Enhancement processing method and device of remote sensing image for soil pollution observation

Abstract

The application provides an enhanced processing method and device for remote sensing images of soil pollution observation, comprising: acquiring remote sensing images of observed soil pollution; performing fast Fourier transform on the remote sensing images to obtain a frequency spectrum; extracting an edge intensity map of the remote sensing images to construct a spatial weight matrix; and applying the spatial weight matrix to the remote sensing images to obtain a spatial domain edge enhanced image; based on high-frequency components and low-frequency components in the frequency spectrum, enhancing the frequency spectrum to obtain an enhanced frequency spectrum; combining the enhanced frequency spectrum with an original phase spectrum of the original remote sensing image, and performing inverse Fourier transform to obtain a frequency domain enhanced image; fusing the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image; based on a brightness gradient map and an edge intensity map of the fused image, performing local contrast optimization processing on the fused image to obtain a final remote sensing image. The method effectively enhances the recognition accuracy and visual distinguishability of the soil pollution area.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to a remote sensing image enhancement processing method and device for soil pollution observation. BACKGROUND

[0002] Traditional remote sensing image enhancement techniques mainly include spatial domain methods and frequency domain methods. The spatial domain methods focus on local adjustment of image gray values, such as histogram equalization and edge enhancement filtering. Although the calculation is simple, the spatial domain methods are sensitive to noise and prone to over-enhancing local regions. The frequency domain methods analyze image spectral characteristics through Fourier transform and adjust the energy distribution of images at different frequencies to achieve detail enhancement and contrast improvement. However, the frequency domain methods have deficiencies in parameter selection and adaptive processing.

[0003] Therefore, the existing methods cannot simultaneously consider detail preservation and contrast optimization in the enhanced remote sensing images, which limits the application effect of remote sensing images in soil pollution detection. SUMMARY

[0004] The present application aims to provide a remote sensing image enhancement processing method and device for soil pollution observation, which can balance detail preservation and contrast optimization in remote sensing images, thereby improving the application effect of remote sensing images in soil pollution detection.

[0005] In a first aspect, the present application provides a remote sensing image enhancement processing method for soil pollution observation, which includes: obtaining an original remote sensing image for observing soil pollution conditions; performing fast Fourier transform on the original remote sensing image to obtain a frequency spectrum corresponding to the original remote sensing image; extracting an edge intensity map of the original remote sensing image through a Scharr operator to construct a spatial weight matrix; and applying the spatial weight matrix to the original remote sensing image through Hadamard product to obtain a spatial domain edge enhancement image; enhancing the frequency spectrum based on a first high-frequency component and a first low-frequency component in the frequency spectrum to obtain an enhanced frequency spectrum; combining the enhanced frequency spectrum with an original phase spectrum of the original remote sensing image and performing inverse Fourier transform to obtain a frequency domain enhancement image; fusing the frequency domain enhancement image and the spatial domain edge enhancement image to obtain a fused image; and performing local contrast optimization processing on the fused image based on a brightness gradient map and an edge intensity map of the fused image to obtain a final enhanced remote sensing image.

[0006] In the preferred embodiment of the present application, the step of enhancing the frequency spectrum based on the first high-frequency component and the first low-frequency component to obtain an enhanced frequency spectrum comprises: separating the first high-frequency component and the first low-frequency component in the frequency spectrum; sharpening the first high-frequency component and adjusting the contrast of the first low-frequency component to obtain a second high-frequency component and a second low-frequency component; and fusing and reconstructing the second high-frequency component and the second low-frequency component to obtain the enhanced frequency spectrum.

[0007] In the preferred embodiment of the present application, before the step of sharpening the first high-frequency component, the method comprises: centering the frequency spectrum to obtain a centered frequency spectrum; calculating the total energy of the centered frequency spectrum; calculating the total energy of the first high-frequency component; constructing a basic gain based on the total energy of the first high-frequency component and the total energy of the centered frequency spectrum; calculating the first amplitude mean value of all frequency points in the first high-frequency component to obtain a high-frequency mean value; constructing a local intensity term based on the amplitude values of the frequency points and the high-frequency mean value; and sharpening the first high-frequency component to obtain a second high-frequency component.

[0008] In the preferred embodiment of the present application, the basic gain is represented by the following formula: ; wherein, is the basic gain, is a global high-frequency gain weight coefficient, is the total energy of the first high-frequency component, is the total energy of the centered frequency spectrum, is a high-frequency component nonlinear adjustment index; the local intensity term is represented by the following formula: ; wherein, is the local intensity term, is a local high-frequency enhancement weight coefficient, is the amplitude value of the frequency point in the first high-frequency component, is the high-frequency mean value, is a minimum constant, is a local enhancement nonlinear adjustment index; the gain coefficient is represented by the following formula: ; wherein, is the gain coefficient, is the energy variance of the first high-frequency component, is a noise threshold parameter.

[0009] In the preferred embodiment of the present application, before the step of adjusting the contrast of the first low-frequency component, the method comprises: calculating the second amplitude mean and standard deviation of all frequency points in the first low-frequency component to obtain a low-frequency mean and a low-frequency standard deviation; constructing a dynamic compression kernel based on the Tukey double weight function according to the low-frequency mean and the low-frequency standard deviation; calculating the frequency domain gradient image of the first low-frequency component; constructing a frequency gradient suppression term according to the frequency domain gradient image; multiplying the dynamic compression kernel and the frequency gradient suppression term to obtain a dynamic compression coefficient; and the step of adjusting the contrast of the first low-frequency component comprises: adjusting the contrast of the first low-frequency component by the dynamic compression coefficient to obtain a second low-frequency component.

[0010] In the preferred embodiment of the present application, the dynamic compression kernel is represented by the following formula ; wherein, is the dynamic compression kernel, is the amplitude value of the frequency point in the first low-frequency component, is the low-frequency mean, is the low-frequency standard deviation, is a spectrum deviation sensitive factor; and the frequency gradient suppression term is represented by the following formula: ; wherein, is the frequency gradient suppression term, is a frequency domain gradient adjustment factor, is the gradient amplitude value of the frequency point in the low-frequency frequency amplitude image corresponding to the first low-frequency component.

[0011] In the preferred embodiment of the present application, the step of fusing the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image comprises: determining a frequency domain enhancement mask according to the frequency domain enhanced image; determining a spatial domain edge enhancement mask according to the spatial domain edge enhanced image; constructing a joint mask based on a first preset weight parameter according to the frequency domain enhancement mask and the spatial domain edge enhancement mask; and fusing the frequency domain enhanced image and the spatial domain edge enhanced image based on the joint mask to obtain the fused image.

