Remote sensing image product fusion method and system based on cyclic matching, and storage medium

Abstract

The invention provides a remote sensing image product fusion method and system based on cyclic matching and a storage medium. According to the method, pixel information of a panchromatic wave band with high spatial resolution and a multispectral wave band is enhanced point by point according to a product algorithm, so that information such as spatial geometric textures, edges and levels of ground objects in the image is comprehensively improved, the visual resolution and the computer analysis capability of the multispectral image are effectively improved, and the image quality is improved. And the application effect and capability of the remote sensing image are improved. The spatial resolution of the fused multispectral image is greatly improved, the spatial information is greatly enriched, the geometric texture, spatial details, edge definition and hierarchy of ground features on the image are comprehensively improved, meanwhile, the ground feature spectral features of the original multispectral image and the stability of color display can be kept, and the method has the advantages of being simple in structure and convenient to operate. And the richness of remote sensing image information and the overall quality of the image are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image product fusion enhancement, and more specifically, to a remote sensing image product fusion method, system and storage medium based on cyclic matching. Background Technique

[0002] With the development of multi-platform, multi-sensor, all-weather, multi-temporal and multi-resolution remote sensing technologies, the types of remote sensing images available for all walks of life are becoming increasingly rich. These rich and colorful remote sensing images not only provide a flexible object selection space for remote sensing image application research, but also pose challenges to the pre-processing research of remote sensing image selection, synthesis, correction, enhancement, etc.

[0003] Different remote sensing image data have different basic image characteristics such as spatial resolution, temporal resolution, and spectral resolution, and have different application performances and application potentials in different application fields. Traditional remote sensing image processing focuses on enhancing general image characteristics such as color, texture, and hierarchy of a single image type; remote sensing image fusion processing focuses on integrating the basic characteristics of different images onto a new remote sensing image map, fully tapping its comprehensive application potential, and improving its application performance. In the past two decades or so, as a new direction of remote sensing image processing, remote sensing image fusion technology has made great progress and achieved a series of new results.

[0004] For remote sensing image fusion, its basic requirement is to ensure both the richness and clarity of geometric spatial information such as texture, hierarchy, and details of the fused multi-spectral image, and to ensure the stability of the fine spectral characteristics and color information of the remote sensing image before and after fusion. Generally, the quality of remote sensing images before and after fusion can be quantitatively evaluated from the following aspects:

[0005] First, the richness and brightness of the colors of the images before and after fusion can be measured by band statistical features - maximum value, minimum value, mean value, standard deviation, and correlation indicators between bands - correlation coefficient, covariance, etc.; second, the richness of the overall information of the images before and after fusion can be measured by indicators such as the information entropy of the image bands; third, the hierarchy (edges), details (texture) and clarity of the images before and after fusion can be measured by indicators such as gradient and average gradient; fourth, the consistency and inheritance of the information of the images before and after fusion, that is, the high or low correlation between the fused multi-spectral image and the panchromatic band and the original multi-spectral bands, can reflect the inheritance degree of the spatial information and spectral information of the original image by the fused image. If the correlation between the fused image and the panchromatic image is high, it means that the spatial geometric detail information of the panchromatic image is well injected. If the correlation between the fused image and the original multi-spectral bands is high, it means that the fine spectral information of the multi-spectral image is well retained, thus enabling the evaluation of the quality of image fusion.

[0006] By comparing the differences in the respective indicators of the images before and after fusion, it is possible to analyze the change directions of the spectral (gray level, tone) information, edge (level, difference) information, and texture (detail) information of the images, thereby judging the quality of the fused image.

[0007] For pan-spectral fusion, there are two basic indicators reflecting geometric sharpness and spectral fidelity - the correlation coefficients between the fusion result and the original multi-spectral data. Generally, the higher the correlation coefficient with the multi-spectral data, the better the spectral characteristics are preserved, and the greater the information entropy; the higher the correlation coefficient between the fusion result and the panchromatic band, the greater the average gradient ratio, the higher the geometric sharpness, and the better the image details are preserved.

