A spectral width coefficient-based panchromatic band simulation method, system and storage medium

Abstract

The application discloses a full-color waveband simulation method and system based on a spectral width coefficient and a storage medium. The method constructs a spectral width coefficient through the spectral correspondence relationship between a full-color waveband and a multi-spectral waveband, iteratively decomposes the high-resolution full-color waveband by using a CN fusion method to obtain optimal structural coefficients of the multi-spectral waveband, simulates a low-resolution full-color waveband by using linear combination of the multi-spectral waveband, constructs a multi-spectral waveband weight coefficient equation based on the spectral width coefficient and the structural coefficient, solves unknown power of the equation by using a cyclic iteration method, determines optimal weight coefficients, and realizes accurate simulation of the low-resolution full-color waveband. Compared with commonly used average value simulation, least square simulation and other methods, the simulated low-resolution full-color waveband image has higher correlation with the high-resolution full-color waveband image, the spectral information of the image is more consistent, the gray color consistency of water, vegetation, bare ground, buildings and other surface features is better, and the simulation of the surface feature spectrum of the full-color waveband is more accurate.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image band simulation technology, and more specifically, to a panchromatic band simulation method, system, and storage medium based on spectral width coefficient. Background Technology

[0002] The development of satellite remote sensing image simulation technology has always revolved around overcoming sensor performance bottlenecks and improving data utilization efficiency. Among these efforts, research on simulating low-resolution panchromatic bands using low-resolution multispectral bands has become a hot topic in recent years because it can provide fundamental data support for improving the spatial accuracy of remote sensing images, such as spatial-spectral fusion. The technical methods in this field have gradually evolved from traditional statistical models to a fusion of physical mechanisms and data-driven approaches. The core idea is to achieve accurate grayscale information reconstruction by mining the correlation between the spectral responses of multispectral and panchromatic bands.

[0003] Traditional simulation methods, centered on simple mathematical models, still hold an important position in practical applications. The average value method, as the most basic approach, directly generates a simulated panchromatic image by calculating the gray-scale mean of all multispectral bands. This method is widely used in preprocessing steps of remote sensing image spatial-spectral fusion, such as Gram-Schmidt spectral sharpening, for rapidly constructing low-resolution panchromatic reference bands. The weighted method optimizes simulation accuracy by introducing the spectral response function (SRF), assigning weights based on the spectral overlap between each multispectral band and the panchromatic band. In recent years, objective weighting methods such as the entropy method have been developed, utilizing the inherent information content or volatility of the data to replace subjective judgment, significantly improving the scientific rigor of weight allocation. The least squares method establishes an optimal fitting model by minimizing the sum of squared errors between simulated and true values. In modern applications, it is often combined with L1 and L2 regularization techniques to avoid overfitting, and optimization algorithms such as gradient descent and singular value decomposition are used to improve solution efficiency. It performs exceptionally well in integrated tasks of image denoising, geometric correction, and simulation.

[0004] With the increasing volume of remote sensing data and the improvement of computing power, hybrid methods integrating physical mechanisms and machine learning have become a new research trend. Some studies have constructed simulation databases based on atmospheric radiative transfer models and fine-tuned neural networks through transfer learning to achieve accurate simulation of low-resolution panchromatic images in complex scenes. This method has shown advantages in tasks requiring high-frequency observation data, such as aerosol inversion. In addition, model-based deep learning networks incorporate the physical formulas of traditional multi-resolution analysis into the network structure design, achieving end-to-end mapping from multispectral features to panchromatic information through iterative optimization. This preserves physical interpretability while possessing data-driven adaptive capabilities. Furthermore, to address the spectral response differences between multispectral and panchromatic bands, researchers have effectively reduced spectral distortion and noise interference during the simulation process by introducing masking, edge-protecting filters, and radiometric consistency loss functions.

[0005] In the future, research in this field will further focus on improving cross-sensor adaptability, optimizing simulation robustness in extreme weather scenarios, and engineering applications of lightweight models. Through multi-source data fusion and advanced algorithm innovation, the accuracy and efficiency of low-resolution panchromatic image simulation technology will be continuously promoted.

[0006] Based on the principle of spatial-spectral linear fusion of remote sensing images, multispectral bands need to be used to simulate low-resolution panchromatic bands, serving as intermediate bands for spatial-spectral fusion of multispectral and high-resolution panchromatic images. Studies show that the higher the correlation coefficient between the intermediate band (low-resolution panchromatic band) and the high-resolution panchromatic band, the higher their similarity, resulting in better spatial-spectral fusion performance. Therefore, constructing an excellent low-resolution panchromatic band is a major fundamental technical problem in the field of spatial-spectral fusion of remote sensing images. Summary of the Invention

[0007] In view of the above problems, the purpose of this invention is to provide a panchromatic band simulation method, system and storage medium based on spectral width coefficient.

