Remote sensing image fusion method, system, equipment and medium based on exponential power function

Abstract

The present invention discloses a remote sensing image fusion method, system, equipment and medium based on exponential power function, the method comprising: obtaining satellite remote sensing images with near infrared band, red band, green band and blue band; calculating characteristic ratio index and normalized index thereof; determining characteristic threshold of characteristic ratio index and characteristic threshold of corresponding normalized index; constructing vegetation feature enhancement step function; constructing vegetation feature enhancement increment coefficient function according to one of characteristic ratio index power transformation smooth constraint basis function, characteristic ratio index corresponding normalized exponential smooth constraint basis function and composite function of the two; constructing vegetation feature enhancement coefficient function to enhance vegetation features of red band and green band; synthesizing the final color image according to the enhanced red band, enhanced green band and original blue band. The present invention can construct vegetation feature enhancement coefficient function with characteristic threshold as independent variable to enhance vegetation features of true color image.

Description

Technical Field

[0001] The invention relates to a remote sensing image fusion method, system, computer equipment and storage medium based on exponential power function, and belongs to the field of remote sensing image fusion. Background Art

[0002] With the development of multi-platform, multi-sensor, multi-weather, multi-temporal and multi-resolution remote sensing technology, the types of remote sensing images that can be used by 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 selection, synthesis, correction and enhancement of remote sensing images and other image pre-processing research.

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

[0004] Most fusion research focuses on the fusion of high-resolution panchromatic images and multispectral images. There are also some literatures studying the fusion of optical images and SAR images, hyperspectral image fusion, and multi-temporal image fusion. However, there are fewer studies on the inter-spectral fusion of multispectral images, which mainly focus on true color image simulation or vegetation enhancement processing. Summary of the invention

[0005] In view of the shortcomings of satellite remote sensing true color images, such as dull vegetation and unnatural colors, the present invention provides a remote sensing image fusion method, system, computer equipment and storage medium based on exponential power function, which can improve the visual resolution and computer analysis of true color remote sensing images, explore the application potential of true color image modes of historical remote sensing data, improve the application effect of true color image modes of current and future remote sensing data, and provide a basis for the navigation application of satellite remote sensing true color image mapping on network platforms such as ordinary users' mobile phones and computers, and the observation and analysis application of physical spaces such as outdoor and indoor spaces.

[0006] The first object of the present invention is to provide a remote sensing image fusion method based on exponential power function.

[0007] The second object of the present invention is to provide a remote sensing image fusion system based on exponential power function.

[0008] A third object of the present invention is to provide a computer device.

[0009] A fourth object of the present invention is to provide a storage medium.

[0010] The first object of the present invention can be achieved by adopting the following technical solutions:

[0011] A remote sensing image fusion method based on exponential power function, the method comprising:

[0012] Obtain satellite remote sensing images with near-infrared band, red band, green band and blue band;

[0013] Calculate a characteristic ratio index and a normalized index thereof according to a near infrared band, a red band, a green band, and a blue band, wherein the characteristic ratio index is an index derived from quasi-Brovey fusion, or an index derived from Gram-Schmidt fusion / Wavelet fusion;

[0014] Determine a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index;

[0015] According to one of the characteristic threshold of the characteristic ratio index and the characteristic threshold of the corresponding normalized index, a vegetation characteristic enhancement step function is constructed;

[0016] Constructing a vegetation feature enhancement increment coefficient function according to one of a characteristic ratio exponential power transformation smoothing constraint basis function, a characteristic ratio exponential corresponding normalized exponential smoothing constraint basis function, and a composite function of a characteristic ratio exponential power transformation and / or a corresponding normalized exponent;

[0017] According to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function, a vegetation feature enhancement coefficient function is constructed;

[0018] The vegetation feature enhancement coefficient function is used to enhance the vegetation features of the red and green bands;

[0019] The final color image is synthesized based on the enhanced red band, the enhanced green band and the original blue band.

[0020] Furthermore, the characteristic threshold of the characteristic ratio index and the characteristic threshold of the corresponding normalized index are determined, specifically including:

[0021] Calculate the minimum, maximum and average values ​​of the characteristic ratio index x, denoted by x min 、x max 、x m , then the characteristic threshold c∈[x min , x max ]; or determine the threshold value x of pure water through human-computer interaction w , the threshold value x of pure bare ground b, the threshold value of pure vegetation x v , then the characteristic threshold c∈[x w , x v ];

[0022] According to the characteristic threshold of the characteristic ratio index, the characteristic threshold of the corresponding normalized index is determined as

[0023] Furthermore, the vegetation feature enhancement step function is constructed as follows:

[0024]

[0025] Or as follows:

[0026]

[0027] Among them, δ(x) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the characteristic ratio index, and δ(NDxI) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the corresponding normalized index.

[0028] Furthermore, according to the composite function of the characteristic ratio exponential power transformation smoothing constraint basis function, the vegetation feature enhancement coefficient function is constructed, which specifically includes:

[0029] Construct a composite function of the characteristic ratio exponential power transformation smooth constraint basis function as follows:

[0030] g 1 (x) = h(x) - h(c)

[0031] Where h(x) = x n , h(c)=c n , n>0, c is the characteristic threshold of the characteristic ratio index;

[0032] Construct the vegetation feature enhancement coefficient function as follows:

[0033]

[0034] Among them, k is the greenness adjustment coefficient, k>0.

[0035] Furthermore, according to the composite function of the characteristic ratio index corresponding to the normalized exponential smoothing constraint basis function, a vegetation feature enhancement coefficient function is constructed, which specifically includes:

[0036] Construct a composite function of the normalized exponential smoothing constraint basis function corresponding to the characteristic ratio index, as follows:

[0037] g 2 (x) = NDhI(x) - NDhI(c)

[0038] in, h(x)=x n , h(c)=c n , n>0, c is the characteristic threshold of the characteristic ratio index;

[0039] Construct the vegetation feature enhancement coefficient function as follows:

[0040]

[0041] Among them, k is the greenness adjustment coefficient, k>0.

[0042] Furthermore, according to the composite function of the characteristic ratio exponential power transformation and / or the corresponding normalized exponent, a vegetation characteristic enhancement increment coefficient function is constructed, which specifically includes:

[0043] Construct a composite function of the characteristic exponential power transformation product mode as follows:

[0044] g 3 (x) = [h 1 (x)-h 1 (c)][h 2 (x)-h 2 (c 1 )]

[0045] in, n 1 >0,n 2 >0,c 1 ∈[0, c], c is the feature threshold of the feature ratio index;

[0046] Construct the vegetation feature enhancement coefficient function as follows:

[0047]

[0048] Wherein, k is the greenness adjustment coefficient, k>0;

[0049] Or construct a composite function of the normalized exponential product mode of the characteristic exponential power transformation, as follows:

[0050] g 4 (x) = [NDhI 1 (x)-NDhI 1 (c)][NDhI 2 (x)-NDhI 2 (c 1 )]

[0051] is equivalent to:

[0052]

[0053] in, n 1 >0,n 2 >0,c 1 ∈[0,c];

[0054] Construct the vegetation feature enhancement coefficient function as follows:

[0055]

[0056] Wherein, k is the greenness adjustment coefficient, k>0;

[0057] Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows:

[0058] g 5 (x) = [h 1 (x)-h 1 (c 1 ))[NJH 2 (x)-NDhI 2 (c)]

[0059] is equivalent to:

[0060]

[0061] in, n 1 >0,n 2 >0,c 1 ∈[0,c];

[0062] Construct the vegetation feature enhancement coefficient function as follows:

[0063]

[0064] Where k is the greenness adjustment coefficient, k>0;

[0065] Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows:

[0066] g 6 (x) = [h 1 (x)-h 1 (c)][NDhI 2 (x)-NDhI 2 (c 1 )]

[0067] is equivalent to:

[0068]

[0069] in, n 1 >0,n 2 >0,c 1 ∈[0,c];

[0070] Construct the vegetation feature enhancement coefficient function as follows:

[0071]

[0072] Where k is the greenness adjustment coefficient, k>0;

[0073] Or construct a composite function of characteristic exponential power transformation and normalized exponential linear combination mode, as follows:

[0074] g 7 (x) = k 1 [h 1 (x)-h 1 (c)]+k 2 [NJI 2 (x)-NDhI 2 (c)]

[0075] in,

[0076] n 1 >0,n 2 >0,c k ∈[0,1 / (x max -c)];

[0077] Construct the vegetation feature enhancement coefficient function as follows:

[0078]

[0079] Among them, k is the greenness adjustment coefficient, k>0.