[0012] In the preferred embodiment of the present application, the step of performing local contrast optimization processing on the fused image based on the brightness gradient image and the edge intensity image of the fused image to obtain a final enhanced remote sensing image comprises: generating an adaptive compression factor map based on the brightness gradient image and the edge intensity image of the fused image; and performing local contrast optimization on the fused image by the adaptive compression factor map to obtain the final enhanced remote sensing image.

[0013] In the preferred embodiment of the present application, based on the luminance gradient map and the edge intensity map of the fused image, the step of generating the adaptive compression factor map comprises: linearly fusing the luminance gradient map and the edge intensity map based on a preset weight parameter to obtain a fused feature map; performing smoothing processing on the fused feature map, and dynamically generating the adaptive compression factor map through a preset exponential function.

[0014] In a second aspect, the embodiments of the present application further provide an enhanced processing device for a remote sensing image for soil pollution observation, comprising: a data acquisition module configured to acquire an original remote sensing image for soil pollution observation; a processing module configured to perform fast Fourier transform on the original remote sensing image to obtain a frequency spectrum corresponding to the original remote sensing image; extract an edge intensity map of the original remote sensing image through a Scharr operator to construct a spatial weight matrix; and apply the spatial weight matrix to the original remote sensing image through Hadamard product to obtain a spatial domain edge enhanced image; enhance the frequency spectrum based on a first high frequency component and a first low frequency component in the frequency spectrum to obtain an enhanced frequency spectrum; combine the enhanced frequency spectrum with an original phase spectrum of the original remote sensing image, and perform inverse Fourier transform to obtain a frequency domain enhanced image; fuse the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image; and an output module configured to perform local contrast optimization processing on the fused image based on a luminance gradient map and an edge intensity map of the fused image to obtain a final enhanced remote sensing image.

[0015] The embodiments of the present application have the following beneficial technical effects:

[0016] The embodiment of the present application provides a kind of for soil pollution observation remote sensing image's enhancement processing method and device, comprising: obtaining the original remote sensing image of observation soil pollution condition;The original remote sensing image is carried out fast Fourier transform, and the frequency spectrum corresponding to the original remote sensing image is obtained;And, the edge intensity map of the original remote sensing image is extracted by Scharr operator to construct space weight matrix;And by Hadamard product, space weight matrix is applied to the original remote sensing image, and the spatial domain edge enhanced image is obtained;Based on the first high frequency component and the first low frequency component in the frequency spectrum, the frequency spectrum is enhanced, and the enhanced frequency spectrum is obtained;The original phase spectrum of the original remote sensing image is combined with the enhanced frequency spectrum, and inverse Fourier transform is carried out, and the frequency domain enhanced image is obtained;The fusion image is obtained by fusing the frequency domain enhanced image and the spatial domain edge enhanced image;Based on the brightness gradient map and the edge intensity map of the fusion image, the fusion image is carried out local contrast optimization processing, and the final enhanced remote sensing image is obtained.The method is fused by spatial domain edge enhancement and frequency domain multi-scale feature optimization, and combined with local contrast adjustment, the detail definition and dynamic range of remote sensing image are significantly improved, and the identification precision and visual distinguishability of soil pollution area are effectively enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0018] Figure 1 A flowchart of a remote sensing image enhancement processing method for soil pollution observation provided by an embodiment of the present application is shown in the figure.

[0019] Figure 2 A flowchart of another remote sensing image enhancement processing method for soil pollution observation provided by an embodiment of the present application is shown in the figure.

[0020] Figure 3 A schematic diagram of an original remote sensing image provided by an embodiment of the present application is shown in the figure.

[0021] Figure 4 A schematic diagram of an enhanced original remote sensing image provided by an embodiment of the present application is shown in the figure.

[0022] Figure 5 A structural schematic diagram of a remote sensing image enhancement processing device for soil pollution observation provided by an embodiment of the present application is shown in the figure.

[0023] Figure 6 A structural schematic diagram of an electronic device provided by an embodiment of the present application is shown.

[0024] Icon: 31-data acquisition module; 32-processing module; 33-output module; 41-memory; 42-processor; 43-bus; 44-communication interface. DETAILED DESCRIPTION

[0025] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0026] Traditional remote sensing image enhancement techniques mainly include spatial domain methods and frequency domain methods. The spatial domain methods focus on local adjustment of image gray values, such as histogram equalization and edge enhancement filtering. Although the calculation is simple, the spatial domain methods are sensitive to noise and prone to over-enhance local areas. The frequency domain methods analyze image spectral characteristics through Fourier transform and adjust the energy distribution of images at different frequencies to achieve detail enhancement and contrast improvement. However, the frequency domain methods have deficiencies in parameter selection and adaptive processing. Therefore, the enhanced remote sensing images obtained by existing methods are difficult to balance detail preservation and contrast optimization simultaneously, which limits the application effect of remote sensing images in soil pollution detection.

[0027] Based on this, the embodiments of the present application provide a remote sensing image enhancement processing method and device for soil pollution observation. The method fuses spatial domain edge enhancement and frequency domain multi-scale feature optimization, and combines local contrast adjustment, which significantly improves the detail clarity and dynamic range of remote sensing images, and effectively enhances the recognition accuracy and visual distinguishability of soil pollution areas. In order to facilitate understanding, first, a remote sensing image enhancement processing method for soil pollution observation is introduced.

[0028] Embodiment 1

[0029] In this embodiment, Figure 1 A flowchart of a remote sensing image enhancement processing method for soil pollution observation provided by an embodiment of the present application is shown.

[0030] As seen from Figure 1 the method comprises:

[0031] Step S101: obtaining an original remote sensing image for observing soil pollution.

[0032] Step S102: Fast Fourier transform is performed on the original remote sensing image to obtain a frequency spectrum corresponding to the original remote sensing image; an edge intensity map of the original remote sensing image is extracted by a Scharr operator to construct a spatial weight matrix; and the spatial weight matrix is applied to the original remote sensing image by Hadamard product to obtain a spatial domain edge enhanced image.

[0033] Step S103: The frequency spectrum is enhanced based on a first high-frequency component and a first low-frequency component in the frequency spectrum to obtain an enhanced frequency spectrum.

[0034] Here, the first high-frequency component is enhanced by a gain coefficient to perform sharpening enhancement, and the first low-frequency component is adjusted in contrast by a dynamic compression coefficient to obtain a second high-frequency component and a second low-frequency component, and fusion reconstruction is performed to obtain the enhanced frequency spectrum.

