A green vegetation enhancement method based on panchromatic and multispectral image fusion

Abstract

The present invention provides a green vegetation enhancement method for panchromatic multispectral image fusion, firstly, the panchromatic image and the multispectral image are roughly aligned based on a rational function model; then the roughly aligned multispectral image and the panchromatic image are subjected to small-surface differential alignment; then the multispectral image subjected to small-surface differential alignment based on the normalized vegetation index is subjected to green vegetation enhancement; then the panchromatic image is fused with the multispectral image subjected to green vegetation enhancement to obtain a final fusion result of green vegetation enhancement. The present invention enhances green vegetation during the fusion process, improves the fusion effect, solves the problem of dark green vegetation in the panchromatic multispectral fusion image, and is conducive to the improvement of the final mapping effect and subsequent interpretation and processing.

Description

Technical Field

[0001] The invention relates to the field of remote sensing image processing, and in particular to a green vegetation enhancement method for panchromatic and multispectral image fusion. Background Art

[0002] When the spatial resolution is improved, the spectral resolution will inevitably decrease relatively. Conversely, when the spectral resolution is improved, the spatial resolution will also decrease. Because for remote sensing cameras, the detector can only maintain the signal-to-noise ratio if it receives enough energy. The energy received by a multi-spectral single band is weak, so the detector size can only be increased, resulting in the spatial resolution of the multi-spectral being lower than that of the panchromatic. Although the spatial resolution of the panchromatic is higher than that of the multi-spectral, in order to increase the intensity of the received energy, its spectral range is wider, and its spectral resolution is lower than that of the multi-spectral. Under equal conditions, improving the spatial resolution will inevitably reduce the spectral resolution, which is an irreconcilable contradiction.

[0003] Therefore, mainstream optical satellites have fully considered this factor during their design, and will simultaneously design panchromatic images with high spatial resolution and multispectral images with high spectral resolution. This also provides a hardware foundation for ground-based remote sensing images with both high spatial resolution and high spectral resolution. The process of acquiring such images is panchromatic and multispectral image fusion. Fused images are also important image products of optical satellites, which are conducive to later interpretation and processing.

[0004] In the existing fusion methods, since satellite images are quantized at high bits, the images appear darker, especially in vegetation areas, which generally appear dark green, which is quite different from the bright green that people are familiar with, and often affects the interpretation of the interpreter. In other words, there is a problem of dark green vegetation in the existing full-color multispectral fusion images. Summary of the invention

[0005] The present invention proposes a green vegetation enhancement method for panchromatic multispectral image fusion, which enhances the green vegetation during the fusion process, is beneficial to the improvement of the final mapping effect and the subsequent interpretation and processing, and solves the technical problem of dark green vegetation in the prior art.

[0006] In order to solve the above technical problems, the present invention provides a green vegetation enhancement method by fusion of panchromatic and multispectral images, comprising:

[0007] S1: Rough registration of panchromatic and multispectral images based on rational function model;

[0008] S2: Perform small-surface differential alignment of the roughly registered multispectral image and the panchromatic image;

[0009] S3: Green vegetation enhancement based on multispectral images after small-surface differential calibration of the NDVI;

[0010] S4: Fuse the panchromatic image with the multispectral image after green vegetation enhancement to obtain the final green vegetation enhanced fusion result.

[0011] In one embodiment, step S1 includes:

[0012] S1.1: Calculate the overlapping area of ​​panchromatic image and multispectral image by rational function model:

[0013] S1.2: Read the panchromatic image of the overlapping area and perform differential geo-registration of the multispectral image based on the panchromatic image of the overlapping area to achieve coarse registration.

[0014] In one embodiment, step S1.2 comprises:

[0015] S1.2.1: Read the panchromatic image into the memory according to the calculated overlapping area, divide the panchromatic image read into the memory into uniform grid points, and obtain the image coordinates (s, l) of the corresponding panchromatic image;

[0016] S1.2.2: Calculate the object coordinates (B, L, H) of each grid point using the rational function model of the panchromatic image;

[0017] S1.2.3: The object-space coordinates of the grid points are calculated by the rational function model of the multispectral image to obtain the image-space coordinates (s′, l′) of each grid point in the multispectral image;

[0018] S1.2.4: Construct a transformation model from the image coordinates (s, l) of the panchromatic image to the image coordinates (s′, l′) of the multispectral image:

[0019] (s, l) = g(s′, l′);

[0020] S1.2.5: Resample the multispectral image to the panchromatic image coordinates through the above transformation model mapping to obtain a multispectral image that is roughly aligned with the panchromatic image.