[0080] Furthermore, the vegetation feature enhancement coefficient function is constructed according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function, specifically including:

[0081] If the vegetation feature enhancement step function is constructed according to the feature threshold of the feature ratio index, the vegetation feature enhancement coefficient function is as follows:

[0082] F(x)=δ(x)f(x)+1

[0083] Among them, δ(x) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the characteristic ratio index, and f(x) is the vegetation feature enhancement increment coefficient function;

[0084] If the vegetation feature enhancement step function is constructed according to the feature threshold of the feature ratio index corresponding to the normalized index, the vegetation feature enhancement coefficient function is as follows:

[0085] F(x)=δ(NDxI)f(x)+1

[0086] Among them, δ(NDxI) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the corresponding normalized index, and f(x) is the vegetation feature enhancement increment coefficient function.

[0087] The second object of the present invention can be achieved by adopting the following technical solutions:

[0088] A remote sensing image fusion system based on exponential power function, the system comprising:

[0089] An acquisition module is used to acquire satellite remote sensing images with near infrared band, red band, green band and blue band;

[0090] A calculation module, used for calculating a characteristic ratio index and a normalized index thereof according to a near infrared band, a red band, a green band and a blue band, wherein the characteristic ratio index is an index derived from a quasi-Brovey fusion or an index derived from a Gram-Schmidt fusion / Wavelet fusion;

[0091] A determination module, used to determine a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index;

[0092] A first construction module is used to construct a vegetation feature enhancement step function according to one of a feature threshold of a feature ratio index and a feature threshold of a corresponding normalized index;

[0093] The second construction module is used to construct a vegetation feature enhancement increment coefficient function according to one of a characteristic ratio exponential power transformation smoothing constraint basis function, a characteristic ratio exponential corresponding normalized exponential smoothing constraint basis function, and a composite function of a characteristic ratio exponential power transformation and / or a corresponding normalized exponent;

[0094] The third construction module constructs a vegetation feature enhancement coefficient function according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function;

[0095] An enhancement module is used to enhance the vegetation features of the red band and the green band using a vegetation feature enhancement coefficient function;

[0096] The synthesis module is used to synthesize the final color image according to the enhanced red band, the enhanced green band and the original blue band.

[0097] The third object of the present invention can be achieved by adopting the following technical solutions:

[0098] A computer device comprises a processor and a memory for storing a program executable by the processor. When the processor executes the program stored in the memory, the remote sensing image fusion method is implemented.

[0099] The fourth object of the present invention can be achieved by adopting the following technical solutions:

[0100] A storage medium stores a program, and when the program is executed by a processor, the remote sensing image fusion method is implemented.

[0101] The present invention has the following beneficial effects compared with the prior art:

[0102] 1. The present invention can improve the visual resolution and computer analysis of true color remote sensing images, explore the application potential of true color image modes of historical remote sensing data, improve the application effect of true color image modes of current and future remote sensing data, and provide a basis for satellite remote sensing true color image mapping in navigation applications on network platforms such as mobile phones and computers of ordinary users, and observation and analysis applications in physical spaces such as outdoor and indoor spaces; especially in the AI ​​era, visual classification of remote sensing images is a key link in the selection of AI automatic classification training samples, which further highlights the basic role of visual classification capabilities of remote sensing images in AI automatic classification, and is also conducive to expanding the application space of visual classification and automatic classification of objects by professional users of true color satellite remote sensing images in all walks of life. In the context of the rapid development of high-resolution satellite remote sensing, it has a huge role in promoting the promotion and application of high-resolution images in all walks of life at home and abroad.

[0103] 2. The present invention provides a technology for enhancing the vegetation features of true color images of visible light satellite remote sensing. Generally speaking, the enhancement algorithm provided is based on the intrinsic characteristics of satellite image data and has strong data adaptability. The characteristic ratio index, power of the characteristic index, characteristic threshold, step function based on the characteristic threshold, smooth constraint function, and greenness adjustment coefficient parameters used have clear physical meanings, clear processing objectives, reliable quality, and easy application. It has the characteristics of multiple indexes, multiple parameters, and multiple modes to meet the individual needs of different users.

[0104] 3. The present invention provides a smooth constraint function for feature classification enhancement, which effectively avoids the situation where noise is generated during the processing: the smooth constraint function for feature classification enhancement is flexibly constructed with the incremental coefficient of 0 at the threshold as the constraint condition, which is original both in theory and in method, ensuring that the enhanced image is smoothly transitioned at the threshold, effectively avoiding the noise phenomenon that often occurs in classification enhancement with non-zero constraints.

[0105] 4. The power exponent and chromaticity adjustment coefficient of the characteristic index provided by the present invention have the function of jointly adjusting the chromaticity of vegetation and highlighting the difference in vegetation. The processing target is clear and the effect is controllable: when the power exponent and greenness adjustment coefficient of the characteristic index change from small to large, the hue of the enhanced image vegetation changes from dark green to yellow-green, and there is a good correspondence between the size changes of the two adjustment coefficients and the direction of vegetation color change. In addition, when the power exponent changes from small to large, the characteristic enhancement coefficient curve changes from a convex curve to a concave curve, and the differences between different vegetation are gradually highlighted. The joint control of the two makes the adjustment direction and enhancement effect predictable and controllable.

[0106] 5. The present invention performs a power transformation on the characteristic index, expands the function space for enhancing the vegetation features of true color images, and effectively improves the vegetation feature enhancement capability: the technology described in the power transformation of the characteristic index retains the monotonic change characteristics of the original index, and can construct a series of feature enhancement function spaces based on power transformation according to the method of constructing the feature enhancement function space according to the original index. The combination of the two effectively expands the function space for feature enhancement, especially for images with less obvious vegetation features and less prominent vegetation differences in winter, the use of power transformation function space can significantly improve their vegetation features. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] 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 only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0108] Figure 1 The present invention is a simplified flowchart of a remote sensing image fusion method based on an exponential power function according to Embodiment 1 of the present invention.

[0109] Figure 2 This is a specific flow chart of the remote sensing image fusion method based on exponential power function of Example 1 of the present invention.

[0110] Figure 3 This is a true color combined image before vegetation feature enhancement according to Example 1 of the present invention.

[0111] Figure 4 This is a schematic diagram of the classification principle of the ratio vegetation index according to Example 1 of the present invention.

[0112] Figure 5 This is a characteristic ratio index diagram of Example 1 of the present invention.

[0113] Figure 6 This is a normalized index diagram corresponding to the characteristic ratio index of Example 1 of the present invention.

[0114] Figure 7 It is a comparison curve diagram of the power transformation of the characteristic exponent and the original exponent in Example 1 of the present invention.

[0115] Figure 8 This is the red band image after being enhanced by the first enhancement scheme of Example 1 of the present invention.

[0116] Fig. 9 This is a green band diagram after being enhanced by the first enhancement scheme of Example 1 of the present invention.

[0117] Fig.10 This is the red band image after being enhanced by the second enhancement scheme of Example 1 of the present invention.

[0118] Fig.11 This is a green band diagram after being enhanced by the second enhancement scheme of Example 1 of the present invention.

[0119] Fig.12 This is the red band image after being enhanced by the third enhancement scheme of Example 1 of the present invention.

[0120] Fig.13 This is a green band diagram after being enhanced by the third enhancement scheme of Example 1 of the present invention.

[0121] Fig.14 This is the red band image after being enhanced by the fourth enhancement scheme of Example 1 of the present invention.

[0122] Fig.15 This is a green band diagram after being enhanced by the fourth enhancement scheme of Example 1 of the present invention.

[0123] Fig.16 This is the red band image after being enhanced by the fifth enhancement scheme of Example 1 of the present invention.

[0124] Fig.17 This is a green band diagram after being enhanced by the fifth enhancement scheme of Example 1 of the present invention.