[0008] The first aspect of this invention provides a panchromatic band simulation method based on spectral width coefficient, characterized in that the method includes the following steps:

[0009] S1: Input panchromatic and multispectral images to obtain high-resolution panchromatic bands. and low-resolution multispectral bands ;in, , The number of bands in a multispectral band;

[0010] S2: Spatial registration of multispectral and panchromatic images to ensure that the geometric spatial position of the same ground feature in the panchromatic and multispectral images is consistent.

[0011] S3: Resample the low-resolution multispectral image as a high-resolution panchromatic image;

[0012] S4: Statistical analysis of all low-resolution multispectral bands and panchromatic band The mean, standard deviation, covariance, and correlation coefficient matrix;

[0013] S5: Based on the panchromatic band of satellite sensors With multispectral bands Spectral parameters, calculate multispectral bands spectral width coefficient ;

[0014] S6: Using the spectral width factor method as a weighting factor, with multispectral bands Linear combination simulation of low-resolution panchromatic band Calculate each low-resolution panchromatic band With high resolution panchromatic band correlation coefficient ;

[0015] S7: Optimal correlation coefficient The basic combination of multispectral bands corresponding to the maximum value The basic combination of this multispectral band As a basic combination scheme; among which... , The number of bands in the basic combination of multispectral bands. ;

[0016] S8: Employing the CN fusion method to combine multispectral bands For panchromatic band Perform iterative decomposition to obtain multispectral bands Structural coefficient sequence;

[0017] S9: Based on multispectral band fundamental combination Determine multispectral bands Optimal structural coefficient ;

[0018] S10: Utilizing multispectral bands spectral width coefficient and optimal structure coefficient Construct the weighting coefficient equation ;

[0019] S11: Low-resolution panchromatic band With high resolution panchromatic band correlation coefficient With the objective of maximizing the value, an iterative method is used to solve for the unknown power. Determine the weighting coefficients Calculate and store low-resolution panchromatic band images Simulation results.

[0020] Preferably, the spectral width coefficient of the multispectral band is... The calculation method is as follows:

[0021] Let the spectral width corresponding to the high-resolution panchromatic band spectral range be... Multispectral bands , ,… …, The spectral widths overlapping with the panchromatic band spectral range are respectively , , ..., , ..., The spectral width coefficients for constructing the multispectral bands are respectively , , ..., , ..., ;

[0022] The specific formula for calculating the spectral width factor is as follows:

[0023]

[0024] in, This represents the number of multispectral bands.

[0025] Preferably, S8 specifically includes:

[0026] Using CN fusion method for high-resolution panchromatic band Decomposition was performed to obtain high-resolution multispectral bands. :

[0027]

[0028] in, ; Low-resolution multispectral bands obtained by CN fusion method The corresponding high-resolution multispectral bands; This represents the spectral width coefficient for the multispectral band. Here are the structure coefficients for each multispectral band, with initial values ​​taken as... ; For low-resolution multispectral bands;

[0029] application For multispectral bands Histogram matching was performed to obtain the multispectral bands after structural optimization. ,Right now:

[0030]

[0031] in, For high resolution multispectral bands standard deviation Multispectral band standard deviation Multispectral band The mean, For high resolution multispectral bands The mean, This refers to the multispectral bands after structural optimization.

[0032] use The sum of analog low-resolution panchromatic bands ,Right now:

[0033]

[0034] With low resolution panchromatic band With panchromatic band When the correlation is used as the evaluation criterion, the above formula is equivalent to:

[0035]

[0036] New structure coefficients in each multispectral band ,Right now:

[0037]

[0038] Similarly, a second decomposition operation can be performed:

[0039]

[0040]

[0041] Even better,

[0042]

[0043] Based on the new structure coefficients of the multispectral bands, iterative decomposition operations are performed successively to obtain the multispectral band structure coefficient sequence after iterative decomposition using the high-resolution panchromatic band.

[0044] Preferably, S9 specifically comprises:

[0045] make:

[0046]

[0047] right Proceed to the first Sub-iteration decomposition;

[0048] calculate With panchromatic image The correlation coefficient;

[0049] Pick With panchromatic image The optimal structure coefficient corresponding to the maximum correlation coefficient The optimal structure coefficient for multispectral bands value.

[0050] Preferably, in step S10, a weighting coefficient equation is constructed. First, determine the basic combination of multispectral bands. In addition to the positive and negative values ​​of the weighting coefficients for the multispectral bands (if If the number of multispectral bands other than one is greater than one, then only the weighting coefficient of the band with the smaller spectral width coefficient is determined (positive or negative sign).

[0051] Preferably, the method for determining the sign of the weighting coefficient is as follows:

[0052] Assume there are n multispectral bands in total, where the multispectral bands are... Not a basic combination of multispectral bands ,assumed Then let:

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060] in, This indicates that, apart from the target band, the rest... Linear weighted combination value of multiple spectral bands;

[0061] like If the calculation result is positive, then its sign is positive; if... If the calculation result is negative, then its sign is negative.