[0008] From the perspective of the fusion principle, remote sensing image fusion methods can be divided into three categories: fusion methods based on color space theory, information analysis principle, and mathematical operations. Fusion methods based on color space theory apply the main color models such as RGB, CMYK, Lab, IHS, and HSV to image fusion. Among them, the IHS model is a classic image fusion method (Te-ming TU et al., A new look at IHS-look image fusion methods, Information Fusion 2, 2001, 177-18). The color model components of this type of method are relatively independent and can be controlled separately, and can accurately describe color characteristics (Li Lin et al., Comparison of panchromatic and multi-spectral image fusion methods for ZY-3 satellite, Transactions of the Chinese Society of Agricultural Engineering, 30(16), 2014, 157-165). The advantages are that it can quickly and simply perform image fusion and the fusion result well preserves the details of the panchromatic image. The disadvantage is that the fused image usually changes the spectral characteristics of the original satellite remote sensing image. Fusion methods based on the information analysis principle apply information analysis techniques such as spatial filtering, principal component analysis (PCA), Fourier transform (FFT), wavelet transform, Gram-Schimdt transform, curvelet transform, contourlet transform, ridgelet transform, bandelet transform, wedgelet transform, and beamlet transform to image fusion. This type of method injects the detail information of the panchromatic band into the multi-spectral image (Xiao Liang et al., Advances and challenges in multi-source pan-spectral remote sensing image fusion methods, Journal of Image and Graphics, 25(05), 2020). The resulting fused image has good spectral continuity, but the algorithm is complex and usually loses some high-frequency information of the image. Fusion methods based on mathematical operations mainly include ratio method, difference method, weighted superposition, multiple amplification, and four arithmetic mixed operation methods. Among these methods, the classic methods are the Brovey fusion method and the CN fusion method.

[0009] Brovey fusion is a fusion method based on product transformation proposed by American scholar Brovey, also known as color normalization transformation fusion method (Vrabel J.Multispectral Imagery Band Sharpening Study.Photogrammetric Engineering&Remote Sensing,1996,62(9):1075-1083). It performs a product operation on three multispectral bands and a high-spatial-resolution panchromatic image, simplifies the image fusion operation coefficients, and retains the spectral information of the source image to a large extent. However, it also has the defect that only image fusion of three bands can be performed. CN fusion (color normalized transformation) is also known as Energy Subdivision Transform. It uses the high-spatial-resolution panchromatic band to enhance the low-spatial-resolution multispectral bands of the input image. This method only fuses the input bands corresponding to the spectral range included in the fusion image bands, and other input bands are directly output without fusion processing. The Brovey fusion method and the CN fusion method require that the spectral response range of the panchromatic band be consistent or similar to that of the multispectral image. Otherwise, it will cause significant deviations in the local brightness and darkness of the fusion image, change the spectral characteristics of the multispectral image before and after fusion, and cause spectral information distortion of the remote sensing image. Therefore, scholars have carried out various optimizations and improvements on fusion methods such as Brovey and CN. Some scholars have introduced an adaptive optimization method to adjust the weight parameters of each multispectral band and the panchromatic band, and improved the Brovey fusion algorithm through weighted average. While retaining the spectral physical characteristics of the multispectral source image, it can also improve the spatial resolution characteristics of the image (Lin Zhilei et al., Research on the ALI image fusion algorithm based on improved Brovey transform.Remote Sensing Technology and Application,2020,35(4):893-900). Some scholars use the nonlinear spectral data mining characteristics of the kernel principal component analysis (KPCA) algorithm to extract the three principal components KPC1, KPC2, and KPC3 with the largest amount of information in the multispectral image; then use the Brovey algorithm to perform normalized fusion operations on the three principal components and the panchromatic band to make the spatial information and spectral information in the fusion result richer (Ke Hongxia et al., A remote sensing image fusion method based on KPCA and Brovey transform, Journal of Chongqing Jiaotong University (Natural Science Edition),2019,38(2)). Some scholars have introduced the method of wavelet analysis to improve the deficiencies such as the large influence of noise on the fusion image and the excessive retention of sporadic details in the high-resolution image in the Brovey transform (Chen Sijin et al., Using wavelet analysis to improve the Brovey remote sensing image fusion method, Journal of Surveying and Mapping College,2004,21(2)). Since the emergence of fusion methods such as Brovey and CN, the path of their improvement has not stopped and there is no end. Summary of the Invention

[0010] In view of the above problems, the object of the present invention is to provide a remote sensing image product fusion method, system and storage medium based on cyclic matching

[0011] In the first aspect of the present invention, a remote sensing image product fusion method based on cyclic matching is provided, and the method includes the following steps:

[0012] Input the panchromatic band P of the remote sensing image and the multispectral band B of the remote sensing image i ;