[0125] Fig.18 This is the red band image after being enhanced by the sixth enhancement scheme of Example 1 of the present invention.

[0126] Fig.19 This is the green band diagram after being enhanced by the sixth enhancement scheme of Example 1 of the present invention.

[0127] Fig. 20 This is the red band image after being enhanced by the seventh enhancement scheme of Example 1 of the present invention.

[0128] Fig.21 This is a green band diagram after being enhanced by the seventh enhancement scheme of Example 1 of the present invention.

[0129] Fig. 22This is a true color composite image enhanced by the first enhancement scheme of Example 1 of the present invention.

[0130] Fig.23 This is a true color composite image enhanced by the second enhancement scheme of Example 1 of the present invention.

[0131] Fig.24 This is a true color composite image enhanced by the third enhancement scheme of Example 1 of the present invention.

[0132] Fig.25 This is a true color composite image enhanced by the fourth enhancement scheme of Example 1 of the present invention.

[0133] Fig.26 This is a true color composite image enhanced by the fifth enhancement scheme of Example 1 of the present invention.

[0134] Fig. 27 This is a true color composite image enhanced by the sixth enhancement scheme of Example 1 of the present invention.

[0135] Fig.28 This is a true color composite image enhanced by the seventh enhancement scheme of Example 1 of the present invention

[0136] Fig.29 This is a structural block diagram of a remote sensing image fusion system based on an exponential power function according to Embodiment 2 of the present invention.

[0137] Fig.30 This is a structural block diagram of a computer device according to Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0138] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0139] Embodiment 1:

[0140] The image generated by placing the red, green and blue bands of visible light multispectral satellite remote sensing data in the red, green and blue channels respectively is a true color image. Its notable feature is that the water bodies and exposed land objects such as rocks, soil and buildings are consistent with the natural colors observed on the ground, and the vegetation colors are like the colors of ground vegetation seen from an airplane at an altitude of 10,000 meters.

[0141] The unnatural and unrealistic vegetation colors in satellite remote sensing true color images are obvious defects. Maintaining the color characteristics of water and exposed objects such as rocks, soil, and buildings on satellite remote sensing true color images and improving the vegetation color characteristics of satellite remote sensing true color images are the key to satellite remote sensing true color image processing. Directly using general enhancement techniques such as traditional histogram adjustment can overcome its external shortcomings to a certain extent, but it is difficult and inefficient, and cannot overcome its inherent defects. How to apply the advantages of vegetation bands and fuse them into different bands of true color images according to the basic principles of fusion, effectively enhance the vegetation characteristics of satellite remote sensing true color images, and then effectively improve the overall quality of true color images, and improve the application potential of true color images have become the focus and difficulty of satellite remote sensing true color image processing technology research. Relevant researchers have made fruitful explorations for this, laying a theoretical and technical foundation for further solving related problems.

[0142] Based on the primary products of satellite remote sensing, Chen Chun et al. corrected the remote sensing data for Rayleigh scattering, making the color signal image close to the ground true color image (Chen Chun et al., Extraction and Reproduction of Color Signals of Remote Sensing Sources, Surveying and Mapping Science, January 2006, Vol. 31, No. 1; Han Xiuzhen et al., Research and Application of Fengyun-3D True Color Image Synthesis Method, Journal of Marine Meteorology, May 2019, Vol. 39, No. 2). You Jing et al. used the white balance method and color correction based on colorimetry to improve the vegetation characteristics of true color images and obtained more realistic true color images (You Jing et al., A White Balance Method for Processing Color Multispectral Images, Journal of Atmospheric and Environmental Optics, July 2012, Vol. 7, No. 4; Huang Honglian et al., True Color Synthesis of Multispectral Remote Sensing Images Based on Artificial Targets, November 2016, Vol. 45, No. 11).

[0143] Based on the secondary processed products of remote sensing images, Fan Xuyan and others obtained relatively good true color images through green band enhancement processing. In the early days, the overall weighted combination operation scheme of green band and near infrared band was mainly used to obtain new green bands (Fan Xuyan et al., Remote Sensing Image Simulation True Color Fusion Method Based on Principal Component Analysis, Journal of Surveying and Mapping Science and Technology, August 2006, Vol. 23, No. 4; Wang Haiyan et al., Discussion on ALOS Natural Color Image Transformation and Fusion Method, Surveying and Mapping Technology and Equipment, Vol. 14, No. 1, 2012; Shi Yuanli et al., Analysis on the Suitability of GF-2 Satellite Remote Sensing Image Mapping, Bulletin of Surveying and Mapping, No. 12, 2017); Later, it gradually developed into The normalized vegetation index is used as the classification function to classify and weight the vegetation pixels in the image to obtain a new green band (Zhang Wei et al., True color synthesis method of multispectral images based on vegetation index, Surveying and Spatial Geographic Information, Vol. 33, No. 6, December 2010); the recent development is to use the normalized vegetation index segmentation to perform Contourlet fusion on the green band and the near-infrared band to obtain a new green band image (Ding Huimei, Research on improving the color naturalness of multispectral remote sensing images using near-infrared, Master's thesis, 2016).

[0144] The color reconstruction effect evaluation technology includes visual evaluation and quantitative evaluation, which are described as follows:

[0145] 1) Visual evaluation: As we all know, although the original remote sensing true color image is consistent with the ground objects in the water body, exposed land and other objects, the color is dim and the layers are unclear in the vegetation area. Generally, true color images need to enhance the vegetation to obtain true color images that are consistent with the ground color in the above-mentioned objects. That is, the vegetation is based on green, and vegetation of different types and coverage presents various shades of green; the water body is based on blue, except for the green, yellow, black and other colors due to different components such as vegetation coverage on the water surface, high concentration of sand, and high pollution, the main body is various shades of blue; other exposed land such as rocks, exposed soil, roads, and residential areas are consistent with the rich colors of the ground, showing gray, black, white, red, orange, yellow, green, cyan, blue, purple and other colors. By visually selecting typical land features such as water areas, exposed land, and vegetation, and qualitatively comparing the color changes between the original true color image and the enhanced true color image, the effect of vegetation enhancement in the true color image can be visually evaluated.

[0146] 2) Quantitative evaluation: To a certain extent, it is the quantification of visual evaluation indicators. For the enhancement of vegetation features in true color images, it is necessary to improve the color of vegetation in true color images and to ensure the richness and clarity of the layers, details, etc. of the enhanced true color images. In general, the reconstructed images can be quantitatively evaluated from two aspects: one is the quantitative description and comparison of the vegetation color enhancement effect. In the models that describe color space such as RGB, CMYK, IHS, and CIE Lab, it is generally believed that the RGB three-primary color model is suitable for screen displays such as computers, and printing models such as CMYK are suitable for color image printouts. The color description of color space models such as IHS and CIE Lab conforms to the visual perception mode of the human eye. Based on this understanding, the method generally used to quantitatively evaluate the effect of true color images is: convert the remote sensing image described in the RGB three primary color space into an image described in the IHS or CIE Lab color space, read the chroma, saturation, intensity, etc. of the vegetation features in these color spaces before and after enhancement, and analyze their changing trends and characteristics; the second is the statistics and comparison of the quality indicators of the reconstructed color image. Generally speaking, the quality of image processing can be evaluated from three aspects: first, the statistics and comparison of the quality indicators of the enhanced true color image. Generally speaking, the quality of image processing can be evaluated from three aspects: first, the information richness of the enhanced image as a whole and the vegetation area, which can be measured by entropy and joint entropy; second, the color richness and brightness of the enhanced image as a whole and the vegetation area, which can be measured by band statistical characteristics - maximum value, minimum value, mean, variance and correlation indicators between bands - correlation coefficient, covariance, etc.; third, the level (edge), details (texture) and clarity of the enhanced image as a whole and the vegetation area can be measured by gradient, average gradient, etc. By comparing the differences in various indicators of the overall image and the vegetation area before and after enhancement, we can analyze the changing directions of spectral (grayscale, hue) information, edge (level, difference) information, and texture (detail) information.