[0062] Preferably, S11 specifically includes:

[0063] Based on multispectral band combination The formula for simulating low-resolution panchromatic imagery is as follows:

[0064]

[0065] in, For the first The power parameters of the optimal structure coefficients for each multispectral band;

[0066] If a certain multispectral band is a basic combination of bands Other Then the weighting coefficient of the multispectral band use Taking the sign of the signal for the basic combination of multispectral bands Among them ,but ;

[0067] The power parameters of the optimal structure coefficients for each multispectral band are obtained sequentially. This continues until each band has been set once. ;

[0068] According to the formula This allows us to obtain all multispectral bands. New weighting coefficient values Then according to the formula Calculate the new correlation coefficient ;

[0069] by The maximum value appears As a multispectral band Final weighting coefficients After normalization, it is used as the optimal weight combination.

[0070] Preferably, the power parameter for obtaining the optimal structure coefficients for each multispectral band is... Specifically:

[0071] set up It is a constant. , making The largest The following formula can be used to obtain:

[0072] make:

[0073]

[0074]

[0075] We can obtain:

[0076]

[0077]

[0078]

[0079]

[0080] but:

[0081]

[0082] A second aspect of the present invention provides a panchromatic band simulation system based on spectral width factor, comprising a memory and a processor. The memory includes a program for a panchromatic band simulation method based on spectral width factor. When the program for the panchromatic band simulation method based on spectral width factor is executed by the processor, it implements the steps of the panchromatic band simulation method based on spectral width factor.

[0083] A third aspect of the present invention provides a computer storage medium, wherein the computer storage medium includes a panchromatic band simulation method program based on spectral width factor, and when the panchromatic band simulation method program based on spectral width factor is executed by a processor, the steps of the panchromatic band simulation method based on spectral width factor are implemented.

[0084] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0085] This invention constructs a spectral width coefficient by establishing the spectral correspondence between panchromatic and multispectral bands, uses the CN fusion method to iteratively decompose the high-resolution panchromatic band to obtain the multispectral band structure coefficient, utilizes a linear combination of multispectral bands to simulate the low-resolution panchromatic band, constructs a multispectral band weight coefficient equation based on the spectral width coefficient and structure coefficient, and uses a cyclic iterative method to solve the unknown powers of the equation to determine the optimal weight coefficient, thereby achieving accurate simulation of the low-resolution panchromatic band.

[0086] The effectiveness of the proposed method was verified by visual discrimination and correlation analysis of the simulated low-resolution panchromatic band image (intermediate band). Compared with the commonly used average value simulation method and least squares simulation method, the low-resolution panchromatic band image simulated by this method has a higher correlation with the high-resolution panchromatic band image, more consistent image spectral information, better consistency of grayscale color of land features such as water bodies, vegetation, bare ground surfaces, and buildings, and more accurate spectral simulation of land features in the panchromatic band. Attached Figure Description

[0087] Figure 1 This is a flowchart of a panchromatic band simulation method based on the spectral width coefficient.

[0088] Figure 2 This is a high-resolution panchromatic (P) image (0.5-meter resolution).

[0089] Figure 3 This is a low-resolution multispectral RGB true-color composite image (2-meter resolution).

[0090] Figure 4 The method described in Example 1 simulates a low-resolution panchromatic band (I) image (2-meter resolution).

[0091] Figure 5 A low-resolution panchromatic band (I) image (2-meter resolution) was simulated using the averaging method.

[0092] Figure 6 A low-resolution panchromatic band (I) image (2-meter resolution) simulated using the least squares method. Detailed Implementation

[0093] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0094] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0095] Example 1

[0096] like Figure 1 As shown in the figure, this embodiment discloses a panchromatic band simulation method based on the spectral width coefficient. The method includes the following steps:

[0097] S1: Input panchromatic and multispectral images to obtain high-resolution panchromatic bands. and low-resolution multispectral bands ;in, , This represents the number of segments in the multispectral bands.

[0098] S2: Spatial registration is performed between the multispectral and panchromatic images to ensure that the geometric spatial position of the same ground feature in the panchromatic and multispectral images is consistent.

[0099] S3: Resample low-resolution multispectral images as high-resolution panchromatic images.

[0100] S4: Statistical analysis of all low-resolution multispectral bands and panchromatic band The mean, standard deviation, covariance, and correlation coefficient matrix.

[0101] S5: Based on the panchromatic band of satellite sensors With multispectral bands Spectral parameters, calculate multispectral bands spectral width coefficient .

[0102] In this embodiment, the spectral width coefficient of the multispectral band The calculation method is as follows:

[0103] Let the spectral width corresponding to the high-resolution panchromatic band spectral range be... Multispectral bands , ,… …, The spectral widths overlapping with the panchromatic band spectral range are respectively , , ..., , ..., The spectral width coefficients for constructing the multispectral bands are respectively , , ..., , ..., ;

[0104] The specific formula for calculating the spectral width factor is as follows:

[0105]

[0106] in, This represents the number of multispectral bands.

