CN113094651B - Hyperspectral imaging parameter optimization design method based on rough set - Google Patents
Hyperspectral imaging parameter optimization design method based on rough set Download PDFInfo
- Publication number
- CN113094651B CN113094651B CN202011366930.0A CN202011366930A CN113094651B CN 113094651 B CN113094651 B CN 113094651B CN 202011366930 A CN202011366930 A CN 202011366930A CN 113094651 B CN113094651 B CN 113094651B
- Authority
- CN
- China
- Prior art keywords
- imaging
- parameter
- data table
- parameters
- influence
- 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 98
- 238000000701 chemical imaging Methods 0.000 title claims abstract description 70
- 238000013461 design Methods 0.000 title claims abstract description 25
- 238000005457 optimization Methods 0.000 title claims abstract description 25
- 238000003384 imaging method Methods 0.000 claims abstract description 111
- 229910052500 inorganic mineral Inorganic materials 0.000 claims abstract description 62
- 239000011707 mineral Substances 0.000 claims abstract description 62
- 230000008569 process Effects 0.000 claims abstract description 46
- 238000005065 mining Methods 0.000 claims abstract description 14
- 238000013507 mapping Methods 0.000 claims abstract description 11
- 238000012417 linear regression Methods 0.000 claims abstract description 6
- 238000011156 evaluation Methods 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 9
- 230000014509 gene expression Effects 0.000 claims description 7
- 238000004445 quantitative analysis Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000000638 solvent extraction Methods 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000000611 regression analysis Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 4
- 230000009471 action Effects 0.000 claims description 3
- 238000010238 partial least squares regression Methods 0.000 claims description 3
- 230000005855 radiation Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 description 6
- 238000011160 research Methods 0.000 description 5
- 238000002310 reflectometry Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011158 quantitative evaluation Methods 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005316 response function Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Operations Research (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Spectroscopy & Molecular Physics (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biochemistry (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Complex Calculations (AREA)
Abstract
A hyperspectral imaging parameter optimization design method based on a rough set comprises the following steps: (1) Analyzing a hyperspectral imaging process, and determining main influencing parameters in the imaging process; (2) Establishing an imaging parameter data table, and dividing imaging key parameter values in the data table by utilizing an entropy-based discretization method; (3) Establishing a discretization imaging parameter data table according to the value interval divided in the step (2), mining the imaging data table by adopting association rules based on Boolean attribute mapping, and establishing an influence constraint relation between key influence parameters in the hyperspectral imaging process and mineral identification capacity; (4) Calculating the attribute importance of the imaging key parameters to the mineral identification capability based on the discretization data table established in the step (3) and the rough set and the knowledge granularity; (5) Establishing a multiple linear regression model based on a partial least square method, and establishing a quantitative equation between imaging parameters and mineral identification capacity; (6) And (3) optimizing and solving the quantitative equation in the step (5) according to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity, so as to obtain the hyperspectral imaging parameter combination with the best mineral identification capacity.
Description
Technical Field
The invention relates to a hyperspectral imaging parameter optimization design method based on a rough set, belongs to the field of hyperspectral load optimization design and imaging performance prediction, and is suitable for hyperspectral imaging performance evaluation and parameter optimization design research facing mineral identification application requirements.