[0013] Perform spatial registration on the panchromatic band P and the multispectral band B i to make the geometric spatial positions of the same ground object in the panchromatic band P and the multispectral band B i consistent;

[0014] Use the multispectral band B i to construct an intermediate band I;

[0015] Resample the intermediate band I, the multispectral band B i and the panchromatic band P according to the high-spatial-resolution image, and synthesize the above three images into an image file;

[0016] Calculate the mean and standard deviation of the panchromatic band P, the intermediate band I, and the multispectral band B in the image file respectively i ;

[0017] Perform image feature cyclic matching on the multispectral band B i , the panchromatic band P, and the intermediate band I;

[0018] According to the cyclic matching results of the multispectral B i , the panchromatic band P, and the intermediate band I, use the product fusion principle to reconstruct the multispectral image;

[0019] Perform feature matching on the reconstructed multispectral image and the spectral band B i to obtain the product fusion image and output the fusion result.

[0020] Preferably, the expression of the intermediate band I is:

[0021]

[0022] where n is the number of multispectral bands participating in the fusion, 1≥ω i ≥0,

[0023] Preferably, the step of calculating the mean and standard deviation of the panchromatic band P, the intermediate band I, and the multispectral band B in the image file respectively includes: i :

[0024] Calculate the mean value μ of the panchromatic band P in the image file P , and the calculation formula is:

[0025]

[0026] Calculate the standard deviation σ of the panchromatic band P in the image file P , and the calculation formula is:

[0027]

[0028] Calculate the mean value μ of the multispectral band B i in the image file i , and the calculation formula is:

[0029]

[0030] B i = B i (l, m)

[0031] where B i is any band image of the multispectral band; l is the ordinate of the pixel, m is the abscissa of the pixel; R is the number of rows of the band B i band, C is the number of columns of the band B i band; B i (l, m) is the multispectral band image corresponding to the pixel with ordinate l and abscissa m; P(l, m) is the panchromatic band image corresponding to the pixel with ordinate l and abscissa m;

[0032] Calculate the standard deviation σ of the multispectral band B i in the image file i , and the calculation formula is:

[0033]

[0034] Calculate the mean value μ of the intermediate band I in the image file I ;

[0035]

[0036] Calculate the standard deviation σ of the intermediate band I in the image file I ;

[0037]

[0038] Preferably, performing cyclic matching of the image features of the multispectral band B i , the panchromatic band P, and the intermediate band I can obtain the following matching method:

[0039] Multispectral B i matching method

[0040] Panchromatic band P matching method

[0041] Intermediate band I matching method

[0042] wherein, hmB i byP is the image feature matching result between multispectral band B i and panchromatic band P; hmB i byl is the image feature matching result between multispectral band B i and intermediate band I; hmPbyB i is the image feature matching result between panchromatic band P and multispectral band B i ; hmIbyP is the image feature matching result between panchromatic band P and intermediate band I; hmIbyB i is the image feature matching result between intermediate band I and multispectral band B i ; hmIbyP is the image feature matching result between intermediate band I and panchromatic band P, μ P and μ I are the means of the multispectral band, panchromatic band and intermediate band images respectively, σ P and σ I are the standard deviations of the multispectral band, panchromatic band and intermediate band images respectively.

[0043] Preferably, according to the cyclic matching results of multispectral B i , panchromatic band P, and intermediate band I, using the product fusion principle, the multispectral image is reconstructed as follows:

[0044] According to the cyclic matching results of multispectral B i , panchromatic band P, and intermediate band I, a product operator is constructed, and the panchromatic band P is injected into the multispectral band product fusion algorithm to obtain a clockwise sequential matching fusion scheme and a counterclockwise sequential matching fusion scheme, thereby reconstructing the multispectral image:

[0045]

[0046] wherein, is the fused multispectral image.

[0047] Preferably, the clockwise sequential matching fusion scheme is:

[0048] B-P-I fusion scheme:

[0049] P-I-B fusion scheme:

[0050] I-B-P fusion scheme:

[0051] Preferably, the counterclockwise sequential matching fusion scheme is as follows:

[0052] B-I-P fusion scheme:

[0053] I-P-B fusion scheme:

[0054] P-B-I fusion scheme:

[0055] Preferably, the step of performing feature matching on the reconstructed multispectral image and the original multispectral band B i to obtain a product fusion image specifically includes:

[0056] According to the fusion requirement, select the corresponding scheme from the counterclockwise sequential matching fusion scheme or the clockwise sequential matching fusion scheme for image fusion, and perform feature matching on the calculation result of the fusion image and the original multispectral band to obtain the final image fusion result.