[0021] In one embodiment, step S2 includes:

[0022] S2.1: Construct a virtual full-color image. The formula is as follows:

[0023] Mss p =r1*Band B +r2*Band G +r3*Band R +r4*Band NIR

[0024] In the above formula, Mss pis a virtual panchromatic image that is closer to the full color after conversion. r1, r2, r3, and r4 are the band range ratios of bands B, G, R, and NIR to the panchromatic image. B 、Band G 、Band R , Baud NIR They represent the blue light band, green light band, red light band and near-infrared band in the virtual multispectral image respectively;

[0025] S2.2: By selecting uniform grid points on the panchromatic image, an m×n matching window is established with the grid points as the center, and a matching window of the same size and shape is established on the virtual panchromatic image to calculate the correlation coefficient. The formula is as follows:

[0026]

[0027] Each correlation coefficient corresponds to a point, and the point corresponding to the largest correlation coefficient is the matching point with the same name. m, n represent the row and column size of the matching window, i, j represent the row and column number in the matching window, and g i，j , g i+r，j+c Respectively represent the grayscale values ​​in the matching window of the M-band and N-band images, c and r are the coordinate differences of the N-band image with respect to the M-band image in rows and columns;

[0028] S2.3: Obtain four points above, below, left and right of the point corresponding to the maximum correlation coefficient, and obtain the correlation coefficients corresponding to the four points, perform Gaussian surface fitting according to the correlation coefficients corresponding to the four points, and obtain the coordinates of the same-name points of the full-color image and the virtual full-color image;

[0029] S2.4: Based on the coordinates of the same-name points of the obtained panchromatic image and the virtual panchromatic image, the multispectral image after the rough registration is fine-tuned.

[0030] In one embodiment, step S3 includes:

[0031] S3.1: Calculate the NDVI index using the following formula:

[0032]

[0033] In the above formula, NIR is the pixel value of the near infrared band, and R is the pixel value of the red light band;

[0034] S3.2: Extract the green vegetation area according to the set threshold, as shown below:

[0035]

[0036] In the above formula, t is the threshold value;

[0037] S3.3: Enhance green vegetation. The formula is as follows:

[0038]

[0039] In the above formula, α is the enhancement ratio, (1+NDVI-t)*α is the enhancement coefficient, Band G It is the green band of the multispectral image.

[0040] In one implementation, the fusion method used in step S4 is brightness transformation based on smoothing filtering, and the formula is:

[0041]

[0042] Among them, MS is the multispectral image after green enhancement by S3, PAN is the panchromatic image after small-surface differential alignment by S2, PAN′ is the low-resolution panchromatic image after small-surface differential alignment by S2, and Fusion is the fusion result.

[0043] The above one or more technical solutions in the embodiments of the present application have at least one or more of the following technical effects:

[0044] The present invention provides a green vegetation enhancement method for panchromatic multispectral image fusion, firstly, the panchromatic image and the multispectral image are roughly aligned based on a rational function model; then the roughly aligned multispectral image and the panchromatic image are subjected to small-surface differential alignment; then the multispectral image subjected to small-surface differential alignment based on the normalized vegetation index is subjected to green vegetation enhancement; then the panchromatic image is fused with the multispectral image subjected to green vegetation enhancement to obtain a final fusion result of green vegetation enhancement. The present invention enhances green vegetation during the fusion process, improves the fusion effect, solves the problem of dark green vegetation in the panchromatic multispectral fusion image, and is conducive to the improvement of the final mapping effect and subsequent interpretation and processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a flow chart of a green vegetation enhancement method of panchromatic multispectral image fusion of the present invention.