[0107] S6: Using the spectral width coefficient as a weighting coefficient, with multispectral bands Linear combination simulation of low-resolution panchromatic band And calculate each low-resolution panchromatic band. With high resolution panchromatic band correlation coefficient .

[0108] S7: Optimal correlation coefficient The basic combination of multispectral bands corresponding to the maximum value The basic combination of this multispectral band As a basic combination scheme; among which... , The number of bands in a multispectral band combination. .

[0109] It should be noted that for the basic combination of multispectral bands The specific operation in this embodiment is as follows:

[0110] Let the low-resolution multispectral band be... , , ... High-resolution panchromatic band is The simulated low-resolution panchromatic band (intermediate band) is Low-resolution panchromatic bands are typically simulated using a linear combination of the original multispectral bands, i.e.:

[0111]

[0112] in, These are the weighting coefficients.

[0113] To ensure the reasonableness of the data values ​​and range in the simulated low-resolution panchromatic band, normalization is generally performed on the linear combination coefficients after they have been determined. That is:

[0114]

[0115] There are a total of (such linear combination schemes) (Number) . Each scheme can be calculated. and Correlation coefficient Clearly, different schemes yield different intermediate bands. Different, therefore how to find The largest linear combination scheme is a major fundamental problem that needs to be studied in depth in the field of space-spectrum fusion.

[0116] This embodiment uses the spectral width factor method to optimize the basic multispectral combination scheme, selecting a high-resolution panchromatic band. The low-resolution multispectral band combination with the highest correlation coefficient This basic combination scheme is used as the multispectral band basic combination for the preferred structural coefficients. .

[0117]

[0118] In the formula, For each multispectral band, represents the spectral width coefficient. The initial structure coefficients for each multispectral band are 1.

[0119] S8: Employing the CN fusion method to combine multispectral bands For panchromatic band Perform iterative decomposition to obtain multispectral bands Structural coefficient sequence.

[0120] In this embodiment, S8 specifically refers to:

[0121] Using CN fusion method for high-resolution panchromatic band Decomposition was performed to obtain high-resolution multispectral bands. :

[0122]

[0123] in, ; Low-resolution multispectral bands obtained by CN fusion method The corresponding high-resolution multispectral bands; This represents the spectral width coefficient for the multispectral band. Here are the structure coefficients for each multispectral band, with initial values ​​taken as... ; It is multispectral;

[0124] application For multispectral bands Histogram matching was performed to obtain the multispectral bands after structural optimization. ,Right now:

[0125]

[0126] in, For high resolution multispectral bands standard deviation Multispectral band standard deviation Multispectral band The mean, For high resolution multispectral bands The mean, This refers to the multispectral bands after structural optimization.

[0127] use The sum of analog low-resolution panchromatic bands ,Right now:

[0128]

[0129] With low resolution panchromatic band With panchromatic band When the correlation is used as the evaluation criterion, the above formula is equivalent to:

[0130]

[0131] New structure coefficients in each multispectral band ,Right now:

[0132]

[0133] Based on the new structure coefficients of the multispectral bands, iterative decomposition operations are performed successively to obtain the multispectral band structure coefficient sequence after iterative decomposition using the high-resolution panchromatic band.

[0134] It should be noted that, for obtaining the multispectral structure coefficient sequence after iterative decomposition using the high-resolution panchromatic band, this embodiment applies the traditional CN fusion method for the high-resolution panchromatic band. Decompose the image to obtain a high-resolution multispectral image. .

[0135] From the CN fusion method, we can obtain:

[0136]

[0137] in, . Obtained by CN fusion method The corresponding high-resolution multispectral bands; For each multispectral band, represents the spectral width coefficient. Here are the structure coefficients for each multispectral band, with initial values ​​taken as... .

[0138] application For multispectral bands Histogram matching was performed to obtain the bands after structural optimization. ,Right now:

[0139]

[0140] As with all high-resolution multispectral images The sum can completely and reversibly reconstruct a high-resolution panchromatic image. One can imagine The sum should be able to effectively simulate low-resolution panchromatic bands. .Right now:

[0141]

[0142] With intermediate band With panchromatic band When the relevance is used as the evaluation criterion, the above formula is equivalent to:

[0143]

[0144] New structure coefficients in each multispectral band ,Right now:

[0145]

[0146] Similarly, a second decomposition operation can be performed:

[0147]

[0148]

[0149] Even better,

[0150]

[0151] Based on the above method, iterative decomposition operations are performed successively to obtain the multispectral band structure coefficient sequence after iterative decomposition using high-resolution panchromatic bands.

[0152] S9: Based on multispectral band fundamental combination Determine multispectral bands Optimal structural coefficient .