[0057] Preferably, the matching method of the feature matching is as follows:

[0058]

[0059] Wherein, is the multispectral band image after panchromatic image spatial-spectral product fusion.

[0060] The second aspect of the present invention provides a remote sensing image product fusion system, including a memory and a processor. The memory includes a remote sensing image product fusion method program. When the remote sensing image product fusion method program is executed by the processor, it implements the steps of a remote sensing image product fusion method based on loop matching.

[0061] The third aspect of the present invention provides a computer storage medium. The computer storage medium includes a remote sensing image product fusion method program. When the remote sensing image product fusion method program is executed by a processor, it implements the steps of a remote sensing image product fusion method based on loop matching.

[0062] Compared with the prior art, the beneficial effects of the technical solution of the present invention are as follows: The present invention provides a remote sensing image product fusion method, system and storage medium based on cyclic matching. The method of the present invention enhances the pixel information point by point according to the product algorithm for the panchromatic band with high spatial resolution and the multispectral band, so that the spatial geometric texture, edges, levels and other information of the ground objects in the image are comprehensively improved, effectively improving the visual resolution and computer analysis ability of the multispectral image, and improving the application effect and ability of the remote sensing image.

[0063] By performing feature matching processing on the multispectral band and the panchromatic band, the present invention enhances the spatial information amount, clarity and layering of the ground objects in the multispectral image while retaining the rich spectral characteristics and color display of the original multispectral image, without changing the image spectral characteristics and display effects of typical ground objects such as water bodies, vegetation, soil, buildings, etc., effectively improving the overall quality of the multispectral image and the potential of remote sensing image mapping applications and the ability of classification applications.

[0064] After the fusion of the present invention, the spatial resolution of the multispectral image is greatly improved, the spatial information is greatly enriched, and the geometric texture, spatial details, edge clarity and layering of the ground objects on the image are comprehensively improved. At the same time, it can maintain the stability of the ground object spectral characteristics and color display of the original multispectral image, greatly improving the richness of remote sensing image information and the overall quality of the image. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a flowchart of a remote sensing image product fusion method based on cyclic matching described in Embodiment 1.

[0066] Figure 2 It is a schematic diagram of sequential clockwise matching.

[0067] Figure 3 It is a schematic diagram of sequential counterclockwise matching.

[0068] Figure 4 It is a panchromatic band (P) image (2-meter resolution).

[0069] Figure 5 It is a multispectral NRG standard false color image (8-meter resolution).

[0070] Figure 6 It is a multispectral RGB true color composite image (8-meter resolution).

[0071] Figure 7 It is an intermediate band I image.

[0072] Figure 8 It is a true color composite map of the fused multispectral R'G'B' band (2-meter resolution).

[0073] Figure 9 It is the fused multi-spectral N’R’G’ band standard false color composite image (2-meter resolution). Specific implementation mode

[0074] In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation modes. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0075] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0076] Embodiment 1

[0077] As Figure 1 shown, this embodiment discloses a remote sensing image product fusion method based on circular matching. The method includes the following steps:

[0078] S1: Input the panchromatic band P of the remote sensing image and the multi-spectral band B of the remote sensing image i .

[0079] S2: Perform spatial registration on the panchromatic band P and the multi-spectral band B i so that the geometric spatial positions of the same ground object in the panchromatic band P and the multi-spectral band B i are consistent.

[0080] S3: Use the multi-spectral band B i to construct an intermediate band I; the expression of the intermediate band I is:

[0081]

[0082] where n is the number of multi-spectral bands participating in the fusion, B i is any multi-spectral band, n≥1, 1≥ω i ≥0, Generally, ω i =1 / n.

[0083] S4: Resample the intermediate band I, the multi-spectral band B i and the panchromatic band P according to the high-spatial-resolution image, and synthesize the above three images into an image file.

[0084] S5: Calculate the mean and standard deviation of the panchromatic band P, the intermediate band I, and the multi-spectral band B i in the image file respectively.