Background
The hyperspectral resolution remote sensing has the characteristics of high spectral resolution and 'map in one', is widely applied to various fields, can accurately acquire detailed spectral information of different ground objects and components, and realizes detection and analysis of complex ground objects and components. Therefore, whether the obtained hyperspectral data can meet the application requirements of the user department is a key problem of the related load development department at present. However, the hyperspectral imaging process is complex, the data acquisition process is affected by a plurality of factors, and in the actual imaging process, parameters such as sensor performance of a hyperspectral imager, response wave bands of the hyperspectral imager, atmospheric conditions in the transmission process, spatial resolution, spectral response functions and the like can directly influence the data quality of the hyperspectral satellite. Therefore, in practical applications, the user department is more prone to find the "optimal value" of these parameters to achieve the best application effect to guide the design and development of the load. However, the relationships between these influencing parameters are not completely independent, and there is some coupling and constraint relationship between them, so that all parameters cannot take optimal values at the same time. The current research has not perfected the establishment of direct connection between load index and application capability, and has not considered the constraint relation between influencing parameters. Therefore, determining the constraint influence relation among key influence parameters in the hyperspectral imaging process and analyzing the influence of the constraint influence relation on the application capability has important significance for the optimization design of hyperspectral load departments. In order to optimize and improve the instrument and make the data more widely used, quantitative evaluation needs to be performed on the data acquired by the remote sensing system, and the current method for evaluating the application capability of the remote sensing image data can be generally divided into: empirical analysis, imaging-based simulation modeling, analytical modeling. The evaluation method based on the empirical analysis method is characterized in that influences of different imaging parameters on the application capacity of the data are analyzed according to the actually acquired data, the influences are parameterized and represented by a statistical method, and the evaluation method is a subjective image interpretation method, and the evaluation method is simple in calculation, is not combined with a load principle and imaging conditions, and cannot essentially reveal the influence rule of the imaging process parameters on the application of the data. The imaging simulation model researches the influence of each link on imaging quality through the simulation of the whole imaging process, and is an application capability evaluation method based on imaging link simulation. The image quality evaluation method based on the analysis model is a parameterized imaging performance evaluation model, and the method has the characteristics of accuracy, flexibility and easy calculation, but lacks consideration of the mutual influence relationship and restriction relationship between key load parameters in the remote sensing physical imaging process, and is difficult to realize the optimization design of guiding load parameters. In summary, in order to guide the optimal design of the load parameters and predict the load application capability, the research in this aspect has become a research hotspot in the hyperspectral remote sensing field, but no perfect imaging performance evaluation standard, theoretical model and technical system for application and combination of load characteristics have been established at present, and comprehensive consideration among all the influencing parameters is less and not combined with the data processing process.
The Rough Set theory is a mathematical tool proposed by Pawlak teaching in 1982 that can quantitatively analyze and process imprecise, inconsistent, incomplete information and knowledge. Because the hyperspectral imaging process has the condition that part of attributes are unknown in an imaging link, the problem can be regarded as an incomplete information system when the quantitative relation between the influence parameters and the data application capacity is researched. The rough set theory can process incomplete data without relying on priori information, and discover the internal relation of each attribute in the incomplete information system, so that the rough set theory is introduced to solve the uncertain relation between imaging parameters and application capability evaluation indexes.
Disclosure of Invention
The invention aims to provide a hyperspectral imaging parameter optimization design method based on a rough set, which comprehensively considers the problems of constraint relation among parameters and information imperfection of an imaging system when optimally designing hyperspectral imaging parameters.
The technical scheme of the invention is as follows: a constraint association relation between imaging parameters and application capacity is established through association rule mining, the influence of each key parameter on mineral identification capacity is calculated based on a rough set and knowledge granularity method, a quantitative evaluation equation of mineral identification capacity evaluation indexes and imaging key parameters is established through multiple regression analysis, and parameter combination of optimal application effects is optimized and solved by combining the parameter constraint relation and attribute importance degree thereof.
The invention relates to a hyperspectral imaging parameter optimization design method based on a rough set, which comprises the following steps:
(1) Analyzing a hyperspectral imaging process, and determining main influencing parameters in the imaging process;
(2) Establishing an imaging parameter data table, and dividing imaging key parameter values in the data table by utilizing an entropy-based discretization method;
(3) Establishing a discretization imaging parameter data table according to the value interval divided in the step (2), mining the imaging data table by adopting association rules based on Boolean attribute mapping, and establishing an influence constraint relation between key influence parameters in the hyperspectral imaging process and mineral identification capacity;
(4) Calculating the attribute importance of the imaging key parameters to the mineral identification capability based on the discretization data table established in the step (3) and the rough set and the knowledge granularity;
(5) Establishing a multiple linear regression model based on a partial least square method, and establishing a quantitative equation between imaging parameters and mineral identification capability evaluation indexes;
(6) And (3) optimizing and solving the quantitative equation in the step (5) according to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity, so as to obtain the hyperspectral imaging parameter combination with the best mineral identification capacity.
The hyperspectral imaging process analysis in the step (1) is carried out, and main influencing parameters in the imaging process are determined: and establishing a scene model, an atmospheric radiation transmission model and a detector model in the hyperspectral imaging process, analyzing the hyperspectral imaging process, and determining key influence parameters influencing imaging performance in the imaging process mainly comprising ground sampling distance, spectral resolution, signal-to-noise ratio and modulation transfer function.