[0035] Step S104: The enhanced frequency spectrum is combined with an original phase spectrum of the original remote sensing image, and inverse Fourier transform is performed to obtain a frequency domain enhanced image.

[0036] Step S105: The frequency domain enhanced image and the spatial domain edge enhanced image are fused to obtain a fusion image.

[0037] In actual operation, the enhanced frequency spectrum is combined with the original phase spectrum to construct an enhanced frequency domain representation, inverse Fourier transform is performed to obtain a frequency domain enhanced image, a joint mask image is constructed, the joint mask image is obtained by weighted superposition of the frequency domain enhanced mask image and the spatial domain edge enhanced mask image, and the frequency domain enhanced image and the spatial domain edge enhanced image are fused by taking the constructed joint mask image as a weight to obtain the fusion image.

[0038] Step S106: The combined image is subjected to local contrast optimization processing based on a brightness gradient map and an edge intensity map of the fusion image to obtain a final enhanced remote sensing image.

[0039] Here, an adaptive compression factor map is constructed, the adaptive compression factor map is jointly generated from the brightness gradient map and the edge intensity map of the fusion image, the fusion image is subjected to local contrast optimization according to the constructed adaptive factor map, and a final enhanced remote sensing image is obtained.

[0040] The fusion image is also referred to as a frequency dual perceptual enhanced image.

[0041] The embodiment of the present application provides a kind of for the enhancement processing method of remote sensing image of soil pollution observation, comprising: obtaining the original remote sensing image of observing soil pollution condition;The fast Fourier transform is carried out to the above-mentioned original remote sensing image, and the frequency spectrum corresponding to the above-mentioned original remote sensing image is obtained;And, the edge intensity map of the above-mentioned original remote sensing image is extracted by Scharr operator to construct space weight matrix;And the space weight matrix is acted on the above-mentioned original remote sensing image by Hadamard product, and the spatial domain edge enhanced image is obtained;Based on the first high frequency component and the first low frequency component in the above-mentioned frequency spectrum, the above-mentioned frequency spectrum is enhanced, and the enhanced frequency spectrum is obtained;The original phase spectrum of the above-mentioned original remote sensing image is combined with the above-mentioned enhanced frequency spectrum, and inverse Fourier transform is carried out, and the frequency domain enhanced image is obtained;The above-mentioned frequency domain enhanced image and the above-mentioned spatial domain edge enhanced image are fused, and the fusion image is obtained;Based on the brightness gradient map and the edge intensity map of the above-mentioned fusion image, the above-mentioned fusion image is carried out local contrast optimization processing, and the final enhanced remote sensing image is obtained.The method is fused by spatial domain edge enhancement and frequency domain multi-scale feature optimization, and combined with local contrast adjustment, the detail definition and dynamic range of remote sensing image are significantly improved, and the identification precision and visual distinguishability of soil pollution area are effectively enhanced.

[0042] Embodiment 2

[0043] On the basis of the above-mentioned embodiment, Figure 2 Another kind of flow chart for the enhancement processing method of remote sensing image of soil pollution observation provided by the embodiment of the present application is shown.

[0044] As seen from Figure 2 The method comprises:

[0045] Step S201: obtaining the original remote sensing image of observing soil pollution condition.

[0046] Step S202: the fast Fourier transform is carried out to the above-mentioned original remote sensing image, and the frequency spectrum corresponding to the above-mentioned original remote sensing image is obtained;And, the edge intensity map of the above-mentioned original remote sensing image is extracted by Scharr operator to construct space weight matrix;And the space weight matrix is acted on the above-mentioned original remote sensing image by Hadamard product, and the spatial domain edge enhanced image is obtained.

[0047] Step S203: separating the above-mentioned first high frequency component and the above-mentioned first low frequency component in the above-mentioned frequency spectrum.

[0048] Step S204: the first high frequency component is sharpened and enhanced, and the first low frequency component is contrast adjusted, and the second high frequency component and the second low frequency component are obtained.

[0049] Step S205: the second high frequency component and the second low frequency component are fused to reconstruct an enhanced frequency spectrum.

[0050] Step S206: the enhanced frequency spectrum is combined with the original phase spectrum of the original remote sensing image, and inverse Fourier transform is performed to obtain a frequency domain enhanced image.

[0051] Step S207: the frequency domain enhanced image and the spatial domain edge enhanced image are fused to obtain a fused image.

[0052] Step S208: based on the brightness gradient map and the edge intensity map of the fused image, a local contrast optimization process is performed on the fused image to obtain a final enhanced remote sensing image.

[0053] Before the step of sharpening and enhancing the first high frequency component, the method comprises: centering the frequency spectrum to obtain a centered frequency spectrum; calculating the total energy of the centered frequency spectrum; calculating the total energy of the first high frequency component; constructing a basic gain according to the total energy of the first high frequency component and the total energy of the centered frequency spectrum; calculating the first amplitude mean value of all frequency points in the first high frequency component to obtain a high frequency mean value; constructing a local intensity term according to the amplitude value of the frequency point and the high frequency mean value; and performing the step of sharpening and enhancing the first high frequency component to obtain a second high frequency component, which comprises: constructing a gain coefficient through the basic gain and the local intensity term; and sharpening and enhancing the first high frequency component through the gain coefficient to obtain a second high frequency component.

[0054] Further, the basic gain is represented by the following formula: ; wherein, is the basic gain, is a global high frequency gain weight coefficient, is the total energy of the first high frequency component, is the total energy of the centered frequency spectrum, is a high frequency component nonlinear adjustment index;

[0055] The local intensity term is represented by the following formula: ; wherein, is the local intensity term, is a local high frequency enhancement weight coefficient, is the amplitude value of the frequency point in the first high frequency component, is the high frequency mean value, is a minimum constant, is a local enhancement nonlinear adjustment index;

[0056] The gain coefficient is represented by the following formula: ; wherein, is the gain coefficient, is the energy variance of the first high frequency component, is the noise threshold parameter.

[0057] Further, before the step of adjusting the contrast of the first low frequency component, the method comprises: calculating a second amplitude mean and a standard deviation of all frequency points in the first low frequency component to obtain a low frequency mean and a low frequency standard deviation; constructing a dynamic compression kernel based on a Tukey double weight function according to the low frequency mean and the low frequency standard deviation; calculating a frequency domain gradient map of the first low frequency component; constructing a frequency gradient suppression term according to the frequency domain gradient map; multiplying the dynamic compression kernel and the frequency gradient suppression term to obtain a dynamic compression coefficient; and the step of adjusting the contrast of the first low frequency component comprises: adjusting the contrast of the first low frequency component by the dynamic compression coefficient to obtain a second low frequency component.