[0147] This embodiment provides a remote sensing image fusion method based on an exponential power function; this method designs a type of characteristic ratio index based on the spectral characteristics of the objects, which can not only roughly classify water bodies (water bodies containing various water components such as sediment and pollutants), exposed objects (including soil, rocks, artificial buildings, etc.), and vegetation (including various health states and various types) according to the threshold method, but also can use it as an independent variable to construct a vegetation feature enhancement coefficient function to enhance the vegetation features of true color images. The power function based on the characteristic exponent has the same mathematical properties and physical functions as the characteristic exponent. The general expression of the vegetation feature superposition enhancement coefficient function is constructed based on the characteristic exponential power function; based on the principle of function continuity, two smooth constraint basis functions for vegetation feature enhancement are constructed - the "0" constraint function with the exponential power function value of 0 at the threshold and its corresponding normalized exponent "0" constraint function, and thus the incremental coefficient function space of vegetation feature enhancement is generated; the incremental coefficient function is substituted into the general expression of the superposition enhancement coefficient function to obtain the specific expression of the vegetation feature enhancement coefficient function, and the vegetation feature enhancement is performed on the original red band and green band of the true color image; the enhanced true color image is synthesized by using the red band and green band after vegetation feature enhancement and the original blue band.

[0148] like Figure 1 and Figure 2 As shown, the remote sensing image fusion method based on exponential power function of this embodiment includes the following steps:

[0149] S101. Acquire satellite remote sensing images having near infrared band, red band, green band, and blue band.

[0150] Open a multispectral remote sensing image with near infrared band (NIR), red band (R), green band (G), and blue band (B); Figure 3 This is a true color composite image before vegetation feature enhancement (Note: all image illustrations are stretched by 0.1% according to the default setting of envi).

[0151] S102. Calculate a characteristic ratio index and a normalized index thereof according to the near infrared band, the red band, the green band, and the blue band.

[0152] The characteristic index of this embodiment is constructed as follows:

[0153] Assume that the near infrared, red, green, and blue bands of the visible light satellite remote sensing image are NIR, R, G, and B respectively. Then the ratio vegetation index VI is:

[0154]

[0155] For satellite remote sensing band data, VI ≥ 0.

[0156] The value of the ratio vegetation index of major types of objects such as water, bare objects, and vegetation on remote sensing images varies from small to large. The product fusion using this index as the enhancement coefficient can effectively enhance the vegetation characteristics on each band of the image. At the same time, the monotonically increasing nature of this index from water to land makes it have the function of object classification, that is, a characteristic threshold can always be found to divide the image into two categories. For example, if a water body threshold is found, the image can be divided into water and non-water categories; if a bare object value range is found, the image can be divided into bare objects and non-bare objects; if a vegetation threshold is found, the image can be divided into vegetation and non-vegetation categories. Of course, due to the widespread existence of "same spectrum but different objects, same object but different spectrum" pixels in images, compared with the precise classification of objects, this classification will always have misclassification and omissions, and can only be called a rough classification. It can be seen that this index has the function of object classification and vegetation feature enhancement. Let the minimum value, maximum value, water body threshold, pure bare object threshold, pure vegetation threshold, and average value of VI be VI respectively. min VI max VI w VI b VI v VI m The schematic diagram of its classification principle is as follows Figure 4 shown.

[0157] When the near-infrared band is used as the numerator and the combination of the three true color bands is used as the denominator, the constructed ratio index has similar functions to the ratio vegetation index. Therefore, the following ratio indices are also suitable rough classification functions, such as the inverse of the ratio water index (IWI), the inverse of the ratio rock index (IRI), the two-band Brovey index (DBI), the three-band Brovey index (TBI), and the fusion derivative index can also be obtained by using fusion methods such as Gram-Schmidt and Wavelet.

[0158] Assume b i is one of the true color bands, and the bands involved in the fusion are NIR and m true color bands. The number of band results after fusion is m, and the fusion result is recorded as b′ i .

[0159] 1) For quasi-Brovey fusion, the fusion result is recorded as:

[0160]

[0161] Where m = 1, 2, 3, i = 1, ..., m, b i 、b j It is the true color band involved in fusion.

[0162] 2) For Gram-Schmidt or Wavelet fusion, the fusion result is recorded as:

[0163] b′ i =Fusion(NIR, b 1 , ..., b m )

[0164] Where m = 1, 2, 3, i = 1, ..., m, b 1 ,,...,b m True color bands involved in fusion. Fusion is GS (Gram-Schmidt) or WL (Wavelet) fusion method

[0165] The characteristic ratio index can be uniformly expressed as:

[0166]

[0167] Wherein, n=1, ..., m.

[0168] Then the normalized index of FI is:

[0169]

[0170] The default characteristic ratio index is:

[0171] b′ i =GS(NIR, b 1 , b 2 , b 3 )

[0172]

[0173] This embodiment takes the characteristic ratio index derived from Gram-Schmidt fusion as an example, applies the image enhancement tool in ENVI software, takes NIR as the enhancement source, adopts the Gram-Schmidt fusion method, and performs fusion operation with the red, green, and blue bands of the true color image to obtain the fused red, green, and blue bands TFR, TFG, and TFB.

[0174]

[0175]

[0176] According to the above formula, the band operation expression (1.0*b1+b2+b3) / (1.0*b4+b5+b6) is established to calculate the characteristic ratio index x, where b1, b2, b3 are the fused red, green, and blue bands, and b4, b5, and b6 are the original red, green, and blue bands. The calculation results are as follows: Figure 5 shown.

[0177] The band operation expression (1.0*b1-1) / (1.0*b1+1) is established to calculate NDxI, where b1 is the characteristic ratio index x. The calculation results are as follows: Figure 6 shown.

[0178] S103, determining a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index.

[0179] From the perspective of classification enhancement only, the threshold can indeed be any value between the maximum and minimum values ​​of the feature index, but as mentioned above, the enhancement goal of this embodiment is to maintain the features of exposed objects such as water bodies, rocks, soil, and buildings on true color images, and to enhance vegetation features. For an image with vegetation, in addition to vegetation features containing pure vegetation pixels, there are also a large number of vegetation features containing mixed pixels containing vegetation. Theoretically, the smaller the feature threshold c is, the more pixels contain vegetation features; conversely, the fewer.

[0180] In this embodiment, the minimum, maximum and average values ​​of the characteristic ratio index x are calculated and are respectively denoted as x min 、x max 、x m , then the characteristic threshold c∈[x min , x max ]; or determine the threshold value x of pure water through human-computer interaction w , the threshold value x of pure bare ground b , the threshold value of pure vegetation x v , then the characteristic threshold c∈[x w , x v ].

[0181] Assume that the threshold of the characteristic index of pure vegetation is x v , the average value of the characteristic index is x m , the threshold value x of pure bare ground b , the threshold value of the characteristic index of pure water is x w . Classify by greater than the threshold, and take the threshold as x m Compared to the threshold x v Will include more mixed pixels containing vegetation; take the threshold x b All mixed pixels containing vegetation will be included; if there is water on the image, for pure water, it does not contain any vegetation features, and the threshold is x w , including pure bare ground objects, pure vegetation, and all mixed pixels containing vegetation. It can be seen that whether the threshold is the pure water threshold, the average value, the pure bare ground threshold, or the pure vegetation threshold, it can play a role in enhancing vegetation features.

[0182] Take x w or x bAs a feature threshold, the vegetation features of all pixels in the image can be enhanced while maintaining the features of pure water bodies and pure exposed objects on land; v As the feature threshold, only the pure vegetation features are enhanced, while the mixed pixels containing vegetation are omitted; the average value x m As the characteristic threshold, the number of pixels for vegetation enhancement is between the above two, so generally speaking, the value range of the characteristic threshold is:

[0183] c∈[x w , x v ]

[0184] Given the complexity of image features, images without water bodies are not important. w , images with water bodies, x w 、x b 、x v Eigenvalues ​​such as x need to be determined through human-computer interaction, which is relatively difficult; m It can be obtained directly through statistical calculation, which is relatively easy. Generally, x m is the default value, and can also be other values ​​according to the actual characteristics of the image or user enhancement needs, such as x w 、x b or x v , or other values.

[0185] If there is no vegetation in the image, but only exposed objects such as water bodies, deserts, soil, rocks, buildings, etc., such true color images do not actually need to be enhanced, and there is no need to determine the value of the feature threshold.