[0047] Figure 2 It is a comparison diagram of the effects before and after the green vegetation enhancement of the present invention. DETAILED DESCRIPTION

[0048] The inventors of this application have discovered through extensive research and practice that fused images are important image products of optical satellites, which are conducive to later interpretation and processing. In addition, because satellite images are high-bit quantized, the images are visually darker, especially in vegetation areas, which generally appear dark green, which is quite different from the bright green that people are familiar with, and often affects the interpretation of the interpreter. The human eye is particularly sensitive to green, and green vegetation that conforms to the characteristics of natural images can greatly improve the output effect and interpretation accuracy. Therefore, the enhancement of green vegetation is also of great significance, especially in fused images with richer usage scenarios.

[0049] In summary, in order to improve the effect and interpretation accuracy of panchromatic multispectral image fusion, the present invention specifically proposes a green vegetation enhancement method for panchromatic multispectral image fusion, which completes the green vegetation enhancement in the fusion process, making the green vegetation information in the fused image more prominent.

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In order to solve the problem of dark green vegetation in panchromatic multispectral fusion images, the present invention enhances green vegetation during the fusion process, which is beneficial to the improvement of the final image effect and subsequent interpretation and processing. The present invention proposes a green vegetation enhancement method for panchromatic multispectral image fusion, comprising:

[0052] S1: Rough registration of panchromatic and multispectral images based on rational function model;

[0053] S2: Perform small-surface differential alignment of the roughly registered multispectral image and the panchromatic image;

[0054] S3: Green vegetation enhancement based on multispectral images after small-surface differential calibration of the NDVI;

[0055] S4: Fuse the panchromatic image with the multispectral image after green vegetation enhancement to obtain the final green vegetation enhanced fusion result.

[0056] Specifically, see Figure 1, which is a flow chart of a green vegetation enhancement method for panchromatic and multispectral image fusion. The rational function model in step S1 is RFM. Due to the influence of satellite platform vibration, calibration random error, etc., the multispectral image and the panchromatic image after coarse registration still need to be precisely registered. Therefore, the multispectral image and the panchromatic image after coarse registration are precisely registered by small facet differential registration through step S2 to eliminate potential random errors.

[0057] Through the aforementioned steps S1 to S2, accurately aligned panchromatic images and multispectral images can be obtained. At this time, the multispectral image can be enhanced for green vegetation. Finally, through step S4, the panchromatic image and the multispectral image after green vegetation enhancement are quickly fused to obtain the final fusion result after green vegetation enhancement. The fusion can use high-fidelity fusion methods such as SFIM (brightness transformation based on smoothing filter), PSD (panchromatic spectral decomposition), GS (orthogonal transformation), etc.

[0058] In one embodiment, step S1 includes:

[0059] S1.1: Calculate the overlapping area of ​​panchromatic image and multispectral image by rational function model:

[0060] S1.2: Read the panchromatic image of the overlapping area and perform differential geo-registration of the multispectral image based on the panchromatic image of the overlapping area to achieve coarse registration.

[0061] Specifically, the rational function model (RFM) can be used to determine the geographical range of panchromatic and multispectral images, including the geodetic latitude B, geodetic longitude L, and geodetic height H (i.e., the object coordinate BLH), where H is the average height. The RFM model can construct the relationship between the object coordinates (B, L, H) and the image coordinates (s, l), as shown in the following formula:

[0062]

[0063] In the above formula, (U, V, W) is the normalized coordinate of the object coordinate (B, L, H):

[0064]

[0065] Among them, LonOff, LatOff, and HeiOff are the translation values ​​of the object coordinates, and LonScale, LatScale, and HeiScale are the scaling values ​​of the object coordinates.

[0066] The numerator and denominator of the polynomial in formula (1) are expressed as follows:

[0067]

[0068] Among them, a i , b i , c i , d i (i=1, 2, ..., 20) are rational polynomial coefficients (RPCs). In general, b1 and d1 are both 1.

[0069] The RFM model can be used to calculate the respective geographical ranges of panchromatic and multispectral images. At this time, the geographical ranges of the two are in a unified geographical coordinate system, and the geographical range of the overlapping area can be calculated. The geographical range of the overlapping area is calculated back to their respective image coordinate systems through the RFM model to obtain the image coordinate range of the overlapping area.

[0070] Next, the panchromatic image of the overlapping area can be read and used as a reference to complete the multispectral differential geographic registration, i.e., coarse registration.