[0085] The mean value μ of the panchromatic band P in the image file P , and the calculation formula is:

[0086]

[0087] Calculate the standard deviation σ of the panchromatic band P in the image file P , and the calculation formula is:

[0088]

[0089] The mean value μ of the multispectral band B in the image file i i , and the calculation formula is:

[0090]

[0091] Where B i is the image of any band of the multispectral band; l is the ordinate of the pixel, m is the abscissa of the pixel; R is the number of rows of band B i band, C is the number of columns of band B i band;

[0092] The standard deviation σ of the multispectral band B in the image file i i , and the calculation formula is:

[0093]

[0094] The mean value μ of the intermediate band I in the image file I ;

[0095]

[0096] The standard deviation σ of the intermediate band I in the image file I ;

[0097]

[0098] S6: Perform image feature cyclic matching on the multispectral band B i , the panchromatic band P, and the intermediate band I. Through image feature cyclic matching, multiple matching methods can be obtained, specifically including:

[0099] Multispectral B i matching method

[0100] Panchromatic band P matching method

[0101] Intermediate band I matching method

[0102] ​​Among them, hmB i byP and hmB i byI are respectively the image feature matching results of the multispectral band B i with the panchromatic band P and the intermediate band I; hmPbyB i and hmPbyI are respectively the image feature matching results of the panchromatic band P with the multispectral band B i and the intermediate band I; hmIbyB i and hmIbyP are respectively the image feature matching results of the intermediate band I with the multispectral band B i and the panchromatic band P.

[0103] S7: According to the cyclic matching results of the multispectral B i the panchromatic band P, and the intermediate band I, use the product fusion principle to reconstruct the multispectral image.

[0104] It should be noted that in this embodiment, according to the cyclic matching results of the multispectral B i the panchromatic band P, and the intermediate band I, multiple product operators can be constructed, and the panchromatic band injection multispectral band product fusion algorithm is adopted A total of the following product fusion schemes are obtained, including:

[0105] (1) Clockwise sequential matching fusion scheme:

[0106] ① B-P-I fusion scheme:

[0107] ② P-I-B fusion scheme:

[0108] ③ I-B-P fusion scheme:

[0109] (2) Counterclockwise sequential matching fusion scheme:

[0110] ④ B-I-P fusion scheme:

[0111] ⑤ I-P-B fusion scheme:

[0112] ⑥ P-B-I fusion scheme:

[0113] Among them, is the fused multispectral image.

[0114] S8: Match the features of the reconstructed multispectral image with the spectral band B i to obtain the product fusion image and output the fusion result.

[0115] It should be noted that in this embodiment, a certain scheme is selected from the fusion scheme described in S7 for image fusion, and the calculation results of the fused image are feature-matched with the original multi-spectral bands to obtain the final image fusion result. The matching method is as follows:

[0116]

[0117] Among them, is the multi-spectral band image after the panchromatic image is fused by the space-spectrum product.

[0118] The method described in this embodiment is applicable to fusing the panchromatic band and the multi-spectral band of remote sensing images to enhance the spatial resolution and geometric texture information of the multi-spectral image. While greatly improving the spatial resolution and geometric detail information of the multi-spectral image, it can achieve a high degree of consistency in the spectral characteristics and color information of the original multi-spectral image and the fused multi-spectral image. The method described in this embodiment has no limitation on the number of bands of the fused multi-spectral image.

[0119] The method described in this embodiment has the following advantages:

[0120] In this embodiment, the method applies a linear combination of multi-spectral bands to simulate the low-resolution panchromatic band as the intermediate band I; the multi-spectral band B i , the panchromatic band P, and the intermediate band I are subjected to cyclic image feature matching, and the ground feature information of the multi-spectral image and the panchromatic image is fused based on the product algorithm principle to reconstruct the fused image result, so as to realize the enhancement of the spatial information and the true preservation of the spectral information before and after the multi-spectral image fusion. Compared with the original image, the geometric details, textures, edges, levels and other spatial information of the ground objects in the fused image are greatly enriched, and compared with the original multi-spectral image, in the fused multi-spectral image, the spectral characteristics and color displays of ground objects such as water bodies, vegetation, bare ground, and buildings remain stable. The fusion function expression in the method described in this embodiment is concise and clear, and the calculation is fast and efficient.

[0121] The method described in this embodiment is applicable to fusing all panchromatic bands and multi-spectral satellite remote sensing images with near-infrared, red, green, and blue bands to improve the spatial accuracy of the multi-spectral bands. At the same time, it is applicable to the fusion between different types of images such as optical images and microwave images to only achieve specific application purposes. The method described in this embodiment only depends on the inherent characteristics of the image data and has universality.