[0186] According to the characteristic threshold of the characteristic ratio index, the characteristic threshold of the corresponding normalized index is determined as

[0187] This example uses the ENVI statistical tool to calculate the average value x of the characteristic ratio index x m , and use it as the feature threshold c, c = x m =1, then NDxI c =0.

[0188] S104: construct a vegetation feature enhancement step function according to one of a feature threshold of a feature ratio index and a feature threshold of a corresponding normalized index.

[0189] Let c be the vegetation feature threshold. For a specific satellite remote sensing image, x has a corresponding statistical maximum and minimum, which are recorded as: min , x max ,c∈(x min , x max). Then the rough classification function of the ground objects on the binary image can be expressed by a vegetation feature enhancement step function as follows:

[0190]

[0191] Obviously, the step function δ(x) is discontinuous at the threshold c.

[0192] Let NDxI be the normalized index corresponding to x, The above step function is equivalent to:

[0193]

[0194] S105, constructing a vegetation feature enhancement incremental coefficient function according to one of the characteristic ratio exponential power transformation smoothing constraint basis function, the characteristic ratio exponent corresponding normalized exponential smoothing constraint basis function, and the composite function of the characteristic ratio exponential power transformation and / or the corresponding normalized exponent.

[0195] The characteristic ratio exponent x to the power of n is h(x) = x n (n>0) has the same mathematical properties as the original exponent h(x)=x, such as Figure 7 shown.

[0196] 1) There is a one-to-one mapping relationship between the independent variable x and the dependent variable h(x)

[0197] h′(x)=nx n-1

[0198] When n>0, for all x greater than 0, h′(x) is always greater than 0, h(x)=x n It has the same monotonically increasing property as h(x)=x. That is, if x can enhance vegetation features, then h(x) can also enhance vegetation features.

[0199] 2) h(x) has the same monotonicity as x

[0200] Therefore, the characteristic exponent x to the power of n, that is, h(x) = x n Like the characteristic index, it can be used as a basic parameter for enhancing vegetation characteristics.

[0201] Regarding the construction of the smooth constraint basis function based on the feature threshold, since the step function δ(x) is a discontinuous function at the threshold c, it is easy to cause the fused image to have noise such as sawtooth and spots, making the color transition between the objects on the image uneven and not smooth. Mathematically speaking, this phenomenon is because the enhanced image destroys the continuity of the original image. The necessary and sufficient condition for the enhanced image b′ to remain smooth is that the function F(h(x))=δ(x)f(h(x)) is a continuous function, that is:

[0202]

[0203] The above condition is equivalent to:

[0204]

[0205] f(h(x)) that satisfies the above constraints is called a smooth constraint function.

[0206] Obviously, there are infinitely many such functions, two of which are concise smooth constraint functions:

[0207] u(h(x))=h(x)-h(c)

[0208] and

[0209] v(h(x))=NDhI-NDhI c

[0210] Among them, NDhI is the homology normalized index of h(x), NDhI c is the normalized index corresponding to the threshold c, and the expressions are:

[0211]

[0212]

[0213]

[0214]

[0215] u(h(x)) is called the characteristic ratio exponential power transform smoothing constrained basis function; v(h(x)) is called the homologous normalized exponential smoothing constrained basis function of the characteristic ratio exponential power transform.

[0216] The composite function of the above two types of smooth constraint basis functions can be used to construct the vegetation feature enhancement coefficient function, among which several typical vegetation feature enhancement coefficient functions are constructed as follows:

[0217] 1) Construct vegetation feature enhancement coefficient function based on composite function of characteristic ratio exponential power transformation smooth constraint basis function

[0218] h(x)=x n

[0219] h(c)=c n

[0220] g 1 (x) = h(x) - h(c)

[0221]

[0222] is equivalent to:

[0223]

[0224] Where n>0, k is a constant coefficient greater than 0.

[0225] 2) Construct vegetation feature enhancement coefficient function based on the composite function of the characteristic ratio index corresponding to the normalized exponential smoothing constraint basis function

[0226] h(x)=x n

[0227] h(c)=c n

[0228]

[0229]

[0230] g 2 (x) = NDhI(x) - NDhI(c)

[0231]

[0232] is equivalent to:

[0233]

[0234] Right now:

[0235]

[0236] Wherein, n>0, k is a constant coefficient greater than 0.

[0237] 3) Constructing vegetation feature enhancement increment coefficient function based on composite function of characteristic exponential power transformation product mode

[0238] When the ratio exponential power transformation product mode is used to construct the feature enhancement incremental coefficient function, one of the product factors is the feature ratio exponential power transformation smooth constraint basis function to ensure that it is 0 at the threshold, and the other factor can relax the constraint and is not necessarily required to be 0 at the threshold, but it needs to satisfy that the factor value is greater than 0 at the threshold, and the product of the two factors still satisfies the smooth constraint condition, that is:

[0239]

[0240]

[0241]

[0242]

[0243]

[0244] is equivalent to:

[0245]

[0246] Among them, n 1 >0,n 2 >0,c 1 ∈[0, c], the default value is c 1 =c, k is a constant coefficient greater than 0.

[0247] 4) Construct vegetation feature enhancement increment coefficient function based on the composite function of normalized exponential product mode of characteristic exponential power transformation

[0248] When the normalized exponential product mode is used to construct the vegetation feature enhancement incremental coefficient function, one of the multiplication factors is a normalized exponential smoothing constraint basis function to ensure that it is 0 at the threshold, and the other multiplication factor can relax the constraint and is not necessarily required to be 0 at the threshold, but it needs to satisfy that the factor value is greater than 0 at the threshold, and the product of the two factors still satisfies the smoothing constraint condition, that is:

[0249]

[0250]

[0251]

[0252]

[0253]

[0254]

[0255]

[0256]

[0257]

[0258]

[0259] g 4 (x) = [NDhI 1 (x)-NDhI 1 (c)][NDhI 2 (x)-NDhI 2 (c 1 )]

[0260] is equivalent to:

[0261]

[0262]

[0263] is equivalent to:

[0264]

[0265] Right now:

[0266]

[0267] Among them, n 1 >0,n 2 >0,c 1 ∈[0, c], the default value is c 1 =c, k is a constant coefficient greater than 0.

[0268] 5) Construct vegetation feature enhancement increment coefficient function based on composite function of characteristic exponential power transformation and normalized exponential product mode

[0269] When the vegetation feature enhancement increment coefficient function is constructed by using the characteristic ratio exponential power transformation and its normalized exponential product mode, one of the product factors is the normalized exponential smoothing constraint basis function to ensure that it is 0 at the threshold, and the other factor can relax the constraint and is not necessarily required to be 0 at the threshold, but it needs to satisfy that the factor value is greater than 0 at the threshold, and the product of the two factors still satisfies the smoothing constraint condition, that is:

[0270]

[0271]

[0272]

[0273]

[0274]

[0275]

[0276]

[0277] g 5 (x) = [h 1 (x)-h 1 (c 1 ))[NJH 2 (x)-NDhI 2 (c)]

[0278] is equivalent to:

[0279]

[0280]

[0281] is equivalent to:

[0282]

[0283] Right now:

[0284]

[0285] or:

[0286]

[0287]

[0288]

[0289]

[0290]

[0291]

[0292]

[0293] g 6 (x) = [h 1 (x)-h 1 (c)][NDhI 2 (x)-NDhI 2 (c 1 )]

[0294] is equivalent to:

[0295]

[0296]

[0297] is equivalent to:

[0298]

[0299] Right now:

[0300]

[0301] Among them, n 1 >0,n 2 >0,c 1 ∈[0, c], the default value is c 1 =c, k is a constant coefficient greater than 0.

[0302] 6) Construct vegetation feature enhancement increment coefficient function based on composite function of characteristic exponential power transformation and normalized exponential linear combination mode

[0303]

[0304]

[0305]

[0306]

[0307]

[0308]

[0309]

[0310] g 7 (x) = k 1 [h 1 (x)-h 1 (c)]+k 2 [NJI 2 (x)-NDhI 2 (c)]

[0311]

[0312] is equivalent to:

[0313] f 7 (x) = k{k 1 [h 1 (x)-h 1 (c)]+k 2 [NJI 2 (x)-NDhI 2 (c)]}

[0314] in,

[0315]

[0316]

[0317]

[0318]

[0319] n 1 >0,n 2 >0,c k ∈[0,1 / (x max-c)], the default value is c k =0, k is a constant coefficient greater than 0.

