor differential box weighting dimension | |
| CN107203991A (en) | A kind of half reference image quality appraisement method based on spectrum residual error |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20191203 Termination date: 20201110 |