[0122] The number of bands of the fused target image in this embodiment only needs to be greater than or equal to 1, that is: multispectral images of one band can be fused, or fused target images of 2 bands, 3 bands, 4 bands or even more bands can be processed. The number of bands of the fused target image for which fusion is performed determines the number of bands of the resulting image fusion output. The method in this embodiment itself has no limit on the number of bands of the fused target image and is open-ended.

[0123] As a specific embodiment, the method in this embodiment will be described below with specific examples:

[0124] 1. Input the remote sensing image.

[0125] Open a domestic GF6 remote sensing image with panchromatic band (P), blue band (B), green band (G), red band (R), and near-infrared (N) (the case image is taken from GF6_PMS_E112_8_N23_2_20210118_L1A1120072629). Figures 4 - 6 They are respectively the panchromatic band image (2-meter resolution), the multispectral NRG standard false-color band image (8-meter resolution), and the multispectral RGB true-color image (8-meter resolution) (the effect image stretched by 1% according to the ENVI default settings).

[0126] 2. Use the 4 multispectral band images in ENVI to construct the intermediate band I, and its band operation expression is I = (1.0 * b1 + b2 + b3 + b4) / 4, where b1, b2, b3, and b4 are the blue band (B), green band (G), red band (R), and near-infrared (N) respectively. The intermediate band I of the calculated fused target image is as Figure 7 shown.

[0127] 3. Resample the intermediate band I and the multispectral bands B, G, R, and N in the ENVI software according to the high-spatial-resolution fused source image, and form an image file with the panchromatic band P, and then calculate the image feature statistical parameters such as the mean μ and standard deviation σ of each band image. The basic feature statistical parameters of each band image are shown in Table 1.

[0128] Table 1 Basic feature statistical parameter table of each band image

[0129] Imaging band Mean / μ Standard deviation / σ Panchromatic band P 820.062867 352.608833 Intermediate band I 895.473494 348.621005 Blue band B 758.198332 195.501797 Green band G 826.7832 297.752981 Red band R 789.444254 415.289068 Near-infrared N 1205.967693 617.606258

[0130] 4. Reconstruct the fused multispectral image

[0131] ① The reconstructed and fused blue band (B’) image has the operation expression uint((0.988691*(b2 - 820.062867)+895.473494)*(1.803609*(b1 - 758.198332)+820.062867) / b3 + 0.5), where b1 is the blue band (B), b2 is the panchromatic band P, and b3 is the intermediate band I. The fused blue band (B’) image (2-meter resolution) is obtained through calculation.

[0132] ② The reconstructed and fused green band (G’) image has the operation expression uint((0.988691*(b2 - 820.062867)+895.473494)*(1.184233*(b1 - 826.7832)+820.062867) / b3 + 0.5), where b1 is the green band (G), b2 is the panchromatic band P, and b3 is the intermediate band I. The fused green band (G’) image (2-meter resolution) is obtained through calculation.

[0133] ③ The reconstructed and fused red band (R’) image has the operation expression uint((0.988691*(b2 - 820.062867)+895.473494)*(0.849068*(b1 - 789.444254)+820.062867) / b3 + 0.5), where b1 is the red band (R), b2 is the panchromatic band P, and b3 is the intermediate band I. The fused red band (R’) image (2-meter resolution) is obtained through calculation.

[0134] ④ The reconstructed and fused near-infrared (N’) image has the operation expression uint((0.988691*(b2 - 820.062867)+895.473494)*(0.570928*(b1 - 1205.967693)+820.062867) / b3 + 0.5), where b1 is the near-infrared band (N), b2 is the panchromatic band P, and b3 is the intermediate band I. The fused near-infrared (N’) image (2-meter resolution) is obtained through calculation.

[0135] The true-color image synthesized from the fused red band (R’), green band (G’), and blue band (B’) according to the red, green, and blue channels is as Figure 8 shown; the standard false-color image synthesized from the fused near-infrared band (N’), red band (R’), and green band (G’) according to the red, green, and blue channels is as Figure 9 shown.

[0136] In this embodiment, band data statistical analysis was carried out on three types of remote sensing images: the original GF6 multispectral image, the multispectral image fused by the method described in this embodiment, and the multispectral image fused by the Gram-Schmidit method. The comparison of the statistical characteristic parameters of the image bands is shown in Table 2.