[0320] The general expression of the incremental coefficient function can be uniformly expressed as:

[0321]

[0322] Wherein, k is the greenness adjustment coefficient, k>0, and k=1 by default; g(x) is a composite function of the smooth constraint basis function, which can be a characteristic ratio exponential power transformation smooth constraint basis function, a corresponding normalized exponential smooth constraint basis function, or a composite function of the two (such as a product mode composite or a linear combination mode composite); g(x max ) is the maximum value of the smooth constrained composite function.

[0323] S106. Construct a vegetation feature enhancement coefficient function according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function.

[0324] If the vegetation feature enhancement step function is constructed according to the characteristic threshold of the characteristic ratio index, according to the superposition fusion principle, the vegetation feature enhancement coefficient can be obtained from the step function and the incremental coefficient function:

[0325] F(x)=δ(x)f(x)+1

[0326] If the vegetation feature enhancement step function is constructed according to the characteristic threshold of the characteristic ratio index corresponding to the normalized index, according to the superposition fusion principle, the vegetation feature enhancement coefficient can be obtained from the step function and the incremental coefficient function:

[0327] F(x)=δ(NDxI)f(x)+1

[0328] S107. Use the vegetation feature enhancement coefficient function to enhance the vegetation features of the red band and the green band.

[0329] When h(x)=x n When , the above step function is equivalent to:

[0330]

[0331] Assume f(x) is a bounded continuous function, and is the incremental coefficient function of vegetation feature enhancement. According to the principle of image superposition enhancement, the general fusion expression of vegetation feature enhancement is as follows:

[0332] b′=[1+δ(x)f(x)]×b

[0333] When h(x) = x n When , the general fusion expression of vegetation feature enhancement is as follows:

[0334] b′=[1+δ(x)f(h(x))]×b

[0335] Where b is the image before fusion, b′ is the image after fusion, 1+δ(x)f(h(x)) is the vegetation feature enhancement coefficient under the condition of characteristic exponential power transformation, and the product of the enhancement coefficient and the original image is the enhanced image.

[0336] Assuming that the red, green, and blue bands of the true color image are R, G, and B respectively, this embodiment uses the above enhancement coefficients to enhance the red and green bands, and the image fusion expression for vegetation feature enhancement is as follows:

[0337] R′=[δ(x)f(x)+1]×R

[0338] G′=[δ(x)f(x)+1]×G

[0339] Or as follows:

[0340] R′=[δ(NDxI)f(x)+1]×R

[0341] G′=[δ(NDxI)f(x)+1]×G

[0342] This embodiment uses the ENVI statistical tool to obtain the maximum value of the characteristic ratio index x, x max =1.770134,NDxI max =0.278013.

[0343] 1) The first enhancement solution:

[0344] g 1 (x) = x n -c n =x 0.5 -1

[0345]

[0346] k=0.5,n=0.5

[0347]

[0348] R′=[δ(x)f(x)+1]×R

[0349] G′=[δ(x)f(x)+1]×G

[0350] According to the above formula, enter the calculation expression in ENVI: uint(((b1 gt 1.0)*(b1^0.5-1.0)*0.5 / 0.330463+1)*b2), where b1 is the characteristic ratio index x, b2 is the red band or green band of the true color image, and the vegetation feature enhancement calculation result is as follows Figure 8and Fig. 9 shown.

[0351] 2) The second enhancement scheme:

[0352] g 2 (x) = NDhI(x) - NDhI(c) = (x n -1) / (x n +1)

[0353] g 2 (x max )=NDxI max -NDX c =0.141801

[0354] k=0.5,n=0.5

[0355]

[0356] R′=[δ(x)f(x)+1]×R

[0357] G′=[δ(x)f(x)+1]×G

[0358] According to the above formula, enter the calculation expression: uint(((b1 gt 1.0)*((b1^0.5-1) / (b1^0.5+1))*0.5 / 0.141801+1)*b2), where b1 is the characteristic ratio index x, and b2 is the red band or green band of the true color image. The vegetation feature enhancement calculation result is as follows Fig.10 and Fig.11 shown.

[0359] 3) The third enhancement solution:

[0360]

[0361]

[0362] k=0.5,n 1 =0.1, n 2 =0.1, c 1 =0

[0363]

[0364] R′=[δ(x)f(x)+1]×R

[0365] G′=[δ(x)f(x)+1]×G

[0366] According to the above formula, enter the calculation expression in ENVI: uint(((b1 gt 1.0)*((b1^0.1-1)*(b1^0.1))*0.5 / 0.062221+1)*b2), where b1 is the characteristic ratio index x, b2 is the red band or green band of the true color image, and the vegetation feature enhancement calculation result is as follows: Fig.12 and Fig.13 shown.

[0367] 4) The fourth enhancement scheme:

[0368]

[0369]

[0370] k=0.5,n 1 =6,n 2 =6, c 1 =0

[0371]

[0372] R′=[δ(x)f(x)+1]×R

[0373] G′=[δ(x)f(x)+1]×G

[0374] According to the above formula, enter the calculation expression in ENVI: uint(((b1 gt 1.0)*(2.0*((b1^6-1)*(b1^6-1)) / ((b1^6+1)*(b1^6+1)))*0.5 / 1.756068+1)*b2), where b1 is the characteristic ratio index x, b2 is the red band or green band of the true color image, and the vegetation feature enhancement calculation results are as follows: Fig.14 and Fig.15 shown.

[0375] 5) The fifth enhancement scheme:

[0376]

[0377]

[0378] k=0.5,n 1 =0.1, n 2 =2, c 1 =0

[0379]

[0380] R′=[δ(x)f(x)+1]×R

[0381] G′=[δ(x)f(x)+1]×G

[0382] According to the above formula, enter the calculation expression: uint(((b1 gt 1.0)*(((b1^2-1)*(b1^0.1)) / (b1^2+1))*0.5 / 0.142014+1)*b2), where b1 is the characteristic ratio index x, and b2 is the red band or green band of the true color image. The calculation results of vegetation feature enhancement are as follows Fig.16 and Fig.17 shown.

[0383] 6) The sixth enhancement scheme:

[0384]

[0385]

[0386] k=0.5,n 1 =0.1, n 2 =6, c 1 =1

[0387]

[0388] R′=[δ(x)f(x)+1]×R

[0389] G′=[δ(x)f(x)+1]×G

[0390] According to the above formula, enter the calculation expression: uint(((b1 gt 1.0)*((((b1^0.1-1)*(b1^6-1)) / (b1^6+1)) / 0.113834)*0.5+1)*b2), where b1 is the characteristic ratio index x, and b2 is the red band or green band of the true color image. The calculation results of vegetation feature enhancement are as follows Fig.18 and Fig.19 shown.

[0391] 7) The seventh enhancement scheme:

[0392] c k ∈(0, 0.564928), take c k =0.564928, k1 = -0.292602, k2 = -21.478080

[0393]

[0394]

[0395] k=0.5,n 1 =0.5, n 2 =0.5

[0396]

[0397] R′=[δ(x)f(x)+1]×R

[0398] G′=[δ(x)f(x)+1]×G

[0399] According to the above formula, enter the calculation expression: uint(((b1 gt 1.0)*(((-0.291602)*(b1^0.5-1)-21.478080*((b1^0.5-1) / (b1^0.5+1))) / (-3.141993))*0.5+1)*b3), where b1 is the characteristic ratio index x, and b2 is the red band or green band of the true color image. The calculation results of vegetation feature enhancement are as follows Fig. 20 and Fig.21 shown.

[0400] S108, synthesizing the enhanced red band, the enhanced green band and the original blue band to obtain a final color image.