[0058] Further, the dynamic compression kernel is represented by the following formula: ; wherein, is the dynamic compression kernel, is the amplitude value of the frequency point in the first low frequency component, is the low frequency mean, is the low frequency standard deviation, is a spectrum deviation sensitive factor; and the frequency gradient suppression term is represented by the following formula: ; wherein, is the frequency gradient suppression term, is a frequency domain gradient adjustment factor, is the gradient amplitude value of the frequency point in the low frequency amplitude map corresponding to the first low frequency component.

[0059] The step of fusing the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image comprises: determining a frequency domain enhancement mask according to the frequency domain enhanced image; determining a spatial domain edge enhancement mask according to the spatial domain edge enhanced image; constructing a joint mask based on a first preset weight parameter according to the frequency domain enhancement mask and the spatial domain edge enhancement mask; and fusing the frequency domain enhanced image and the spatial domain edge enhanced image based on the joint mask to obtain the fused image.

[0060] Finally, based on the brightness gradient map and the edge intensity map of the fusion image, a step of performing local contrast optimization processing on the fusion image to obtain a final enhanced remote sensing image, comprising: generating an adaptive compression factor map based on the brightness gradient map and the edge intensity map of the fusion image; and performing local contrast optimization on the fusion image through the adaptive compression factor map to obtain the final enhanced remote sensing image.

[0061] Here, the step of generating an adaptive compression factor map based on the brightness gradient map and the edge intensity map of the fusion image comprises: performing linear fusion on the brightness gradient map and the edge intensity map based on a preset weight parameter to obtain a fused feature map; performing smoothing processing on the fused feature map, and dynamically generating the adaptive compression factor map through a preset exponential function.

[0062] The embodiment of the present application provides a kind of for soil pollution observation remote sensing image enhancement processing method, comprising: obtaining the original remote sensing image of observation soil pollution situation;The Fourier transform of the original remote sensing image is carried out, and the frequency spectrum corresponding to the original remote sensing image is obtained;And, the edge intensity map of the original remote sensing image is extracted by Scharr operator to construct spatial weight matrix;And spatial weight matrix is acted on the original remote sensing image by Hadamard product, and spatial domain edge enhanced image is obtained;The first high frequency component and the first low frequency component in the frequency spectrum are separated;The first high frequency component is sharpened and enhanced, and the first low frequency component is contrast adjusted, to obtain second high frequency component and second low frequency component;The second high frequency component and the second low frequency component are fused and reconstructed, to obtain enhanced frequency spectrum;The original phase spectrum of the enhanced frequency spectrum and the original remote sensing image is combined, and inverse Fourier transform is carried out, to obtain frequency domain enhanced image;Fusion frequency domain enhanced image and spatial domain edge enhanced image are obtained, to obtain fusion image;Based on the brightness gradient map and the edge intensity map of the fusion image, the fusion image is processed by local contrast optimization, to obtain the final enhanced remote sensing image.The method is fused spatial domain edge enhancement and frequency domain high frequency sharpening and low frequency contrast adjustment, and combined with local contrast optimization, which significantly improves the detail definition, contrast and edge sharpness of remote sensing image, effectively enhances the distinguishability and observation accuracy of soil pollution characteristics.

[0063] Embodiment 3

[0064] In order to facilitate the understanding of the present application, the following actual case is described in detail. The present application includes the following steps in specific implementation:

[0065] Step S1: Fast Fourier transform is performed on the original remote sensing image to obtain a frequency spectrum corresponding to the original remote sensing image, spectral energy analysis is performed according to energy distribution characteristics in the frequency spectrum, and a first high-frequency component and a first low-frequency component of the frequency spectrum are adaptively separated.

[0066] Step S2: A gain coefficient is used for the first high-frequency component to perform sharpening enhancement, and a dynamic compression coefficient is used for the first low-frequency component to perform contrast adjustment, the second high-frequency component and the second low-frequency component after enhancement are fused and reconstructed to obtain an enhanced frequency spectrum.

[0067] Step S3: An edge intensity map of the original remote sensing image is extracted using a Scharr operator, a spatial weight matrix is constructed based on the edge intensity map, and the spatial weight matrix is applied to the original remote sensing image through Hadamard multiplication to obtain a spatial domain edge enhancement image.

[0068] Step S4: The enhanced frequency spectrum is combined with an original phase spectrum to construct an enhanced frequency domain representation, a frequency domain enhancement image is obtained through inverse Fourier transform, a joint mask image is constructed, the joint mask image is obtained through weighted superposition of a frequency domain enhancement mask image and a spatial domain edge enhancement mask image, and the frequency domain enhancement image and the spatial domain edge enhancement image are fused by taking the constructed joint mask image as a weight to obtain a fused image.

[0069] Step S5: Because soil pollution often appears as a mixture of weak high-frequency characteristics and low-frequency background in remote sensing images, directly enhancing the full frequency band can cause noise amplification or loss of details, therefore, the application realizes differentiated processing of high-frequency and low-frequency through adaptive frequency spectrum separation, and step S1 specifically comprises: inputting an original remote sensing image, using a two-dimensional fast Fourier transform algorithm to convert the original remote sensing image from a spatial domain to a frequency domain to obtain a frequency domain representation, extracting an amplitude spectrum from the frequency domain representation to obtain a frequency spectrum graph; using the conjugate symmetry of the frequency spectrum, performing centering processing on the frequency spectrum graph, and moving the zero-frequency component to the center position of the frequency spectrum graph through quadrant exchange, so that the low-frequency component is concentrated in the center of the frequency spectrum graph and the high-frequency component is distributed in the edge area, to obtain a centered frequency spectrum graph; based on the centered frequency spectrum graph, calculating the energy of each frequency point and the distance of each frequency point to the center of the frequency spectrum to obtain a radial distance matrix, grouping the energy of each frequency point in the frequency spectrum graph according to the radial distance matrix, and counting the total energy of each radial distance interval to obtain a frequency spectrum energy distribution function; in order of increasing radial distance, accumulating the energy of each radial distance interval in the frequency spectrum energy distribution function, and performing normalization processing on the accumulation result to obtain a normalized cumulative energy distribution curve; according to the soil pollution detection requirement, setting an energy accumulation threshold, searching from the normalized cumulative energy distribution curve through binary search to find the minimum radial distance that satisfies the normalized cumulative energy first reaching or exceeding the energy accumulation threshold, using an optimized binary search algorithm, the whole process can be completed in seconds, taking the minimum radial distance as an adaptive separation frequency radius of the frequency spectrum energy distribution, through automatic calculation of the separation frequency radius, the original remote sensing image with different resolution and different pollution characteristics can be adapted to, without manual pre-adjustment of parameters, the separation frequency is determined through the normalized cumulative energy distribution curve, which can effectively avoid the influence of individual high-frequency noise points on the separation result; constructing a frequency domain mask graph, the frequency domain mask graph comprises a low-frequency mask and a high-frequency mask, the construction method of the low-frequency mask is that when the radial distance corresponding to a frequency point is less than or equal to the adaptive separation frequency radius, the mask value corresponding to the point is set to 1, otherwise it is set to 0, the high-frequency mask is the complementary area of the low-frequency mask, and the mask value of each frequency point in the high-frequency mask is 1 minus the low-frequency mask value at the corresponding position; performing point-by-point multiplication operation on the centered frequency spectrum graph and the low-frequency mask and the high-frequency mask respectively to extract the low-frequency component and the high-frequency component in the frequency spectrum graph.