[0153] It should be noted that this is based on a multispectral band fundamental combination. Simulated low-resolution panchromatic band It can more clearly reflect the analog low-resolution panchromatic band. With panchromatic image correlation coefficient Therefore, this embodiment is based on a multispectral band fundamental combination due to the changes. The preferred structure coefficients in the above multispectral structure coefficient sequence are those with the following characteristics. The specific method is as follows:

[0154] make:

[0155]

[0156] If after the l-th iteration of decomposition... With panchromatic image The correlation coefficient is larger. This is still not optimal and requires further decomposition. If With panchromatic image The correlation coefficient is smaller, so no further iterative decomposition calculation is needed. The optimal structural coefficient is .

[0157] Thus, the optimal structure coefficients for each multispectral band are determined. value.

[0158] S10: Utilizing multispectral bands spectral width coefficient and optimal structure coefficient Construct the weighting coefficient equation .

[0159] In this embodiment, the weighting coefficient equation is constructed in step S10. First, determine the sign of the weighting coefficient for the band with the smaller spectral width coefficient.

[0160] In this embodiment, the method for determining the sign of the weighting coefficient for the band with the smaller spectral width coefficient is as follows:

[0161] set up Not a basic combination of multispectral bands ,assumed Then let:

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169] in, This indicates that, apart from the target band, the rest... Linear weighted combination value of multiple spectral bands;

[0170] like If the calculation result is positive, then its sign is positive; if... If the calculation result is negative, then its sign is negative.

[0171] It should be noted that, in this embodiment, the optimal structure coefficient for the multispectral band is determined. Then, using the spectral width coefficient With structural coefficient Multiplication yields a linear combination of multispectral bands. Simulated low-resolution panchromatic band The weighting coefficients, i.e.:

[0172]

[0173] To better describe the linear combination of multispectral bands The change in the weighting coefficients, which can generally be written as:

[0174]

[0175] Iterative calculations are needed to solve for the unknown powers of the weighting coefficients for each multispectral band. Thus, the weighting coefficients are determined. Due to the default weight coefficients in the loop iteration. It is a positive value. If a certain multispectral band does not belong to the basic band combination. Among them Its weighting coefficient When the actual value is negative, the iterative calculation will terminate. Therefore, the basic band combination must be determined first. Other Band weighting coefficient The sign of the value (positive or negative).

[0176] As a specific example, using four multispectral band images, the method for determining the sign of the weighting coefficients is as follows:

[0177] like Not a basic combination of multispectral bands ,assumed Then let:

[0178]

[0179]

[0180]

[0181]

[0182]

[0183]

[0184]

[0185] like If the calculation result is positive, then its sign is positive (1); if If the calculation result is negative, then its sign is negative (-1).

[0186] S11: Low-resolution panchromatic band With high resolution panchromatic band correlation coefficient With the objective of maximizing the value, an iterative method is used to solve for the unknown power. Determine the weighting coefficients Calculate and store low-resolution panchromatic band images Simulation results.

[0187] In this implementation, S11 specifically refers to:

[0188] Based on multispectral band combination The formula for simulating low-resolution panchromatic imagery is as follows:

[0189]

[0190] in, For the first The power parameters of the optimal structure coefficients for each multispectral band;

[0191] If a certain multispectral band is a basic combination of bands Other Then the weighting coefficient of the multispectral band use Take the sign of the positive or negative value (if) If the number of multispectral bands other than one is greater than one, then only the weighting coefficient (sign of positive or negative value) of the band with the smaller spectral width coefficient is determined. This applies to the selection of a basic multispectral band combination. Among them ,but ;

[0192] The power parameters of the optimal structure coefficients for each multispectral band are obtained sequentially. This continues until each band has been set once. ;

[0193] According to the formula This allows us to obtain all multispectral bands. New weighting coefficient values Then according to the formula Calculate the new correlation coefficient ;

[0194] by The maximum value appears As a multispectral band Final weighting coefficients After normalization, it is used as the optimal weight combination.

[0195] In this embodiment, the power parameter for obtaining the optimal structure coefficients of each multispectral band is... Specifically:

[0196] set up It is a constant. , making The largest The following formula can be used to obtain:

[0197] make:

[0198]

[0199]

[0200] We can obtain:

[0201]

[0202]

[0203]

[0204]

[0205] but:

[0206]

[0207] It should be noted that in this embodiment, when determining the spectral width coefficient of the multispectral image... and structural coefficient Subsequently, it can be based on multispectral band combinations The formula for simulating low-resolution panchromatic imagery is as follows:

[0208]

[0209] Among them, if a certain multispectral band is a basic combination of bands Other Then the weighting coefficient of the multispectral band is used The sign is determined, and the value is 1 or -1; for the basic combination of selected bands Among them , .

[0210] The above low-resolution panchromatic image simulation formula uses the correlation coefficient With the objective of maximizing, the unknown power is solved using a cyclic iterative method. .when Let be any real number, according to the known... , There is always a corresponding Since they are both possible, they are equivalent.

[0211] This structure can be described as: a simulated low-resolution panchromatic image. Based on multispectral imagery A linear combination is performed, and the linear combination coefficients for each band are exponential functions of the spectral width coefficient of each band multiplied by the optimal structure optimization coefficient. Thus, the coefficients of a linear combination in a mathematical sense acquire a clear physical meaning.