The imaging parameter data table is established in the step (2), and imaging key parameter values in the data table are divided by utilizing an entropy-based discretization method: establishing an imaging parameter data table, adopting an entropy-based attribute value dividing method, selecting a break point which enables the overall entropy of the imaging parameter data table to be minimum as an interval dividing break point, determining the number of discrete intervals of each parameter attribute by utilizing a minimum description length principle, and discretizing the attribute values of imaging key influence parameters;
the process is as follows:
first, the imaging parameters c are set for each piece of imaging data d 1 The values are arranged in order from small to large, and the obtained sequence is expressed as follows: delta 1 ,δ 2 ,…,δ n The method comprises the steps of carrying out a first treatment on the surface of the Then, record in imaging parameter c 1 Each potential boundary in the range of values is:
wherein ,Li Dividing the candidate interval of the imaging parameter into points, wherein the dividing point L i Sample set D of each imaging parameter i The division into two subsets is noted as:
D 1i ＝{d∈D|D c1 ≤L i }
D 2i ＝{d∈D|D c1 ＞L i }
when selecting L i As a partitioning point, it separates the sample set of the data table into two subsets, the information gain of the imaging data table is expressed as:
wherein ,Di Is a sample set, is (D i ) Information entropy of the sample set;
suppose D in a sample set 1i The number of categories is m 1 ，D 2i The number of categories is m 2 Then when:
terminating partitioning of data at time, wherein Δ=log 2 (3 m-2 -[Ent(D)-m 1 Ent(D 1i )-m 2 Ent(D 2i )]) And obtaining a discretized data table.
Wherein step (3) is according to step(2) Establishing a discretization imaging parameter data table in the divided value interval, mining the imaging data table by adopting association rules based on Boolean attribute mapping, and establishing an influence constraint relation between key influence parameters in a hyperspectral imaging process and mineral identification capacity: performing Boolean attribute mapping on the discretized data table, performing association rule mining on the data table by adopting an Apriori association rule mining algorithm, and setting the set of all attribute values of the hyperspectral imaging database to be I= { I 1 ,i 2 ,…,i m -consisting of m items, called item sets, i k The method is characterized in that the method is represented as a certain parameter attribute value in a hyperspectral imaging database, the parameter attribute value is called an item, data of all attribute values under different imaging parameter combinations are recorded as D, T is recorded as each group of imaging parameters in the D, each group of T is a subset of an item set I, rules meeting requirements are selected through the support degree, the lifting degree and the confidence degree of a calculation rule, and the mined association rule is represented as:
wherein ,
x is a front item of an association rule, namely a key influence parameter, Y is an association rule rear item, and is an evaluation index of mineral identification capacity, and an influence constraint relation between the key influence parameter and the mineral identification capacity in a hyperspectral imaging process is established through the obtained association rule.
The step (4) is based on the discretized data table established in the step (3), and the attribute importance of the imaging key parameters on the mineral identification capacity is calculated based on the rough set and the knowledge granularity: the hyperspectral imaging parameter data sheet is T= (U, C, D, V, f), U= { x 1 ,x 2 ,…,x n The method comprises the steps that (1) an object set in a data table is obtained, A is a non-empty limited set of all parameter attributes in the data table, A=C U.D, C is a key influence parameter in an imaging system, D is an evaluation index of mineral identification capability, V is a set of parameter values of the data table, and an information function is obtained: f, u×a→v, x=u/d= { X 1 ,X 2 ,…,X n And the equal division of the domain U under the decision attribute D is shown, U/C is the equal division of the conditional characteristics on the domain U, and then X=U/D= { X 1 ,X 2 ,…,X n Knowledge feature resolution on } is:
KCDis(C)＝ρ C (X)·Dis(C)
wherein ,ρR (X) is the accuracy of each group of data X in its equivalence relation (knowledge) R, dis (C) is the resolution of the knowledge, and in the hyperspectral imaging parameter data table, for any key influencing parameter C E C, its characteristic resolution is:
KCDis(c)＝KCDis(C)-KCDis(C-{c})
therefore, the attribute importance of any key influence parameter on the mineral identification capacity is calculated:
and establishing the influence degree action relation of each influence parameter on the mineral identification capacity based on the attribute importance.