[0070] In some examples, the first high-frequency component contains edge, texture and other detail characteristics of the original remote sensing image, which is easily disturbed by noise, and through calculation of the high-frequency energy proportion to dynamically adjust the gain coefficient, the noise can be suppressed while enhancing the detail characteristics.

[0071] Therefore, the first high frequency component is sharpened and enhanced in step S2, specifically including: inputting the centralized spectrum diagram obtained in step S1, calculating the modulus square of each frequency point in the centralized spectrum diagram, summing the modulus square of all frequency points to obtain the total energy of the centralized spectrum diagram; according to the global distribution characteristics of the spectrum energy, a basic gain is constructed as the core adjustment factor of the first high frequency component enhancement, by introducing a nonlinear mapping function, the adaptive enhancement of high frequency energy is realized, and the problem of excessive enhancement of noise caused by direct linear amplification is avoided, the high frequency mask and the high frequency component obtained in step S1 are input, the energy of each frequency point in the first high frequency component is calculated, the energy of each frequency point is weighted by the high frequency mask, the weighted energy of all frequency points is summed to obtain the total energy of the first high frequency component, the ratio of the total energy of the high frequency component and the total energy of the centralized spectrum diagram is calculated to obtain the first high frequency energy proportion, the basic gain is constructed according to the high frequency energy proportion, and the mathematical model of the basic gain is: ; wherein, is the basic gain, is the global high frequency gain weight coefficient, is the total energy of the first high frequency component, is the total energy of the above-mentioned centralized spectrum diagram, is the high frequency component nonlinear regulation index, to control the nonlinear characteristics of the gain mapping of the first high frequency energy proportion; a local intensity term is constructed to realize the non-uniform enhancement of the first high frequency component, by quantifying the deviation degree of each frequency point in the first high frequency component relative to the high frequency mean, the significant edges and texture features of the original remote sensing image are adaptively strengthened, the amplitude mean of all frequency points in the first high frequency component is calculated to obtain the high frequency mean, for each frequency point in the first high frequency component, the normalized ratio of the amplitude value of the frequency point and the high frequency mean is calculated, and the local intensity term is constructed according to the normalized ratio, and the mathematical model of the local intensity term is: ; wherein, is the local intensity term, is the local high frequency enhancement weight coefficient, is the amplitude value of the frequency point in the above-mentioned first high frequency component, is the above-mentioned high frequency mean, is a minimum constant, is a local enhancement nonlinear regulation index, which determines the sensitivity and curve shape of high frequency feature enhancement; the gain coefficient dynamically fuses the basic gain and the local intensity term through a noise adaptive weighting mechanism, sets a noise threshold parameter, the setting of the noise threshold needs to be dynamically adjusted according to the spectral characteristics of the soil target pollutant, calculates the energy variance of the high frequency component as a noise evaluation index, constructs an exponential noise suppression factor, and the basic gain and the local gain term are adaptively weighted according to the exponential noise suppression factor to obtain the gain coefficient, and the mathematical model of the gain coefficient is: ; wherein, is a gain coefficient, is an energy variance of the first high-frequency component, is a noise threshold parameter; the first high-frequency component is adaptively sharpened according to the gain coefficient to obtain a second high-frequency component.

[0072] Further, the first low-frequency component mainly reflects the overall brightness and contrast of the original remote sensing image, but there are areas of radiation anomaly or excessive smoothing. By constructing a dynamic compression kernel item through a Tukey double weight function, the abnormal values deviating from the low-frequency mean value can be adaptively suppressed, and the gentle transition characteristics are retained. In step S2, the contrast of the first low-frequency component is adjusted, specifically including:

[0073] The low-frequency mask and the first low-frequency component obtained in step S1 are input, the amplitude mean and standard deviation of all frequency points in the low-frequency component are calculated, and the low-frequency mean and low-frequency standard deviation are obtained. A dynamic compression kernel item is constructed by introducing a Tukey double weight function. For each frequency point in the low-frequency component, the low-frequency component is nonlinearly and adaptively optimized according to the degree of deviation of the low-frequency amplitude from the low-frequency mean value. The mathematical model of the dynamic compression kernel item is: ; wherein, is a dynamic compression kernel item, is the amplitude value of the frequency point in the low-frequency component, is a low-frequency mean, is a low-frequency standard deviation, is a frequency spectrum deviation sensitive factor, which dynamically adjusts the compression range and intensity of the low-frequency component by setting a threshold; the amplitude of the first low-frequency component is extracted to obtain a low-frequency frequency amplitude map. The gradient amplitude value of each frequency point in the low-frequency frequency amplitude map in the frequency domain direction is calculated using a difference operator to obtain a frequency domain gradient map. A frequency gradient suppression item is constructed according to the frequency domain gradient map. The mathematical model of the frequency gradient suppression item is: ; wherein, is a frequency gradient suppression item, is a frequency domain gradient adjustment factor for controlling the smoothing intensity of the low-frequency component in the frequency domain gradient direction, is the gradient amplitude value of the frequency point in the low-frequency frequency amplitude map; the dynamic compression kernel item and the frequency gradient suppression item are point multiplied to obtain a dynamic compression coefficient. The mathematical model of the dynamic compression coefficient is ; wherein, is a dynamic compression coefficient; the contrast of the low-frequency component is adjusted according to the dynamic compression coefficient to obtain an enhanced low-frequency component.