[0401] This embodiment uses the three bands R', G', and B to correspond to the red, green, and blue channels of the color image to synthesize a true color image; the true color synthesis image obtained by the first enhancement scheme is as follows Fig. 22 The statistical characteristics of the enhanced color image and the original true color image are shown in Table 1. The true color composite image obtained by the second enhancement scheme is shown in Fig.23 The statistical characteristics of the enhanced color image and the original true color image are shown in Table 2. The true color composite image obtained by the third enhancement scheme is shown in Fig.24 The statistical characteristics of the enhanced color image and the original true color image are compared and analyzed in Table 3. The true color composite image obtained by the fourth enhancement scheme is shown in Fig.25 The statistical characteristics of the enhanced color image and the original true color image are shown in Table 4. The true color composite image obtained by the fifth enhancement scheme is shown in Fig.26 The statistical characteristics of the enhanced color image and the original true color image are shown in Table 5. The true color composite image obtained by the sixth enhancement scheme is shown in Fig. 27 The statistical characteristics of the enhanced color image and the original true color image are compared in Table 6. The true color composite image obtained by the seventh enhancement scheme is shown in Fig.28 As shown in Table 7, the statistical characteristics of the enhanced color image results and the original true color image are compared and analyzed.

[0402] Table 1 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (first enhancement scheme)

[0403]

[0404]

[0405] Table 2 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (second enhancement scheme)

[0406]

[0407] Table 3 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (third enhancement scheme)

[0408]

[0409]

[0410] Table 4 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (the fourth enhancement scheme)

[0411]

[0412] Table 5 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (fifth enhancement scheme)

[0413]

[0414]

[0415] Table 6 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (sixth enhancement scheme)

[0416]

[0417] Table 7 Comparison of statistical characteristics of true color images after vegetation feature enhancement and original true color images (seventh enhancement scheme)

[0418]

[0419]

[0420] From Tables 1 to 7, we can see that there are two main effects:

[0421] First, the green vegetation features of true color images are effectively enhanced, and the visual separability and computer parsing capabilities of vegetation are improved: the vegetation information of visible light satellite remote sensing true color images is not prominent, natural, or real, and the tone of pure vegetation pixels is gray, which is easily confused with other ground objects; the vegetation color of mixed vegetation pixels is submerged in the colors of other ground objects, and the vegetation information is obscured; resulting in low visual resolution of true color image vegetation information, affecting the application effect of the true color image mode; the method of this embodiment enhances the vegetation pixels of the true color combined image point by point, and the color, texture, and layering of the vegetation are comprehensively improved, effectively improving the visual resolution and computer parsing capabilities of the true color image vegetation information, and improving the vegetation analysis capability and effect of the true color image mode.

[0422] Secondly, the overall characteristics of true color images have been greatly improved, expanding their application scope and potential: a large amount of historical remote sensing data has true color bands, which are often abandoned due to true color processing technology. With the development of high-resolution images, although the true color image mode has become a standardized application product, its inherent defects have affected its application effect and potential; while enhancing the vegetation characteristics on the true color image, the method of this embodiment maintains the characteristics of water bodies and exposed objects such as soil, rocks, and buildings on the true color image, greatly improving the overall visual characteristics and effects of the true color image. At the same time, the correlation between the bands of the true color image is reduced, and the image color, texture, and level are richer, which effectively improves the visual resolution and computer analysis capabilities of the true color image, which is conducive to exploring the application potential of the true color image mode of historical remote sensing data and improving the application effect of the true color image mode of current and future remote sensing data; it is conducive to expanding the navigation application of satellite remote sensing true color image mapping on network platforms such as ordinary users' mobile phones and computers, and the observation and analysis application of physical spaces such as outdoor and indoor spaces, and is conducive to playing its basic role in AI automatic classification.

[0423] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned embodiment method may be implemented by instructing related hardware through a program, and the corresponding program may be stored in a computer-readable storage medium.

[0424] It should be noted that although the method operations of the above embodiments are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted can be performed in a different order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be decomposed into multiple steps.

[0425] Embodiment 2:

[0426] like Fig.29As shown, this embodiment provides a remote sensing image fusion system based on an exponential power function, which includes an acquisition module 2901, a calculation module 2902, a determination module 2903, a first construction module 2904, a second construction module 2905, a third construction module 2906, an enhancement module 2907 and a synthesis module 2908. The specific functions of each module are as follows:

[0427] The acquisition module 2901 is used to acquire satellite remote sensing images with near infrared band, red band, green band and blue band;

[0428] A calculation module 2902 is used to calculate a characteristic ratio index and a normalized index thereof according to a near infrared band, a red band, a green band, and a blue band, wherein the characteristic ratio index is an index derived from a quasi-Brovey fusion, or an index derived from a Gram-Schmidt fusion / Wavelet fusion;

[0429] A determination module 2903 is used to determine a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index;

[0430] A first construction module 2904 is used to construct a vegetation feature enhancement step function according to one of a feature threshold of a feature ratio index and a feature threshold of a corresponding normalized index;

[0431] The second construction module 2905 is used to construct a vegetation feature enhancement increment coefficient function according to one of a characteristic ratio exponential power transformation smoothing constraint basis function, a characteristic ratio exponential corresponding normalized exponential smoothing constraint basis function, and a composite function of a characteristic ratio exponential power transformation and / or a corresponding normalized exponent;

[0432] The third construction module 2906 is used to construct a vegetation feature enhancement coefficient function according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function;

[0433] An enhancement module 2907 is used to enhance the vegetation features of the red band and the green band using a vegetation feature enhancement coefficient function;

[0434] The synthesis module 2908 is used to synthesize the final color image according to the enhanced red band, the enhanced green band and the original blue band.

[0435] The specific implementation of each module in this embodiment can refer to the above-mentioned embodiment 1, which will not be described one by one here; it should be noted that the system provided in this embodiment is only illustrated by the division of the above-mentioned functional modules. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above.

[0436] Embodiment 3:

[0437] like Fig.30 As shown, this embodiment provides a computer device, which includes a processor 3002, a memory, an input device 3003, a display device 3004 and a network interface 3005 connected through a system bus 3001. The processor 3002 is used to provide computing and control capabilities, and the memory includes a non-volatile storage medium 3006 and an internal memory 3007. The non-volatile storage medium 3006 stores an operating system, a computer program and a database. The internal memory 3007 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium 3006. When the computer program is executed by the processor 3002, the remote sensing image fusion method of the above-mentioned embodiment 1 is implemented.

[0438] Embodiment 4:

[0439] This embodiment provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the remote sensing image fusion method of the above-mentioned embodiment 1 is implemented.

[0440] In summary, the satellite remote sensing true color image vegetation feature enhancement technology of the present invention is mainly aimed at the inherent defects of the vegetation features of satellite remote sensing images with near-infrared, red, green and blue bands. According to the intrinsic relationship of the remote sensing band data, a multi-index, multi-parameter and multi-mode enhancement method is constructed to effectively improve the vegetation features on the original true color synthetic image and greatly improve the overall effect of the true color image. The technical method has a solid theoretical basis, clear physical meaning and a wide range of applications. The enhanced image has bright colors, rich information, and is easy to visually and automatically classify. Especially in the current context of the rapid development of high-resolution satellite remote sensing, it has a huge role in promoting the popularization and application of high-resolution images in various industries at home and abroad.

[0441] The above is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and inventive concept of the present invention within the scope disclosed by the present invention, which shall fall within the protection scope of the present invention.