[0071] In one embodiment, step S1.2 comprises:

[0072] S1.2.1: Read the panchromatic image into the memory according to the calculated overlapping area, divide the panchromatic image read into the memory into uniform grid points, and obtain the image coordinates (s, l) of the corresponding panchromatic image;

[0073] S1.2.2: Calculate the object coordinates (B, L, H) of each grid point using the rational function model of the panchromatic image;

[0074] S1.2.3: The object-space coordinates of the grid points are calculated by the rational function model of the multispectral image to obtain the image-space coordinates (s′, l′) of each grid point in the multispectral image;

[0075] S1.2.4: Construct a transformation model from the image coordinates (s, l) of the panchromatic image to the image coordinates (s′, l′) of the multispectral image:

[0076] (s, l) = g(s′, l′) (4)

[0077] S1.2.5: Resample the multispectral image to the panchromatic image coordinates through the above transformation model mapping to obtain a multispectral image that is roughly aligned with the panchromatic image.

[0078] Specifically, the panchromatic image is read into the memory according to the calculated overlapping area, and the multispectral image can be differentially geo-referenced based on this area. In S1.2.4, g() represents a transformation model, which can be a linear model, a radial transformation model, a perspective transformation model, etc. Considering the accuracy and efficiency, the present embodiment uses the perspective transformation model.

[0079] In the mapping process, the multispectral image needs to be resampled. One of the resampling methods such as nearest neighbor resampling, bilinear resampling, and bicubic resampling can be selected. Based on efficiency and accuracy considerations, this embodiment uses bilinear resampling.

[0080] In one embodiment, step S2 includes:

[0081] S2.1: Construct a virtual full-color image. The formula is as follows:

[0082] Mss p =r1*Band B +r2*Band G +r3*Band R +r4*Band NIR (5)

[0083] In the above formula, Mss p is the converted virtual panchromatic image that is closer to the panchromatic image (compared to the multispectral image, the virtual panchromatic image is closer to the panchromatic image). r1, r2, r3, and r4 are the band range ratios of bands B, G, R, and NIR to the panchromatic image. B 、Band G 、Band R 、Band NIR They represent the blue light band, green light band, red light band and near-infrared band in the virtual multispectral image respectively;

[0084] S2.2: By selecting uniform grid points on the panchromatic image, an m×n matching window is established with the grid points as the center, and a matching window of the same size and shape is established on the virtual panchromatic image to calculate the correlation coefficient. The formula is as follows:

[0085]

[0086] Each correlation coefficient corresponds to a point, and the point corresponding to the largest correlation coefficient is the matching point with the same name. m, n represent the row and column size of the matching window, i, j represent the row and column number in the matching window, and g i，j , g i+r，j+c Respectively represent the grayscale values ​​in the matching window of the M-band and N-band images, c and r are the coordinate differences of the N-band image with respect to the M-band image in rows and columns;

[0087] S2.3: Obtain four points above, below, left and right of the point corresponding to the maximum correlation coefficient, and obtain the correlation coefficients corresponding to the four points, perform Gaussian surface fitting according to the correlation coefficients corresponding to the four points, and obtain the coordinates of the same-name points of the full-color image and the virtual full-color image;

[0088] S2.4: Based on the coordinates of the same-name points of the obtained panchromatic image and the virtual panchromatic image, the multispectral image after the rough registration is fine-tuned.

[0089] Specifically, since the multispectral range is relatively narrow, its correlation with the panchromatic image is not high, so it is necessary to construct a virtual panchromatic image to better match the panchromatic image. The constructed formula is shown in (5), where r1, r2, r3, and r4 can be obtained by looking up the design parameters of the remote sensing satellite.

[0090] In S2.2, for multispectral images that have been registered, c=r=0; for images that have not been registered, the sizes of c and r are determined according to the camera design values.

[0091] Since the image is discrete data, the coordinates of the same-name points obtained by calculating the correlation coefficient in step S2.2 are also discrete and not continuous. Therefore, in step S2.3, Gaussian surface fitting is performed according to the correlation coefficients corresponding to the four points to obtain continuous coordinates of the same-name points, which are the coordinates of the same-name points of the full-color image and the virtual full-color image.