[0074] Further, the spatial domain edge enhancement of the original remote sensing image extracts an edge intensity map through a Scharr operator and constructs a spatial weight matrix, which can highlight the edges and structure information in the original remote sensing image. Step S3 specifically includes:

[0075] The original remote sensing image is input, and a Scharr operator is used to perform edge detection on the original remote sensing image. A horizontal gradient convolution kernel of the Scharr operator is used to perform convolution operation on the original remote sensing image to obtain a gradient component in the horizontal direction. A vertical gradient convolution kernel of the Scharr operator is used to perform convolution operation on the original remote sensing image to obtain a gradient component in the vertical direction. The square sum of the gradient component in the horizontal direction and the gradient component in the vertical direction is squared to obtain the edge intensity of each pixel point in the original remote sensing image. The edge intensities of all pixel points of the original remote sensing image constitute an edge intensity map.

[0076] Further, the edge intensity map is normalized to construct a spatial weight matrix. The spatial weight matrix uses an S-shaped curve function to perform nonlinear adjustment on the normalized edge intensity map, so that the weight of the edge region is close to 1 and the weight of the smooth region is close to 0.

[0077] Then, the spatial weight matrix is multiplied point by point with the original remote sensing image to obtain a spatial domain edge enhancement image.

[0078] Further, the frequency domain enhancement image and the spatial domain edge enhancement image respectively improve the quality of the original remote sensing image from different angles, but each has its own advantages. By constructing a joint mask image, the advantages of the two images can be dynamically mixed. Step S4 specifically includes:

[0079] The enhanced frequency spectrum obtained in step S2 is input, and the enhanced frequency spectrum is combined with the phase spectrum represented in the frequency domain obtained in step S1 to obtain an enhanced frequency domain representation. The enhanced frequency domain representation is inverse Fourier transformed to obtain a frequency domain enhancement image. A frequency domain enhancement mask image is constructed, and the frequency domain enhancement image is Gaussian pyramid decomposed to generate image layers at different resolutions. Each layer of image is obtained through Gaussian filtering and down-sampling. On each pyramid level, a Laplacian convolution kernel is used to perform convolution operation to obtain a Laplacian response matrix of each level. Each response matrix is normalized, and the global histogram of the normalized response matrix is calculated to determine the 25% quantile and the 75% quantile The upper threshold and the lower threshold are calculated. , , the region with a response value greater than or equal to the upper threshold value is divided into a high response area, the high response area is given a fixed weight, the region with a response value between the upper and lower threshold values is divided into a transition area, the transition area adopts a linear weight mapping, the region with a response value less than or equal to the lower threshold value is divided into a smooth area, and the weight of the smooth area is set to 0; the spatial domain edge enhancement image obtained in the input step S3 is used to construct a spatial domain edge enhancement mask image, a circular LBP operator with a radius of 1, 2 and 3 pixels is used on the spatial domain edge enhancement image, a multi-scale LBP image is calculated, three scales of LBP images are fused, for each pixel in the spatial domain edge enhancement image, the difference between the maximum gray value and the minimum gray value of the neighborhood is calculated, all the difference values are normalized to obtain a local contrast image, the multi-scale LBP image and the local contrast image are linearly combined according to a certain weight to obtain a feature image, the Ostu algorithm is used to automatically determine the best segmentation threshold value, and the feature image is thresholded according to the threshold value, the region greater than the threshold value is masked with a value of 1, and the remaining regions are masked with a value of 0; the constructed frequency domain enhancement mask image and the spatial domain edge enhancement mask image are weighted and superimposed to obtain a joint mask image, and for each pixel, the frequency domain enhancement image and the spatial domain edge enhancement image are dynamically mixed according to the weight value of the joint mask image to obtain a fusion image. Figure Two

[0080] The fusion image generates an adaptive compression factor map through the fusion of the brightness gradient map and the edge intensity map, and can dynamically adjust the contrast of different regions. Step S5 specifically includes:

[0081] The fusion image obtained in step S4 is input, and the Sobel operator is used to calculate the horizontal gradient and the vertical gradient of each pixel, the gradient amplitude of each pixel is calculated according to the horizontal gradient and the vertical gradient, the gradient amplitude of each pixel is normalized, and the normalized gradient amplitudes of all pixels constitute a brightness gradient; the Scharr operator is used for edge detection of the fusion image, the horizontal gradient kernel and the vertical gradient kernel are used for convolution operation on the fusion image respectively to obtain the horizontal gradient and the vertical gradient, the edge intensity of each pixel point is calculated according to the horizontal gradient and the vertical gradient, the edge intensity of each pixel point is normalized, and the normalized edge intensities of all pixel points constitute an edge intensity map; the brightness gradient map and the edge intensity map are linearly fused according to a weight coefficient through a set weight coefficient to obtain a fused feature map, the fused feature map is subjected to Gaussian smoothing filtering, and an adaptive compression factor map is dynamically generated through an exponential function according to the smoothed feature map; for each pixel of the fusion image, a fixed size field window is set with each pixel point as the center, the gray mean value of all pixels in the field window is calculated, and the contrast of the fusion image is dynamically adjusted according to the adaptive compression factor map and the gray mean value to obtain an enhanced original remote sensing image.

[0082] In order to facilitate understanding, Figure 3 ​A schematic diagram of an original remote sensing image provided by an embodiment of the present application; Figure 4 A schematic diagram of an enhanced original remote sensing image provided by an embodiment of the present application.

[0083] After comparison of Figure 3 and Figure 4 It can be seen that the enhanced original remote sensing image improves the detail definition, contrast and edge sharpness of the remote sensing image compared with the original remote sensing image.

[0084] Embodiment 3

[0085] On the basis of the above embodiments, Figure 5 A structural schematic diagram of an enhanced processing device for a remote sensing image for soil pollution observation provided by an embodiment of the present application.

[0086] As seen from Figure 5 , the device comprises:

[0087] A data acquisition module 31 for acquiring an original remote sensing image for soil pollution observation.

[0088] A processing module 32 for performing fast Fourier transform on the original remote sensing image to obtain a frequency spectrum corresponding to the original remote sensing image; and extracting an edge intensity map of the original remote sensing image by a Scharr operator to construct a spatial weight matrix; and applying the spatial weight matrix to the original remote sensing image by Hadamard product to obtain a spatial domain edge enhanced image; enhancing the frequency spectrum based on a first high frequency component and a first low frequency component in the frequency spectrum to obtain an enhanced frequency spectrum; combining the enhanced frequency spectrum with an original phase spectrum of the original remote sensing image and performing inverse Fourier transform to obtain a frequency domain enhanced image; and fusing the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image.