Claims

1. A remote sensing image fusion method based on exponential power function, characterized in that: The method comprises: Obtain satellite remote sensing images with near-infrared band, red band, green band and blue band; Calculating a characteristic ratio index and a normalized index thereof according to a near infrared band, a red band, a green band, and a blue band, wherein the characteristic ratio index is an index derived from a quasi-Brovey fusion, an index derived from a Gram-Schmidt fusion, or an index derived from a Wavelet fusion; Determine a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index; According to one of the characteristic threshold of the characteristic ratio index and the characteristic threshold of the corresponding normalized index, a vegetation characteristic enhancement step function is constructed; Constructing a vegetation feature enhancement increment coefficient function according to one of a characteristic ratio exponential power transformation smoothing constraint basis function, a characteristic ratio exponential corresponding normalized exponential smoothing constraint basis function, and a composite function of a characteristic ratio exponential power transformation and / or a corresponding normalized exponent; According to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function, a vegetation feature enhancement coefficient function is constructed; The vegetation feature enhancement coefficient function is used to enhance the vegetation features of the red and green bands; The final color image is synthesized based on the enhanced red band, the enhanced green band and the original blue band; Among them, according to the composite function of the characteristic ratio exponential power transformation smooth constraint basis function, the vegetation feature enhancement coefficient function is constructed, which specifically includes: Construct a composite function of the characteristic ratio exponential power transformation smooth constraint basis function as follows: g1(x)=h(x)-h(c) Where h(x) = x n , h(c)=c n , n = 0.5, x is the characteristic ratio index, c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Among them, according to the composite function of the normalized exponential smoothing constraint basis function corresponding to the characteristic ratio index, the vegetation feature enhancement coefficient function is constructed, which specifically includes: Construct a composite function of the normalized exponential smoothing constraint basis function corresponding to the characteristic ratio index, as follows: g2(x)=NDhI(x)-NDhI(C) in, h(x)=x n , h(c)=c n , n = 0.5, x is the characteristic ratio index, c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Among them, according to the composite function of the characteristic ratio exponential power transformation and / or the corresponding normalized exponent, the vegetation feature enhancement increment coefficient function is constructed, which specifically includes: Construct a composite function of the characteristic exponential power transformation product mode as follows: g3(x)=[h1(x)-h1(c)][h2(x)-h2(c1)] in, n1=0.1,n2=0.1,c1∈[0,c],x is the characteristic ratio index,c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of the normalized exponential product mode of the characteristic exponential power transformation, as follows: g4(x)=[NDhI1(x)-NDhI1(c)][NDhI2(x)-NDhI2(c1)] is equivalent to: in, n1=6, n2=6, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows: g5(x)=[h1(x)-h1(c1))[NDhI2(x)-NDhI2(c)] is equivalent to: in, n1=0.1, n2=2, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows: g6(x)=[h1(x)-h1(c)][NDhI2(x)-NDhI2(c1)] is equivalent to: in, n1=0.1, n2=6, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential linear combination mode, as follows: g7(x)=k1[h1(x)-h1(c)]+k2[NDhI2(x)-NDhI2(c)] in, n1=0.5,n2=0.5,c k ∈[0,1 / (x max -c)]; Construct the vegetation feature enhancement coefficient function as follows: Among them, k is the greenness adjustment coefficient, k=0.

5.

2. The remote sensing image fusion method according to claim 1, characterized in that: The characteristic threshold of determining the characteristic ratio index and the characteristic threshold of the corresponding normalized index specifically include: Calculate the minimum, maximum and average values ​​of the characteristic ratio index x, denoted by x min 、x max 、x m , then the characteristic threshold c∈[x min , x max ]; or determine the threshold value x of pure water through human-computer interaction w , the threshold value x of pure bare ground b , the threshold value of pure vegetation x v , then the characteristic threshold c∈[x w , x v ]; According to the characteristic threshold of the characteristic ratio index, the characteristic threshold of the corresponding normalized index is determined as 3. The remote sensing image fusion method according to claim 1, characterized in that: The vegetation feature enhancement step function is constructed as follows: Or as follows: Among them, δ(x) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the characteristic ratio index, and δ(NDxI) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the corresponding normalized index.

4. The remote sensing image fusion method according to any one of claims 1 to 3, characterized in that: The step of constructing the vegetation feature enhancement coefficient function according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function specifically includes: If the vegetation feature enhancement step function is constructed according to the feature threshold of the feature ratio index, the vegetation feature enhancement coefficient function is as follows: F(x)=δ(x)f(x)+1 Among them, δ(x) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the characteristic ratio index, and f(x) is the vegetation feature enhancement increment coefficient function; If the vegetation feature enhancement step function is constructed according to the feature threshold of the feature ratio index corresponding to the normalized index, the vegetation feature enhancement coefficient function is as follows: F(x)=δ(NDxI)f(x)+1 Among them, δ(NDxI) is the vegetation feature enhancement step function constructed according to the characteristic threshold of the corresponding normalized index, and f(x) is the vegetation feature enhancement increment coefficient function.

5. A remote sensing image fusion system based on exponential power function, characterized in that: The system comprises: An acquisition module is used to acquire satellite remote sensing images with near infrared band, red band, green band and blue band; A calculation module, used for calculating a characteristic ratio index and a normalized index thereof according to a near infrared band, a red band, a green band and a blue band, wherein the characteristic ratio index is an index derived from a quasi-Brovey fusion, an index derived from a Gram-Schmidt fusion, or an index derived from a Wavelet fusion; A determination module, used to determine a characteristic threshold of a characteristic ratio index and a characteristic threshold of a corresponding normalized index; A first construction module is used to construct a vegetation feature enhancement step function according to one of a feature threshold of a feature ratio index and a feature threshold of a corresponding normalized index; The second construction module is used to construct a vegetation feature enhancement increment coefficient function according to one of a characteristic ratio exponential power transformation smoothing constraint basis function, a characteristic ratio exponential corresponding normalized exponential smoothing constraint basis function, and a composite function of a characteristic ratio exponential power transformation and / or a corresponding normalized exponent; The third construction module constructs a vegetation feature enhancement coefficient function according to the vegetation feature enhancement step function and the vegetation feature enhancement increment coefficient function; An enhancement module is used to enhance the vegetation features of the red band and the green band using a vegetation feature enhancement coefficient function; A synthesis module, used for synthesizing the enhanced red band, the enhanced green band and the original blue band to obtain a final color image; Among them, according to the composite function of the characteristic ratio exponential power transformation smooth constraint basis function, the vegetation feature enhancement coefficient function is constructed, which specifically includes: Construct a composite function of the characteristic ratio exponential power transformation smooth constraint basis function as follows: g1(x)=h(x)-h(c) Where h(x) = x n , h(c)=c n , n = 0.5, x is the characteristic ratio index, c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Among them, according to the composite function of the normalized exponential smoothing constraint basis function corresponding to the characteristic ratio index, the vegetation feature enhancement coefficient function is constructed, which specifically includes: Construct a composite function of the normalized exponential smoothing constraint basis function corresponding to the characteristic ratio index, as follows: g2(x)=NDhI(x)-NDhI(c) in, h(x)=x n , h((c)=c n , n = 0.5, x is the characteristic ratio index, c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Among them, according to the composite function of the characteristic ratio exponential power transformation and / or the corresponding normalized exponent, the vegetation feature enhancement increment coefficient function is constructed, which specifically includes: Construct a composite function of the characteristic exponential power transformation product mode as follows: g3(x)=[h1(x)-h1(c)][h2(x)-h2(c1)] in, n1=0.1,n2=0.1,c1∈[0,c],x is the characteristic ratio index,c is the characteristic threshold of the characteristic ratio index; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of the normalized exponential product mode of the characteristic exponential power transformation, as follows: g4(x)=[NDhI1(x)-NDhI1(c)][NDhI2(x)-NDhI2(c1)] is equivalent to: in, n1=6, n2=6, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows: g5(x)=[h1(x)-h1(c1))[NDhI2(x)-NDhI2(c)] is equivalent to: in, n1=0.1, n2=2, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential product mode, as follows: g6(x)=[h1(x)-h1(c)][NDhI2(x)-NDhI2(c1)] is equivalent to: in, n1=0.1, n2=6, c1∈[0,c]; Construct the vegetation feature enhancement coefficient function as follows: Wherein, k is the greenness adjustment coefficient, k = 0.5; Or construct a composite function of characteristic exponential power transformation and normalized exponential linear combination mode, as follows: g7(x)=k1[h1(x)-h1(c)]+k2[NDhI2(x)-NDhI2(c)] in, n1=0.5,n2=0.5,c k ∈[0,1 / (x max -c)]; Construct the vegetation feature enhancement coefficient function as follows: Among them, k is the greenness adjustment coefficient, k=0.

5.

6. A computer device comprising a processor and a memory for storing a program executable by the processor, characterized in that: When the processor executes the program stored in the memory, the remote sensing image fusion method described in any one of claims 1 to 4 is implemented.

7. A storage medium storing a program, characterized in that: When the program is executed by a processor, the remote sensing image fusion method described in any one of claims 1 to 4 is implemented.

Images