[0212] According to the principle of scale invariance, when Maximum To achieve what is desired.

[0213] set up:

[0214]

[0215]

[0216] but:

[0217]

[0218] in, The mean square error of the panchromatic band. The covariance between the panchromatic band and the intermediate band. This represents the variance of the intermediate band.

[0219] This embodiment takes a multispectral band of 4 bands as an example to calculate the power of the exponential function corresponding to each band. Then we have:

[0220]

[0221]

[0222] assumed It is a constant. , making The largest It can be obtained using the following formula:

[0223] make:

[0224]

[0225]

[0226] We can obtain:

[0227]

[0228]

[0229]

[0230]

[0231] but:

[0232]

[0233] The above method can be used to obtain , The solution is repeated iteratively until each band has been set once. According to the formula This allows us to obtain all multispectral bands. New weighting coefficient values Then according to the formula Calculate the new correlation coefficient .

[0234] If the difference in the correlation coefficient between the nth iteration and the (n-1)th iteration is ( Then repeat the above. Iterative solution; if If the iteration stops, then stop.

[0235] by The maximum value appears As a multispectral band Final weighting coefficients After normalization, it becomes the optimal weight combination.

[0236] The method described in this embodiment is applicable to simulating low-resolution panchromatic band images using low-resolution multispectral images and high-resolution panchromatic images, so as to truly restore the spectral information of low-resolution panchromatic images and lay the foundation for high-quality space-spectral fusion of multispectral and panchromatic images.

[0237] As a specific embodiment, the method described in this embodiment will be explained below with reference to specific examples:

[0238] To achieve the goal of simulating low-resolution panchromatic band images, this embodiment mainly utilizes ENVI remote sensing image processing software, and is further described using a satellite remote sensing image with panchromatic band (P), near-infrared (N), red band (R), green band (G), and blue band (B).

[0239] 1. Input remote sensing image.

[0240] Open a Beijing-3 (BJ3) remote sensing image with panchromatic (P), near-infrared (N), red (R), green (G), and blue (B) bands. Figure 2 , Figure 3 These are panchromatic band images (0.5-meter resolution) and multispectral RGB composite images (2-meter resolution) (stretched images according to the default settings of the ENVI software).

[0241] 2. Calculate image statistical parameters.

[0242] Calculate the mean, standard deviation, covariance, and correlation coefficient matrix of each multispectral and panchromatic image. The statistical parameters of the remote sensing images are shown in Table 1.

[0243] Table 1 Statistical Parameters of Remote Sensing Images

[0244]

[0245] 3. Calculate the spectral width coefficient of the multispectral band.

[0246] Query panchromatic imagery from Beijing-3 satellite sensor With multispectral images The spectral parameters were used to calculate the spectral width coefficients of the multispectral bands (as shown in Table 2). The spectral width coefficients of the N, R, G, and B bands were obtained. The values ​​are 0.041666667, 0.291666667, 0.333333333 and 0.333333333 respectively.

[0247] Table 2 Calculation Table of Multispectral Width Coefficient of Beijing No. 3

[0248]

[0249] 4. Prefer multispectral bands.

[0250] This BJ3 remote sensing image contains There are a total of 4 multispectral bands. Let the initial structure coefficients of each multispectral band be set. Both are 1, known Spectral width coefficient of the band The values ​​are 0.041666667, 0.291666667, 0.333333333, and 0.333333333, respectively. Simulating low-resolution panchromatic images using different multispectral band combinations Calculate each With high-resolution panchromatic images correlation coefficient Table 2 shows that combination scheme 5... The maximum value is 0.916124865; therefore, the corresponding multispectral bands are combined. As a basic combination scheme.

[0251] Table 3. Statistical table of r(P,I) for different combinations of multispectral bands

[0252]

[0253] 5. Determine the optimal structure coefficients for the multispectral system.

[0254] Assuming the initial structure coefficients of each band in the multispectral bands are all 1, using the multispectral bands For panchromatic band Perform the first decomposition to obtain high-resolution multispectral images. ,use right Perform histogram matching to obtain new low-resolution multispectral images. Calculate the structural optimization coefficients The correlation coefficient r(P,I) with the optimal band simulation I under the corresponding structural coefficient is calculated. This process is repeated for each subsequent decomposition and matching step to obtain the structural optimization coefficients and calculate the correlation coefficient r(P,I). When r(P,I) reaches its maximum value, the structural optimization coefficient obtained in that decomposition is taken as the optimal structural coefficient.

[0255] In this case, the maximum correlation coefficient r(P,I) obtained during the third decomposition is 0.91610529825. Therefore, the structure optimization coefficient S3 is taken as the optimal structure optimization coefficient for this multispectral image, and the structure optimization coefficients for the N, R, G, and B bands are 1.134107827, 1.159202817, 1.206257955, and 1.27882902, respectively.