The method comprises the following steps of (5) establishing a multiple linear regression model based on a partial least square method, and establishing a quantitative equation between imaging parameters and evaluation indexes of mineral identification capacity: establishing quantitative analysis expression of mineral identification type and hyperspectral imaging performance key influence parameters by adopting a partial least square multiple regression analysis method, wherein in a quantitative equation, the evaluation index of mineral identification capacity is a random variable y and a common variable x 1 ,x 2 ,…,x p As key influence parameters, y and p (p is more than or equal to 2) common variables x are established through partial least squares regression analysis 1 ,x 2 ,…,x p Is that
wherein ,
for regression equationEpsilon is a random variable.
According to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity, the quantitative equation in the step (5) is optimized and solved to obtain the hyperspectral imaging parameter combination with the optimal mineral identification capacity: taking the quantitative analysis expression obtained in the step (5) as an objective function, and establishing the following optimization model by combining the constraint relation of the step (3) and the step (4):
min f(x),x∈R n
s.t.c i (x)＝0,i∈E＝{1,2,…,l},
c i (x)≤0,i∈I＝{l+1,l+2,…,l+m}.
based on the optimality condition, solving the constraint optimization problem to obtain an optimal parameter combination, and realizing the optimal design of hyperspectral imaging parameters.
Compared with the prior art, the invention has the advantages that: and establishing direct connection between key indexes such as load parameters and application capacity, and comprehensively considering incomplete characteristics of an imaging process and constraint relations of all parameters. The method utilizes an association rule mining algorithm and an attribute importance calculation method based on rough set and knowledge granularity to effectively consider the constraint association relation of imaging parameters, analyzes the influence of different imaging parameters on the application capacity of the data under an incomplete information system, establishes direct connection between the imaging parameters and the application capacity, and realizes the solution of the optimal combination of hyperspectral imaging parameters. It has the following advantages: (1) And an association rule mining algorithm is effectively applied to mine association influence constraint relations of all imaging parameters on mineral identification capacity in the imaging process. (2) And calculating the dependence degree of the evaluation index of the mineral identification capability on the key influence parameter by utilizing the rough set theory, and objectively calculating the relation between the imaging parameter and the evaluation index of the mineral identification capability, thereby determining the influence degree and the influence degree of the imaging parameter on the mineral identification capability. (3) And establishing quantitative analysis expression between the mineral identification type and the key influence parameters by utilizing multiple regression analysis, and establishing direct connection between the application capacity and the load parameters.
Detailed Description
In order to better explain the hyperspectral imaging parameter optimization design method based on the rough set, hyperspectral imaging parameter optimization design is performed by taking Hymap visible light to short wave infrared (VNIR-SWIR) hyperspectral reflectivity data as source data, changing imaging parameters of the hyperspectral imaging parameter data to perform data simulation, performing mineral mapping on simulation data by adopting a method based on mixed modulation matched filtering, taking a mineral identification type as an evaluation index of mineral identification capability, and performing hyperspectral imaging parameter optimization design based on the hyperspectral imaging parameter. The invention relates to a hyperspectral imaging parameter optimization design method based on a rough set theory, which comprises the following specific implementation steps:
(1) Hyperspectral imaging process analysis, determining main influencing parameters in the imaging process: and establishing a scene model, an atmospheric radiation transmission model and a detector model in the hyperspectral imaging process, analyzing the hyperspectral imaging process, and determining key influence parameters influencing imaging performance in the imaging process mainly comprising ground sampling distance, spectral resolution, signal-to-noise ratio and modulation transfer function.