[0137] Table 2 Comparison of the statistical characteristic parameters of the original multispectral image, the image fused by this method, and the GS-fused image

[0138]

[0139]

[0140] As can be seen from the data in the table, compared with the original multispectral image, the spatial resolution of the multispectral image fused by the method of the present invention has increased from the original 8 meters to 2 meters, and the spatial accuracy of the image has been greatly improved; the average value of the correlation coefficients of each band has decreased from 0.816 to 0.813, the data structure of each band has been further optimized, and the total amount of spectral information of the image has increased; the fused image combines the image information of the original multispectral band and the panchromatic band at the same time, and the information consistency index has increased from 0.934 to 0.940; the information entropy of the multispectral band has slightly decreased, but the gradient information of each multispectral band has been greatly enhanced, and the ground object spatial information of the fused multispectral image is more abundant, which is conducive to the development of various image applications such as ground object recognition, interpretation, and analysis. As can be seen from the data in the table, compared with the multispectral image fused by the Gram-Schmidt method, the correlation coefficient between the multispectral images fused by the method of the present invention is lower, and the information redundancy is smaller. The two image fusion methods have their own characteristics.

[0141] The method, system, medium and device for remote sensing image product fusion based on cyclic matching described in this embodiment are mainly aimed at remote sensing images with panchromatic bands and multispectral bands such as near-infrared, red, green, and blue bands. First, a low-resolution panchromatic band is simulated using the multispectral image to be fused as the intermediate band I for fusion. Then, based on the principle of image feature matching, cyclic feature matching is performed on the intermediate band I, the panchromatic band P, and the multispectral band Bi to form multiple fusion schemes and construct the image fusion product operator for each multispectral band. Finally, using the product algorithm principle, product fusion is performed on the multispectral bands to obtain the fused multispectral remote sensing image. The method described in this embodiment is applicable to fusing the panchromatic band and the multispectral band to enhance the spatial geometry, texture, edges, levels, and other information of the ground objects in the multispectral image, improve its image clarity and spatial resolution. At the same time, the fused image can highly maintain the stability of the spectral characteristics and color display of various ground objects in the original multispectral image. The method described in this embodiment has a solid theoretical basis, clear physical meaning, a wide range of application objects, is easy to operate, and has high operation efficiency. The fused image has distinct colors, rich information, stable spectral information, and is easy for visual and automatic classification. Especially in the context of the rapid development of high-resolution satellite remote sensing, it has a huge promoting effect on promoting the popularization and application of domestic high-resolution images in various industries at home and abroad.

[0142] Embodiment 2

[0143] This embodiment discloses a remote sensing image product fusion system, including a memory and a processor. The memory includes a remote sensing image product fusion method program, and when the remote sensing image product fusion method program is executed by the processor, it realizes the steps of a method for remote sensing image product fusion based on cyclic matching described in Embodiment 1.

[0144] Embodiment 3

[0145] This embodiment discloses a computer storage medium, which includes a remote sensing image product fusion method program. When the remote sensing image product fusion method program is executed by a processor, it realizes the steps of a method for remote sensing image product fusion based on cyclic matching described in Embodiment 1.

[0146] In 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 the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the displayed or discussed components can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0147] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0148] In addition, each functional unit in the embodiments of the present invention can be all integrated in a processing unit, or each unit can be separately used as a unit, or two or more units can be integrated in a unit; the above-mentioned integrated units can be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0149] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: removable storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical disks and other various media that can store program codes.

[0150] Alternatively, if the above-mentioned integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present invention. And the foregoing storage medium includes: removable storage devices, ROM, RAM, magnetic disks, or optical disks and other various media that can store program codes.

Claims

1. A remote sensing image product fusion method based on cyclic matching, characterized in that: The method comprises the following steps: Input remote sensing image panchromatic band P and remote sensing image multispectral band B i ; Combine the panchromatic band P with the multispectral band B i Perform spatial registration to align the panchromatic band P with the multispectral band B i The geometrical spatial position of the same feature is consistent; Using multispectral band B i Construct Intermediate Band I; Intermediate band I, multispectral band B i The panchromatic band P is resampled according to the high spatial resolution image, and the above three images are combined into one image file; Calculate the panchromatic band P, intermediate band I, and multispectral band B in the image file respectively i The mean and standard deviation of Multispectral band B i , panchromatic band P, and intermediate band I for image feature cyclic matching; According to multispectral band B i , the cyclic matching results of the panchromatic band P and the intermediate band I, and the multispectral image is reconstructed using the product fusion principle; The reconstructed multispectral image is combined with the original multispectral band B i Perform feature matching, obtain the product fusion image, and output the fusion result.