[0089] An output module 33 for performing local contrast optimization processing on the fused image based on a brightness gradient map and an edge intensity map of the fused image to obtain a final enhanced remote sensing image.

[0090] Among them, the data acquisition module 31, the processing module 32 and the output module 33 are connected in sequence.

[0091] In one embodiment, the processing module 32 is further configured to separate the first high frequency component and the first low frequency component in the frequency spectrum; sharpen the first high frequency component and adjust the contrast of the first low frequency component to obtain a second high frequency component and a second low frequency component; and fuse and reconstruct the second high frequency component and the second low frequency component to obtain the enhanced frequency spectrum.

[0092] In one of the embodiments, the processing module 32 is further configured to center the spectrum diagram to obtain a centered spectrum diagram, calculate a total energy of the centered spectrum diagram, calculate a total energy of the first high-frequency component, construct a base gain according to the total energy of the first high-frequency component and the total energy of the centered spectrum diagram, calculate a first amplitude mean value of all frequency points in the first high-frequency component to obtain a high-frequency mean value, construct a local intensity term according to the amplitude values of the frequency points and the high-frequency mean value, and sharpen the first high-frequency component to obtain a second high-frequency component.

[0093] In one of the embodiments, the processing module 32 is further configured to calculate a second amplitude mean value and a standard deviation of all frequency points in the first low-frequency component to obtain a low-frequency mean value and a low-frequency standard deviation, construct a dynamic compression kernel term based on a Tukey double weight function according to the low-frequency mean value and the low-frequency standard deviation, calculate a frequency domain gradient diagram of the first low-frequency component, and adjust the contrast of the first low-frequency component by the dynamic compression coefficient to obtain a second low-frequency component.

[0094] In one of the embodiments, the processing module 32 is further configured to determine a frequency domain enhancement mask according to the frequency domain enhancement image, determine a spatial domain edge enhancement mask according to the spatial domain edge enhancement image, construct a joint mask based on a first preset weight parameter according to the frequency domain enhancement mask and the spatial domain edge enhancement mask, and fuse the frequency domain enhancement image and the spatial domain edge enhancement image based on the joint mask to obtain the fusion image.

[0095] In one of the embodiments, the output module 33 is further configured to generate an adaptive compression factor map based on a brightness gradient map and an edge intensity map of the fusion image, and perform local contrast optimization on the fusion image by the adaptive compression factor map to obtain a final enhanced remote sensing image.

[0096] In one of the embodiments, the output module 33 is further configured to perform linear fusion on the brightness gradient map and the edge intensity map based on a preset weight parameter to obtain a fused feature map, perform smoothing processing on the fused feature map, and dynamically generate the adaptive compression factor map by a preset exponential function.

[0097] The implementation principle and the technical effects of the device for enhancing remote sensing images for soil pollution observation provided in the embodiment of the present application are the same as those of the method embodiment, and for brief description, the part of the device for enhancing remote sensing images for soil pollution observation not mentioned in the embodiment of the device is described with reference to the corresponding content in the method embodiment.

[0098] Embodiment 4

[0099] The embodiment provides an electronic device, including a processor and a memory, the memory stores computer executable instructions capable of being executed by the processor, and the processor executes the computer executable instructions to implement the steps of the defect detection method of the municipal pipeline.

[0100] The embodiment provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the defect detection method of the municipal pipeline.

[0101] Referring to Figure 6 The electronic device includes a memory 41 and a processor 42, the memory 41 stores a computer program capable of running on the processor 42, and the processor implements the steps provided by the defect detection method of the municipal pipeline when executing the computer program.

[0102] As Figure 6 The device further includes a bus 43 and a communication interface 44, the processor 42, the communication interface 44 and the memory 41 are connected through the bus 43, and the processor 42 is used to execute the executable modules stored in the memory 41, for example, the computer program.

[0103] The memory 41 can include a high-speed random access memory (RAM) and can also include a non-volatile memory, for example, at least one disk memory. The communication connection between the device network element and at least one other network element is realized through at least one communication interface 44 (which can be wired or wireless), and the Internet, a wide area network, a local area network, a metropolitan area network, etc. can be used.

[0104] The bus 43 can be an ISA bus, a PCI bus or an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 6 Only one bidirectional arrow is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0105] The memory 41 is configured to store a program, and the processor 42 is configured to execute the program after receiving an execution instruction. The method performed by the defect detection device for the municipal pipeline disclosed in any of the embodiments of the present application can be applied to the processor 42 or implemented by the processor 42. The processor 42 can be an integrated circuit chip with a signal processing capability. In the implementation process, each step of the above method can be completed by the integrated logic circuit of hardware in the processor 42 or the instruction in the form of software. The processor 42 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; or a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Each method, step and logic block disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register or other mature storage medium in the art. The storage medium is located in the memory 41, and the processor 42 reads the information in the memory 41 and combines the hardware to complete the steps of the above method.

[0106] Further, the embodiments of the present application also provide a machine readable storage medium, which stores machine executable instructions. When the machine executable instructions are called and executed by the processor 42, the machine executable instructions cause the processor 42 to implement the above defect detection method for the municipal pipeline.

[0107] The electronic device and the computer readable storage medium provided by the embodiments of the present application have the same technical features, so they can also solve the same technical problems and achieve the same technical effects.

[0108] In the description of the application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the application and simplifying the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application. In addition, the terms "first", "second", "third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

Claims

1. A method for enhancing remote sensing images used for soil pollution observation, characterized in that, include: Acquire raw remote sensing images of soil pollution; Perform a fast Fourier transform on the original remote sensing image to obtain the spectrum corresponding to the original remote sensing image; Furthermore, the edge intensity map of the original remote sensing image is extracted using the Scharr operator to construct a spatial weight matrix; and the spatial weight matrix is ​​applied to the original remote sensing image using the Hadamard product to obtain a spatial domain edge-enhanced image. Based on the first high-frequency component and the first low-frequency component in the spectrum, the spectrum is enhanced to obtain an enhanced spectrum. The enhanced spectrogram is combined with the original phase spectrum of the original remote sensing image, and an inverse Fourier transform is performed to obtain a frequency domain enhanced image. The frequency domain enhanced image and the spatial domain edge enhanced image are fused to obtain a fused image; Based on the brightness gradient map and edge intensity map of the fused image, local contrast optimization processing is performed on the fused image to obtain the final enhanced remote sensing image; The step of enhancing the spectrum based on the first high-frequency component and the first low-frequency component in the spectrum to obtain the enhanced spectrum includes: Separate the first high-frequency component and the first low-frequency component in the spectrum; The first high-frequency component is sharpened and enhanced, and the contrast of the first low-frequency component is adjusted to obtain the second high-frequency component and the second low-frequency component. The second high-frequency component and the second low-frequency component are fused and reconstructed to obtain the enhanced spectrum.