(2) Establishing an imaging parameter data table, and dividing imaging key parameter values in the data table by using an entropy-based discretization method: based on hyperspectral reflectivity data from Hymap visible light to short wave infrared (VNIR-SWIR), changing imaging parameters of the hyperspectral reflectivity data to perform data simulation, and performing mineral mapping on the hyperspectral reflectivity data; establishing an imaging parameter data table, discretizing key image parameter attributes of four hyperspectral imaging systems by adopting a continuous attribute discretization method based on entropy values, and obtaining a discretized imaging data table by calculating the average value of the maximum value and the minimum value of two discrete intervals when dividing the attribute value range according to the discretization result;
the process is as follows:
first, the imaging parameters c are set for each piece of imaging data d 1 The values are arranged in order from small to large, and the obtained sequence is expressed as follows: delta 1 ,δ 2 ,…,δ n The method comprises the steps of carrying out a first treatment on the surface of the Then, record in imaging parameter c 1 Each potential boundary in the range of values is:
wherein ,Li Dividing the candidate interval of the imaging parameter into points, wherein the dividing point L i Sample set D of each imaging parameter i The division into two subsets is noted as:
when selecting L i As a partitioning point, it separates the sample set of the data table into two subsets, the information gain of the imaging data table is expressed as:
wherein ,Di Is a sample set, is (D i ) Information entropy of the sample set;
suppose D in a sample set 1i The number of categories is m 1 ，D 2i The number of categories is m 2 Then when:
terminating partitioning of data at time, wherein Δ=log 2 (3 m-2 -[Ent(D)-m 1 Ent(D 1i )-m 2 Ent(D 2i )]) And obtaining a discretized data table.
(3) Establishing a discretization imaging parameter data table according to the value interval divided in the step (2), mining the imaging data table by adopting association rules based on Boolean attribute mapping, and establishing key influence parameters in the hyperspectral imaging process and oresInfluence constraint relation of object recognition capability: performing Boolean attribute mapping on the discretized data table, performing association rule mining on the data table by adopting an Apriori association rule mining algorithm, and setting the set of all attribute values of the hyperspectral imaging database to be I= { I 1 ,i 2 ,…,i m -consisting of m items, called item sets, i k The method is characterized in that the method is expressed as a certain parameter attribute value in a hyperspectral imaging database, the item is called, data of all attribute values under different imaging parameter combinations are recorded as D, T is recorded as each group of imaging parameters in the D, each group T is a subset of an item set I, the support, the promotion degree and the confidence degree of a calculation rule are expressed as:
wherein ,
x is the front item of the association rule, namely the key influence parameter, Y is the rear item of the association rule, and is the mineral identification type, wherein the minimum value support degree is 4%, the minimum confidence degree is 80%, 20 association rules are obtained through excavation, and based on the rules, the constraint relation between the key influence parameters and the mineral identification capacity in the four hyperspectral imaging processes is established.
(4) Calculating the attribute importance of the imaging key parameters to the mineral identification capability based on the discretization data table established in the step (3) and the rough set and the knowledge granularity: the hyperspectral imaging parameter data sheet is T= (U, C, D, V, f), U= { x 1 ,x 2 ,…,x n The method comprises the steps that } is a set of objects in a data table, A is a non-empty limited set of all parameter attributes in the data table, A=C U.D, C is a key influence parameter in an imaging system, D is a mineral identification type, V is a set of parameter values of the data table, and an information function is that: f, u×a→v, x=u/d= { X 1 ,X 2 ,…,X n And the equal division of the domain U under the decision attribute D is shown, U/C is the equal division of the conditional characteristics on the domain U, and then X=U/D= { X 1 ,X 2 ,…,X n Knowledge feature resolution on } is:
KCDis(C)＝ρ C (X)Dis(C)
wherein ,ρR (X) is the accuracy of each group of data X in its equivalence relation (knowledge) R, dis (C) is the resolution of the knowledge, and in the hyperspectral imaging parameter data table, for any key influencing parameter C E C, the feature resolution is:
KCDis(c)＝KCDis(C)-KCDis(C-{c})
therefore, the attribute importance of any key influence parameter on the mineral identification capacity is calculated:
and establishing the influence degree action relation of each influence parameter on the mineral identification capacity based on the attribute importance.
(5) Establishing a multiple linear regression model based on a partial least square method, and establishing a quantitative equation between imaging parameters and mineral identification types: establishing quantitative analysis expressions of key influence parameters of mineral identification types and hyperspectral imaging performance by adopting a partial least square multiple regression analysis method; in the quantitative equation, the mineral identification type is a random variable y and a common variable x 1 ,x 2 ,…,x p As key influence parameters, y and p (p is more than or equal to 2) common variables x are established through partial least squares regression analysis 1 ,x 2 ,…,x p Is that
wherein ,
epsilon is a random variable, which is the regression coefficient of the regression equation.