[0092] After obtaining the coordinates of the same-name points of the continuous panchromatic image and the virtual panchromatic image, the multispectral image after the rough alignment can be fine-tuned. The matching method is the same as step S1. The fine-tuning can also use a linear model, a radial transformation model, a perspective transformation model, etc. The resampling method can also use one of the nearest neighbor resampling, bilinear resampling, bicubic resampling, etc. The same considerations as the above steps, this embodiment uses perspective transformation and bilinear resampling.

[0093] In one embodiment, step S3 includes:

[0094] S3.1: Calculate the NDVI index using the following formula:

[0095]

[0096] In the above formula, NIR is the pixel value of the near infrared band, and R is the pixel value of the red light band;

[0097] S3.2: Extract the green vegetation area according to the set threshold, as shown below:

[0098]

[0099] In the above formula, t is the threshold value;

[0100] S3.3: Enhance green vegetation. The formula is as follows:

[0101]

[0102] In the above formula, α is the enhancement ratio, (1+NDVI-t)*α is the enhancement coefficient, Band G It is the green band of the multispectral image.

[0103] Specifically, in S3.1, the NDVI index quantifies vegetation by measuring the difference between near-infrared (strongly reflected by vegetation) and red light (absorbed by vegetation). When the value is negative, it may be water. On the other hand, if the NDVI value is close to +1, it is dense green leaves. However, when the NDVI is close to zero, there is no green leaves and it may be an urbanized area. Therefore, the calculation formula as described in formula (7) is obtained.

[0104] Since green vegetation needs to be enhanced, the green vegetation area needs to be extracted first. From the definition of NDVI, when NDVI is greater than 0 and closer to +1, it indicates dense vegetation. Therefore, a threshold can be set to distinguish green vegetation from non-vegetation, as shown in formula (8). The threshold t can be changed according to the actual image requirements. In this implementation, t is set to 0.3.

[0105] After the green vegetation area is extracted, the green vegetation can be enhanced. The color image is composed of the three primary colors of RGB, and the green color intensity is mainly determined by the G component. Therefore, the green enhancement mainly increases the pixel value of the G component. The vegetation area obtained in the above steps can be enhanced by adding an enhancement coefficient. The amplitude of the enhancement coefficient is determined by the NDVI index. As mentioned above, the higher the NDVI value, the denser the vegetation. Therefore, the G component Band can be enhanced. G Reconstruction is performed, and the formula is shown in formula (9). The value of α can be adjusted according to actual needs. In this embodiment, α is set to 1. When α is smaller, the enhancement effect is weaker, and when α is larger, the enhancement effect is stronger.

[0106] In one implementation, the fusion method used in step S4 is brightness transformation based on smoothing filtering, and the formula is:

[0107]

[0108] Among them, MS is the multispectral image after green enhancement by S3, PAN is the panchromatic image after small-surface differential alignment by S2, PAN′ is the low-resolution panchromatic image after small-surface differential alignment by S2, and Fusion is the fusion result.

[0109] Specifically, step S4 quickly fuses the panchromatic image with the multispectral image after green vegetation enhancement to obtain the final fusion result after green vegetation enhancement. The fusion can use high-fidelity fusion methods such as SFIM, PSD, GS, etc. This embodiment uses brightness transformation based on smoothing filtering, that is, the SFIM model. The ratio of Pan and Pan' only retains the edge detail information of the high-resolution image, while basically eliminating the spectral and contrast information. The SFIM model can be regarded as adding high-resolution details to the low-resolution image, and its fusion result has nothing to do with the spectral difference of the high-resolution image. The SFIM model has good color retention ability.

[0110] For a comparison of the effects before and after green vegetation enhancement, see Figure 2 , where part A is the original image, part B is the image after green enhancement, A1 and B1 are the full images before and after green enhancement, respectively, and A2, B2, A3, B3 are the partial enlarged images before and after green enhancement, respectively. (For details of the color drawings, please refer to the substantive examination reference materials).

[0111] Compared with the existing technology, this method can significantly enhance the green vegetation after fusion. It is suitable for units that have the need to enhance the green vegetation in fused images. It can improve the effect and interpretation accuracy of the fused images and provide strong support for subsequent processing.