2. The remote sensing image product fusion method based on cyclic matching according to claim 1, characterized in that: The expression of the intermediate band I is: Where n is the number of multispectral bands involved in fusion, 1 ≥ ω i ≥0, 3. The remote sensing image product fusion method based on cyclic matching according to claim 2 is characterized in that: The panchromatic band P, intermediate band I and multispectral band B in the image file are calculated respectively. i The mean and standard deviation of , including: Calculate the mean μ of the panchromatic band P in the image file P ; Calculate the standard deviation σ of the panchromatic band P in the image file P ; Calculate the multispectral band B in the image file i The mean μ i , the calculation formula is: B i =B i (l,m) Calculate the multispectral band B in the image file i The standard deviation σ i , the calculation formula is: Calculate the mean μ of the intermediate band I in the image file I ; Calculate the standard deviation σ of the intermediate band I in the image file I ; Among them, B i is any band image of the multispectral band; l is the ordinate of the pixel, m is the abscissa of the pixel; R is the band B i The number of band rows, C is band B i The number of columns of the band; B i (l,m) is the multispectral band image corresponding to the pixel with ordinate l and abscissa m; P(l,m) is the panchromatic band image corresponding to the pixel with ordinate l and abscissa m.

4. The remote sensing image product fusion method based on cyclic matching according to claim 3 is characterized in that: The multispectral band B i , panchromatic band P, and intermediate band I are used to perform image feature cyclic matching, and the following matching methods can be obtained: Multispectral B i Matching method Panchromatic band P matching method Intermediate Band I Matching Method Among them, hmB i byP is multispectral band B i Matching result with the image features of the panchromatic band P; hmB i byI is multispectral band B i Matching results with the image features of the intermediate band I; hmPbyB i The panchromatic band P and the multispectral band B i hmPbyI is the image feature matching result of the panchromatic band P and the intermediate band I; hmIbyB i Intermediate band I and multispectral band B i The image feature matching result hmIbyP is the image feature matching result of the intermediate band I and the panchromatic band P. μ P , μ I are the mean values ​​of the multispectral band, panchromatic band and intermediate band images respectively, σ P , σ I They are the standard deviations of the multispectral, panchromatic, and intermediate band images respectively.

5. The remote sensing image product fusion method based on cyclic matching according to claim 4 is characterized in that: According to the multi-spectral B i , panchromatic band P, and intermediate band I, and reconstruct the multispectral image using the product fusion principle, specifically: According to multispectral B i The product operator is constructed by the cyclic matching results of the panchromatic band P and the intermediate band I. The panchromatic band P is injected into the multispectral band product fusion algorithm to obtain the clockwise matching fusion scheme and the counterclockwise matching fusion scheme, thereby reconstructing the multispectral image: in, The fused multispectral image.

6. The method for product fusion of remote sensing images based on cyclic matching according to claim 5, characterized in that: The clockwise matching fusion scheme is: BPI Fusion Solution: PIB Fusion Solution: IBP Fusion Solution: The counterclockwise matching fusion scheme is: BIP fusion solution: IPB fusion solution: PBI Fusion Solution:

7. The remote sensing image product fusion method based on cyclic matching according to claim 6, characterized in that: The reconstructed multispectral image is combined with the original multispectral band B i Perform feature matching to obtain product fusion images, including: According to the fusion needs, the corresponding scheme is selected from the counterclockwise matching fusion scheme or the clockwise matching fusion scheme for image fusion, and the fused image calculation result is feature matched with the original multispectral band to obtain the final image fusion result.

8. The remote sensing image product fusion method based on cyclic matching according to claim 7, characterized in that: The matching method of the feature matching is as follows: in, It is a multispectral band image formed by the spatial-spectral product fusion of the panchromatic image.

9. A remote sensing image product fusion system, characterized in that: It comprises a memory and a processor, wherein the memory comprises a remote sensing image product fusion method program, and when the remote sensing image product fusion method program is executed by the processor, the steps of a remote sensing image product fusion method based on cyclic matching as described in any one of claims 1 to 8 are implemented.

10. A computer storage medium, characterized in that: The computer storage medium includes a remote sensing image product fusion method program. When the remote sensing image product fusion method program is executed by a processor, the steps of a remote sensing image product fusion method based on cyclic matching as described in any one of claims 1 to 8 are implemented.

Images