(6) According to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity, optimizing and solving the quantitative equation in the step (5), and obtaining the hyperspectral imaging parameter combination with the best mineral identification capacity: taking the quantitative analysis expression obtained in the step (5) as an objective function, and establishing the following optimization model by combining the constraint relation of the step (3) and the step (4):
min f(x),x∈R n
s.t.c i (x)＝0,i∈E＝{1,2,…,l},
c i (x)≤0,i∈I＝{l+1,l+2,…,l+m}.
based on the optimality condition, solving the constraint optimization problem to obtain an optimal parameter combination, and realizing the optimal design of hyperspectral imaging parameters.
Claims (6)
1. A hyperspectral imaging parameter optimization design method based on a rough set is characterized by comprising the following steps of: it comprises the following steps:
(1) Analyzing a hyperspectral imaging process, and determining main influencing parameters in the imaging process;
(2) Establishing an imaging parameter data table, and dividing imaging key parameter values in the data table by utilizing an entropy-based discretization method;
(3) Establishing a discretization imaging parameter data table according to the value interval divided in the step (2), mining the imaging data table by adopting association rules based on Boolean attribute mapping, and establishing an influence constraint relation between key influence parameters in the hyperspectral imaging process and mineral identification capacity;
(4) Calculating the attribute importance of the imaging key parameters to the mineral identification capability based on the discretization data table established in the step (3) and the rough set and the knowledge granularity;
(5) Establishing a multiple linear regression model based on a partial least square method, and establishing a quantitative equation between imaging parameters and mineral identification capacity;
(6) Optimizing and solving the quantitative equation in the step (5) according to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity to obtain a hyperspectral imaging parameter combination with the best mineral identification capacity;
step (4) is based on the discretized data table established in step (3)Calculating attribute importance of imaging key parameters on mineral identification capacity based on the rough set and knowledge granularity: the hyperspectral imaging parameter data sheet is T= (U, C, D, V, f), U= { x 1 ,x 2 ,...,x n The method comprises the steps that a is a set of objects in a data table, A is a non-empty limited set of all parameter attributes in the data table, A=CUD, C is a key influence parameter in an imaging system, D is an evaluation index of mineral identification capability, V is a set of parameter values of the data table, and an information function is achieved: f, u×a→v, x=u/d= { X 1 ,X 2 ,...,X n And the equal division of the domain U under the decision attribute D is shown, U/C is the equal division of the conditional characteristics on the domain U, and then X=U/D= { X 1 ,X 2 ,...,X n Knowledge feature resolution on } is:
KCDis(C)＝ρ C (X)gDis(C)
wherein ,ρR (X) is the accuracy of each set of data X in its equivalence relation (knowledge) R, dis (C) is the resolution of the knowledge, and in the hyperspectral imaging parameter data table, for any key influencing parameter C ε C, its characteristic resolution is:
KCDis(c)＝KCDis(C)-KCDis(C-{c})
therefore, the attribute importance of any key influence parameter on the mineral identification capacity is calculated:
and establishing the influence degree action relation of each influence parameter on the mineral identification capacity based on the attribute importance.
2. The method for optimizing design of hyperspectral imaging parameters based on rough set as claimed in claim 1, wherein the hyperspectral imaging process in step (1) is analyzed to determine the main influencing parameters in the imaging process: and establishing a scene model, an atmospheric radiation transmission model and a detector model in the hyperspectral imaging process, analyzing the hyperspectral imaging process, and determining key influence parameters influencing imaging performance in the imaging process mainly comprising ground sampling distance, spectral resolution, signal-to-noise ratio and modulation transfer function.
3. The hyperspectral imaging parameter optimization design method based on the rough set, according to claim 1, wherein the step (2) is to build an imaging parameter data table, and the imaging key parameter values in the data table are divided by using an entropy-based discretization method: establishing an imaging parameter data table, adopting an entropy-based attribute value dividing method, selecting a break point which enables the overall entropy of the imaging parameter data table to be minimum as an interval dividing break point, determining the number of discrete intervals of each parameter attribute by utilizing a minimum description length principle, and discretizing the attribute values of imaging key influence parameters;
the process is as follows:
first, the imaging parameters c are set for each piece of imaging data d 1 The values are arranged in order from small to large, and the obtained sequence is expressed as follows: delta 1 ,δ 2 ,K,δ n The method comprises the steps of carrying out a first treatment on the surface of the Then, record in imaging parameter c 1 Each potential boundary in the range of values is:
wherein ,Li Dividing the candidate interval of the imaging parameter into points, wherein the dividing point L i Sample set D of each imaging parameter i The division into two subsets is noted as:
when selecting L i As a partitioning point, it separates the sample set of the data table into two subsets, the information gain of the imaging data table is expressed as:
wherein ,Di Is a sample set, is (D i ) Information entropy of the sample set;
suppose D in a sample set 1i The number of categories is m 1 ，D 2i The number of categories is m 2 Then when:
terminating partitioning of data at time, wherein Δ=log 2 (3 m-2 -[Ent(D)-m 1 Ent(D 1i )-m 2 Ent(D 2i )]) And obtaining a discretized data table.