[0112] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A green vegetation enhancement method based on panchromatic and multispectral image fusion, characterized in that: include: S1: Rough registration of panchromatic and multispectral images based on rational function model; S2: Perform small-surface differential alignment of the roughly registered multispectral image and the panchromatic image; S3: Green vegetation enhancement based on multispectral images after small-surface differential calibration of the NDVI; S4: Fusing the panchromatic image with the multispectral image after green vegetation enhancement to obtain the final green vegetation enhanced fusion result; Wherein, step S1 comprises: S1.1: Calculate the overlapping area of ​​panchromatic image and multispectral image by rational function model: S1.2: Read the panchromatic image of the overlapping area and perform differential geo-registration of the multispectral image based on the panchromatic image of the overlapping area to achieve coarse registration; Step S1.2 includes: S1.2.1: Read the panchromatic image into the memory according to the calculated overlapping area, divide the panchromatic image read into the memory into uniform grid points, and obtain the image coordinates (s, l) of the corresponding panchromatic image; S1.2.2: Calculate the object coordinates (B, L, H) of each grid point using the rational function model of the panchromatic image; S1.2.3: The object-space coordinates of the grid points are calculated by the rational function model of the multispectral image to obtain the image-space coordinates (s′, l′) of each grid point in the multispectral image; S1.2.4: Construct a transformation model from the image coordinates (s, l) of the panchromatic image to the image coordinates (s′, l′) of the multispectral image: (s,l)=g(s′,l′); S1.2.5: Resample the multispectral image to the panchromatic image coordinates through the above transformation model mapping to obtain a multispectral image that is roughly aligned with the panchromatic image.

2. The green vegetation enhancement method of panchromatic and multispectral image fusion as claimed in claim 1, characterized in that: Step S2 includes: S2.1: Construct a virtual full-color image. The formula is as follows: Mss p =r1*Band B +r2*Band G +r3*Band R +r4*Band NIR In the above formula, Mss p is a virtual panchromatic image that is closer to the panchromatic image after conversion. r1, r2, r3, and r4 are the band range ratios of bands B, G, R, and NIR to the panchromatic image. B 、Band G 、Band R 、Band NIR They represent the blue light band, green light band, red light band and near-infrared band in the virtual multispectral image respectively; S2.2: By selecting uniform grid points on the panchromatic image, an m×n matching window is established with the grid points as the center, and a matching window of the same size and shape is established on the virtual panchromatic image to calculate the correlation coefficient. The formula is as follows: Among them, each correlation coefficient corresponds to a point, and the point corresponding to the largest correlation coefficient is the matching point with the same name. Among them, m, n represent the row and column size of the matching window, i, j represent the row number and column number in the matching window, and g i,j , g i+r,j+c Respectively represent the grayscale values ​​in the matching window of the M-band and N-band images, c and r are the coordinate differences of the N-band image with respect to the M-band image in rows and columns; S2.3: Obtain four points above, below, left and right of the point corresponding to the maximum correlation coefficient, and obtain the correlation coefficients corresponding to the four points, perform Gaussian surface fitting according to the correlation coefficients corresponding to the four points, and obtain the coordinates of the same-name points of the full-color image and the virtual full-color image; S2.4: Based on the coordinates of the same-name points of the obtained panchromatic image and the virtual panchromatic image, the multispectral image after the rough registration is fine-tuned.

3. The green vegetation enhancement method of panchromatic multispectral image fusion as claimed in claim 1, characterized in that: Step S3 includes: S3.1: Calculate the NDVI index using the following formula: In the above formula, NIR is the pixel value of the near infrared band, R is the pixel value of the red light band, and the higher the NDVI value, the denser the vegetation; S3.2: Extract the green vegetation area according to the set threshold, as shown below: In the above formula, t is the threshold value; S3.3: Enhance green vegetation. The formula is as follows: In the above formula, α is the enhancement ratio, (1+NDVI-t)*α is the enhancement coefficient, Band G It is the green band of the multispectral image.

4. The green vegetation enhancement method of panchromatic multispectral image fusion as claimed in claim 1, characterized in that: The fusion method used in step S4 is brightness transformation based on smoothing filtering, and the formula is: Among them, MS is the multispectral image after green enhancement by S3, PAN is the panchromatic image after small-surface differential alignment by S2, PAN′ is the low-resolution panchromatic image after small-surface differential alignment by S2, and Fusion is the fusion result.

Images