2. The method for enhancing remote sensing images for soil pollution observation according to claim 1, characterized in that, Before the step of sharpening and enhancing the first high-frequency component, the method includes: The spectrum is centered to obtain a centered spectrum. Calculate the total energy of the centered spectrum; Calculate the total energy of the first high-frequency component; The basic gain is constructed based on the total energy of the first high-frequency component and the total energy of the centered spectrum. Calculate the first amplitude mean of all frequency points in the first high-frequency component to obtain the high-frequency mean; Based on the amplitude value at the frequency point and the high-frequency mean, a local intensity term is constructed; The step of sharpening and enhancing the first high-frequency component to obtain the second high-frequency component includes: The gain coefficients are constructed using the base gain and the local intensity terms. The first high-frequency component is sharpened and enhanced using the gain coefficient to obtain the second high-frequency component.

3. The method for enhancing remote sensing images for soil pollution observation according to claim 2, characterized in that, The fundamental gain is expressed by the following formula: ； in, It is the base gain. It is the global high-frequency gain weighting coefficient. It is the total energy of the first high-frequency component. It is the total energy of the centralized spectrum. It is the nonlinear adjustment index of the high-frequency component; The local strength term is expressed by the following formula: ； in, It is a local strength term. It is a local high-frequency enhancement weighting coefficient. It is the amplitude value of the frequency point in the first high-frequency component. It is the high-frequency mean. It is a very small constant. It is a locally enhanced nonlinear adjustment index; The gain coefficient is expressed by the following formula: in, It is the gain coefficient, It is the energy variance of the first high-frequency component. It is the noise threshold parameter.

4. The method for enhancing remote sensing images for soil pollution observation according to claim 1, characterized in that, Before the step of contrast adjustment of the first low-frequency component, the method includes: Calculate the second amplitude mean and standard deviation of all frequency points in the first low-frequency component to obtain the low-frequency mean and low-frequency standard deviation; Based on the Tukey double-weight function, a dynamic compression kernel term is constructed according to the low-frequency mean and the low-frequency standard deviation; Calculate the frequency domain gradient plot of the first low-frequency component; Based on the frequency domain gradient plot, a frequency gradient suppression term is constructed; Multiply the dynamic compression kernel term and the frequency gradient suppression term by the dot product to obtain the dynamic compression coefficient; The step of adjusting the contrast of the first low-frequency component includes: The contrast of the first low-frequency component is adjusted using the dynamic compression coefficient to obtain the second low-frequency component.

5. The method for enhancing remote sensing images for soil pollution observation according to claim 4, characterized in that, The dynamic compression kernel term is expressed by the following formula: in, It is the aforementioned dynamic compression kernel term. It is the amplitude value at a frequency point in the first low-frequency component. It is the low-frequency mean. It is the low-frequency standard deviation, It is a spectral deviation sensitive factor; The frequency gradient suppression term is expressed by the following formula: ； in, It is the frequency gradient suppression term. It is a frequency domain gradient adjustment factor. It is the gradient amplitude value of the frequency point in the low-frequency amplitude diagram corresponding to the first low-frequency component.

6. The method for enhancing remote sensing images for soil pollution observation according to claim 1, characterized in that, The step of fusing the frequency domain enhanced image and the spatial domain edge enhanced image to obtain a fused image includes: Based on the frequency domain enhanced image, a frequency domain enhanced mask is determined; and based on the spatial domain edge enhanced image, a spatial domain edge enhanced mask is determined. Based on the first preset weight parameters, a joint mask map is constructed according to the frequency domain enhancement mask map and the spatial domain edge enhancement mask map; The frequency domain enhanced image and the spatial domain edge enhanced image are fused based on the joint mask image to obtain the fused image.

7. The method for enhancing remote sensing images for soil pollution observation according to claim 1, characterized in that, The steps of performing local contrast optimization processing on the fused image based on the brightness gradient map and edge intensity map of the fused image to obtain the final enhanced remote sensing image include: An adaptive compression factor map is generated based on the brightness gradient map and edge intensity map of the fused image; The fused image is then subjected to local contrast optimization using the adaptive compression factor map to obtain the final enhanced remote sensing image.

8. The method for enhancing remote sensing images for soil pollution observation according to claim 7, characterized in that, The step of generating an adaptive compression factor map based on the brightness gradient map and edge intensity map of the fused image includes: Based on preset weight parameters, the brightness gradient map and the edge intensity map are linearly fused to obtain a fused feature map; The fused feature map is smoothed, and the adaptive compression factor map is dynamically generated using a preset exponential function.

9. An enhancement processing device for remote sensing images used for soil pollution observation, characterized in that, include: The data acquisition module is used to acquire raw remote sensing images for soil pollution observation. The processing module is used to perform a fast Fourier transform on the original remote sensing image to obtain the spectrum map corresponding to the original remote sensing image; and to extract the edge intensity map of the original remote sensing image through the Scharr operator to construct a spatial weight matrix; and to apply the spatial weight matrix to the original remote sensing image through the Hadamard product to obtain a spatial domain edge enhancement image. Based on the first high-frequency component and the first low-frequency component in the spectrum, the spectrum is enhanced to obtain an enhanced spectrum. The enhanced spectrogram is combined with the original phase spectrum of the original remote sensing image, and an inverse Fourier transform is performed to obtain a frequency domain enhanced image. The frequency domain enhanced image and the spatial domain edge enhanced image are fused to obtain a fused image; The output module is used to perform local contrast optimization processing on the fused image based on the brightness gradient map and edge intensity map of the fused image to obtain the final enhanced remote sensing image; The processing module is further configured to: separate the first high-frequency component and the first low-frequency component in the spectrum; sharpen and enhance the first high-frequency component and adjust the contrast of the first low-frequency component to obtain the second high-frequency component and the second low-frequency component. The second high-frequency component and the second low-frequency component are fused and reconstructed to obtain the enhanced spectrum.

Images