4. The hyperspectral imaging parameter optimization design method based on the rough set, according to the value interval divided in the step (2), the step (3) establishes a discretization imaging parameter data table, the imaging data table is mined by adopting association rules based on Boolean attribute mapping, and the influence constraint relation between key influence parameters of the hyperspectral imaging process and mineral identification capacity is established: performing Boolean attribute mapping on the discretized data table, performing association rule mining on the data table by adopting an Apriori association rule mining algorithm, and setting the set of all attribute values of the hyperspectral imaging database to be I= { I 1 ,i 2 ,…,i m -consisting of m items, called item sets, i k The method is characterized in that the method is represented as a certain parameter attribute value in a hyperspectral imaging database, the parameter attribute value is called an item, data of all attribute values under different imaging parameter combinations are recorded as D, T is recorded as each group of imaging parameters in the D, each group of T is a subset of an item set I, rules meeting requirements are selected through the support degree, the lifting degree and the confidence degree of a calculation rule, and the mined association rule is represented as:
wherein ,
x is the front item of the association rule, namely the key influence parameter, Y is the rear item of the association rule, and is the mineral identification capacity, and the influence constraint relation between the key influence parameter and the mineral identification capacity in the hyperspectral imaging process is established through the obtained association rule.
5. The hyperspectral imaging parameter optimization design method based on the rough set, according to claim 1, wherein the step (5) is to build a multiple linear regression model based on a partial least square method, and build a quantitative equation between imaging parameters and mineral identification capacity: establishing quantitative analysis expression of mineral identification capacity evaluation index and hyperspectral imaging performance key influence parameter by adopting partial least square multiple regression analysis method, wherein in the quantitative equation, the evaluation index of the mineral identification capacity is random variable y and common variable x 1 ,x 2 ,...,x p As key influence parameters, y and p (p is more than or equal to 2) common variables x are established through partial least squares regression analysis 1 ,x 2 ,...,x p Is that
wherein ,
epsilon is a random variable, which is the regression coefficient of the regression equation.
6. The hyperspectral imaging parameter optimization design method based on the rough set, according to the influence constraint relation in the step (3) and the attribute importance of each key influence parameter in the step (4) on the mineral identification capacity, in the step (6), optimizing and solving the quantitative equation in the step (5), and obtaining the hyperspectral imaging parameter combination with the best mineral identification capacity: taking the quantitative analysis expression obtained in the step (5) as an objective function, and establishing the following optimization model by combining the constraint relation of the step (3) and the step (4):
min f(x),x∈R n
s.t.c i (x)＝0,i∈E＝{1,2,L,l},
c i (x)≤0,i∈I＝{l+1,l+2,L,l+m}.