[0256] Table 4. Successive Calculation of Structural Coefficients and Corresponding Correlation Coefficients

[0257]

[0258] 6. Determine the weighting coefficients for bands with smaller spectral width coefficients. Positive and negative value signs.

[0259] In this case, the near-infrared band (N) in the multispectral bands does not belong to the basic band combination, and its spectral width coefficient is 0.041666667. Its weighting coefficient is calculated. The sign of its weight coefficient value Positive, with a value of 1.

[0260] 7. Determine the weighting coefficients of the linear combination with the objective of maximizing the correlation coefficient. value.

[0261] The spectral width coefficients of the above multispectral bands are known. and structural coefficient The linear combination weighting coefficients of the multispectral bands simulating the low-resolution panchromatic band I are used. The goal is to maximize the correlation coefficient r(P,I) between the low-resolution panchromatic band I and the high-resolution panchromatic band P. A power-law solution is obtained using a cyclic iterative method. Determine the weighting coefficients .

[0262] As can be seen from Table 5, in this case when During the fifth iteration, the correlation coefficient r(P,I) reached its maximum value of 0.916226906513344. Subsequent iterations failed to exceed this value, thus determining the linear combination weight coefficients of multispectral bands N, R, G, and B to be 0.014327717, 0.283650690, 0.410544877, and 0.291476717, respectively, achieving accurate simulation of the low-resolution panchromatic band image I.

[0263] like Figure 4-6As shown, compared to the traditional multispectral averaging method, the correlation coefficient r(P,I) between the simulated low-resolution panchromatic band image and the high-resolution panchromatic band image increases from 0.89944382717187 in the averaging method to 0.916226906513344 in this method, improving the information consistency between the simulated low-resolution panchromatic band I and the high-resolution panchromatic band P by 1.87%. Compared to the least squares simulation method, the correlation coefficient r(P,I) increases from 0.916224821477691 in the least squares method to 0.916226906513344 in this method, improving the information consistency between the simulated low-resolution panchromatic band I and the high-resolution panchromatic band P by 0.0002%.

[0264] Table 5. Weighting coefficients for linear combinations Value Iteration Calculation Table

[0265]

[0266] As can be seen from the above examples, the low-resolution panchromatic image simulated by the method described in this embodiment has a higher correlation coefficient with the high-resolution panchromatic image. The overall grayscale color and ground feature consistency between the simulated panchromatic image and the real panchromatic image are better, the information is more realistic and consistent, the simulated image better reflects the spectral information of the high-resolution panchromatic image, and the accuracy of the simulation results of the low-resolution panchromatic band image is greatly improved, which helps to improve the accuracy of the spatial-spectral fusion results of remote sensing images.

[0267] Example 2

[0268] This embodiment discloses a panchromatic band simulation system based on spectral width factor, including a memory and a processor. The memory includes a panchromatic band simulation method program based on spectral width factor. When the processor executes the panchromatic band simulation method program based on spectral width factor, it implements the steps of a panchromatic band simulation method based on spectral width factor as described in Embodiment 1.

[0269] Example 3

[0270] This embodiment discloses a computer storage medium, which includes a panchromatic band simulation method program based on spectral width factor. When the panchromatic band simulation method program based on spectral width factor is executed by a processor, it implements the steps of a panchromatic band simulation method based on spectral width factor as described in Embodiment 1.

[0271] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0272] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0273] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0274] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0275] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A panchromatic band simulation method based on spectral width coefficient, characterized in that, The method includes the following steps: S1: Input panchromatic and multispectral images to obtain high-resolution panchromatic bands. and low-resolution multispectral bands ;in, , The number of bands in a multispectral band; S2: Spatial registration of multispectral and panchromatic images to ensure that the geometric spatial position of the same ground feature in the panchromatic and multispectral images is consistent. S3: Resample the low-resolution multispectral image as a high-resolution panchromatic image; S4: Statistical analysis of all low-resolution multispectral bands and panchromatic band The mean, standard deviation, covariance, and correlation coefficient matrix; S5: Based on the panchromatic band of satellite sensors With multispectral bands Spectral parameters, calculate multispectral bands spectral width coefficient ; S6: Using the spectral width coefficient as a weighting coefficient, with multispectral bands Linear combination simulation of low-resolution panchromatic band Calculate each low-resolution panchromatic band With high resolution panchromatic band correlation coefficient ; S7: Optimal correlation coefficient The basic combination of multispectral bands corresponding to the maximum value The basic combination of this multispectral band As a basic combination scheme; among which... , The number of bands in a multispectral band combination. ; S8: Employing the CN fusion method to combine multispectral bands For panchromatic band Perform iterative decomposition to obtain multispectral bands Structural coefficient sequence; S9: Based on multispectral band fundamental combination Determine multispectral bands Optimal structural coefficients ; S10: Utilizing multispectral bands spectral width coefficient and optimal structure coefficient Construct the weighting coefficient equation ; S11: Low-resolution panchromatic band With high resolution panchromatic band correlation coefficient With the objective of maximizing the value, an iterative method is used to solve for the unknown power. Determine the weighting coefficients Calculate and store low-resolution panchromatic band images Simulation results.