based on the optimality condition, solving the constraint optimization problem to obtain an optimal parameter combination, and realizing the optimal design of hyperspectral imaging parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011366930.0A CN113094651B (en) | 2020-11-27 | 2020-11-27 | Hyperspectral imaging parameter optimization design method based on rough set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011366930.0A CN113094651B (en) | 2020-11-27 | 2020-11-27 | Hyperspectral imaging parameter optimization design method based on rough set |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113094651A CN113094651A (en) | 2021-07-09 |
| CN113094651B true CN113094651B (en) | 2023-06-09 |
Family
ID=76663706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011366930.0A Active CN113094651B (en) | 2020-11-27 | 2020-11-27 | Hyperspectral imaging parameter optimization design method based on rough set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113094651B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103488802A (en) * | 2013-10-16 | 2014-01-01 | 国家电网公司 | EHV (Extra-High Voltage) power grid fault rule mining method based on rough set association rule |
| CN103675518A (en) * | 2013-09-29 | 2014-03-26 | 广州供电局有限公司 | Maintenance method and device for oil-immersed high-voltage equipment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9147265B2 (en) * | 2012-06-04 | 2015-09-29 | Raytheon Company | System and method for rapid cluster analysis of hyperspectral images |
-
2020
- 2020-11-27 CN CN202011366930.0A patent/CN113094651B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103675518A (en) * | 2013-09-29 | 2014-03-26 | 广州供电局有限公司 | Maintenance method and device for oil-immersed high-voltage equipment |
| CN103488802A (en) * | 2013-10-16 | 2014-01-01 | 国家电网公司 | EHV (Extra-High Voltage) power grid fault rule mining method based on rough set association rule |
Non-Patent Citations (3)
| Title |
|---|
| A ROUGH SET BASED BAND SELECTION TECHNIQUE FOR THE ANALYSIS OF HYPERSPECTRAL IMAGES;Swarnajyoti Patra等;《2015 IEEE International Geoscience and Remote Sensing Symposium(IGARSS)》;20150731;第497-500页 * |
| Hyperspectral Band Selection Based on Rough Set;Swarnajyoti Patra等;《IEEE Transactions on Geoscience and Remote Sensing》;20150508;第5495-5502页 * |
| 基于高光谱粗糙集分析的矿化信息挖掘;詹云军等;《地质科技情报》;20090531;第28卷(第3期);第63-66页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113094651A (en) | 2021-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Saylor et al. | Quantifying comparison of large detrital geochronology data sets | |
| Bao et al. | Vis-SWIR spectral prediction model for soil organic matter with different grouping strategies | |
| Wadoux et al. | Multi-source data integration for soil mapping using deep learning | |
| Rocchini et al. | Satellite remote sensing to monitor species diversity: Potential and pitfalls | |
| Otgonbayar et al. | Mapping pasture biomass in Mongolia using partial least squares, random forest regression and Landsat 8 imagery | |
| Baisantry et al. | Change vector analysis using enhanced PCA and inverse triangular function-based thresholding | |
| CN104408705A (en) | Anomaly detection method of hyperspectral image | |
| Chen et al. | Rapid determination of soil classes in soil profiles using vis–NIR spectroscopy and multiple objectives mixed support vector classification | |
| CN113959974B (en) | Transmission method of universal near infrared spectrum model | |
| Cheng et al. | Combining multivariate method and spectral variable selection for soil total nitrogen estimation by Vis–NIR spectroscopy | |
| CN108827909B (en) | Rapid soil classification method based on visible near infrared spectrum and multi-target fusion | |
| CN113094651B (en) | Hyperspectral imaging parameter optimization design method based on rough set | |
| CN107764797A (en) | A kind of Raman spectral image data preprocessing method based on low-rank tensor algorithm | |
| CN110659686A (en) | Fuzzy coarse grain outlier detection method for mixed attribute data | |
| Tan et al. | Mutual information-induced interval selection combined with kernel partial least squares for near-infrared spectral calibration | |
| Zhang et al. | A New Approach for Mineral Mapping Using Drill-Core Hyperspectral Image | |
| CN117115652A (en) | Forest land degradation target detection and identification method based on visible spectrum visual remote sensing | |
| CN110321528B (en) | Hyperspectral image soil heavy metal concentration assessment method based on semi-supervised geospatial regression analysis | |
| Tarnik et al. | Minimum variance based-Bayes Combination for prediction of soil properties on Vis-NIR reflectance spectroscopy | |
| CN115392375A (en) | Intelligent evaluation method and system for multi-source data fusion degree | |
| Zhao et al. | Prediction of low Zn concentrations in soil from mountainous areas of central Yunnan Province using a combination of continuous wavelet transform and Boruta algorithm | |
| Castanys et al. | Identification of Raman spectra through a case‐based reasoning system: application to artistic pigments | |
| Nidamanuri et al. | Existence of characteristic spectral signatures for agricultural crops–potential for automated crop mapping by hyperspectral imaging | |
| CN101109658A (en) | High spectrum mineral map plotting method | |
| Zhu et al. | Pixel-level deep spectral features and unsupervised learning for detecting aflatoxin B1 on peanut kernels |
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 |