2. The panchromatic band simulation method based on spectral width coefficient according to claim 1, characterized in that, The spectral width coefficient of the multispectral band The calculation method is as follows: Let the spectral width corresponding to the high-resolution panchromatic band spectral range be... Multispectral bands , ,… …, The spectral widths overlapping with the panchromatic band spectral range are respectively , , ..., , ..., The spectral width coefficients for constructing the multispectral bands are respectively , , ..., , ..., ; The specific formula for calculating the spectral width coefficient is as follows: in, This represents the number of multispectral bands.

3. The panchromatic band simulation method based on spectral width coefficient according to claim 2, characterized in that, Specifically, S8 is: Using CN fusion method for high-resolution panchromatic band Decomposition was performed to obtain high-resolution multispectral bands. : in, ; Low-resolution multispectral bands obtained by CN fusion method The corresponding high-resolution multispectral bands; This represents the spectral width coefficient for the multispectral band. Here are the structure coefficients for each multispectral band, with initial values ​​taken as... ; For low-resolution multispectral bands; application For multispectral bands Histogram matching was performed to obtain the multispectral bands after structural optimization. ,Right now: in, For high resolution multispectral bands standard deviation Multispectral band standard deviation Multispectral band The mean, For high resolution multispectral bands The mean, For the optimized multispectral bands; use The sum of analog low-resolution panchromatic bands ,Right now: With low resolution panchromatic band With panchromatic band When the correlation is used as the evaluation criterion, the above formula is equivalent to: New structure coefficients in each multispectral band ,Right now: Similarly, a second decomposition operation can be performed: Even better, Based on the new structure coefficients of the multispectral bands, iterative decomposition operations are performed successively to obtain the multispectral band structure coefficient sequence after iterative decomposition using the high-resolution panchromatic band.

4. The panchromatic band simulation method based on spectral width coefficient according to claim 3, characterized in that, Specifically, S9 is: make: right Proceed to the first Sub-iteration decomposition; calculate With panchromatic image The correlation coefficient; Pick With panchromatic image When the correlation coefficient is at its maximum, the corresponding optimal structure coefficient is... The optimal structure coefficient for multispectral bands value.

5. The panchromatic band simulation method based on spectral width coefficient according to claim 4, characterized in that, The weighting coefficient equation is constructed in S10. First, the basic combination of multispectral bands needs to be determined. In addition to the sign of the weighting coefficients for the multispectral bands, if the multispectral band basic combination If the number of multispectral bands other than one is greater than one, then only the sign of the weight coefficient of the band with the smaller spectral width coefficient is determined.

6. The panchromatic band simulation method based on spectral width coefficient according to claim 5, characterized in that, The method for determining the sign of the weighting coefficients for the multispectral bands is as follows: Assume there are n multispectral bands in total, where the multispectral bands are... Not a basic combination of multispectral bands ,assumed Then let: in, This indicates that, apart from the target band, the rest... Linear weighted combination value of multiple spectral bands; like If the calculation result is positive, then its sign is positive; if... If the calculation result is negative, then its sign is negative.

7. The panchromatic band simulation method based on spectral width coefficient according to claim 6, characterized in that, Specifically, S11 is: Based on multispectral band combination The formula for simulating low-resolution panchromatic imagery is as follows: in, For the first The power parameters of the optimal structure coefficients for each multispectral band; If a certain multispectral band is a basic combination of bands Other Then the weighting coefficient of the multispectral band use Taking the sign of the signal for the basic combination of multispectral bands Among them ,but ; The power parameters of the optimal structure coefficients for each multispectral band are obtained sequentially. This continues until each band has been set once. ; According to the formula This allows us to obtain all multispectral bands. New weighting coefficient values Then according to the formula Calculate the new correlation coefficient ; by The maximum value appears As a multispectral band Final weighting coefficients After normalization, it is used as the optimal weight combination.

8. The panchromatic band simulation method based on spectral width coefficient according to claim 7, characterized in that, The power parameter for obtaining the optimal structure coefficients for each multispectral band. Specifically: set up It is a constant. , making The largest The following formula can be used to obtain: make: We can obtain: but: 。 9. A panchromatic band simulation system based on spectral width coefficient, characterized in that, The device includes a memory and a processor. The memory includes a program for a panchromatic band simulation method based on a spectral width factor. When the processor executes the program for the panchromatic band simulation method based on a spectral width factor, it implements the steps of a panchromatic band simulation method based on a spectral width factor as described in any one of claims 1 to 8.

10. A computer storage medium, characterized in that, The computational read storage medium includes a panchromatic band simulation method program based on spectral width factor. When the panchromatic band simulation method program based on spectral width factor is executed by the processor, it implements the steps of a panchromatic band simulation method based on spectral width factor as described in any one of claims 1 to 8.

Images