A remote sensing image enhancement method and system based on homomorphic filtering terrain correction

Abstract

The present invention relates to the technical field of remote sensing image enhancement, and discloses a remote sensing image enhancement method and system based on homomorphic filtering terrain correction. The method comprises: after obtaining the remote sensing image to be enhanced, converting it from the RGB color space to the HSV color space, and extracting the brightness component. Based on a high-pass filter, homomorphic filtering is performed on the brightness component to enhance the high-frequency component and suppress the low-frequency component. The processed brightness component is combined with the saturation and color, and inverted back to the RGB color space, thereby completing the image enhancement. Then, the grayscale distribution of the enhanced image is obtained, and the image information entropy is calculated based on its grayscale distribution. The contrast of the image is further calculated by the grayscale value of each pixel. Based on the information entropy and contrast, the image enhancement score is calculated, and compared with the preset score to determine whether the image enhancement is completed. If the preset score is not reached, the homomorphic filtering process is performed again until the standard is met. The present invention can effectively improve the image quality of remote sensing images.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image enhancement, and in particular to a remote sensing image enhancement method and system based on homomorphic filtering terrain correction. Background Art

[0002] The quality of remote sensing images directly impacts the accuracy of image analysis and interpretation, thus image enhancement plays a crucial role in remote sensing image processing. Remote sensing images are often affected by imaging conditions, sensor characteristics, and atmospheric and topographical factors, resulting in image noise, low contrast, and loss of detail. Traditional image enhancement methods, such as histogram equalization and contrast stretching, while able to improve overall image quality, often fail to effectively enhance detail in distinct image regions and are prone to introducing artifacts in high-noise environments, compromising image accuracy.

[0003] Homomorphic filtering, a method that uses frequency-domain filtering to enhance specific frequency components in an image, is widely used to enhance image brightness and contrast. By separating low-frequency and high-frequency components, homomorphic filtering can enhance image detail and is particularly suitable for remote sensing images with uneven illumination or low contrast. However, traditional homomorphic filtering methods often rely on manually set filtering parameters and can result in over- or under-enhancement of the image during processing, lacking automation and precision.

[0004] Therefore, there is an urgent need to invent a remote sensing image enhancement technology to solve the problem that when remote sensing images are enhanced by homomorphic filtering in the existing technology, the influence of terrain cannot be effectively eliminated, resulting in low image quality. Summary of the Invention

[0005] In view of this, the present invention proposes a remote sensing image enhancement method and system based on homomorphic filtering terrain correction, aiming to solve the problem of low image quality caused by the inability to effectively eliminate the terrain influence when enhancing remote sensing images through homomorphic filtering in current technology.

[0006] The present invention proposes a remote sensing image enhancement method based on homomorphic filtering terrain correction, comprising:

[0007] Acquire a remote sensing image to be enhanced, convert the remote sensing image from an RGB color space to an HSV color space, and extract a brightness component from the remote sensing image;

[0008] Performing homomorphic filtering on the luminance component based on a high-pass filter, and enhancing high-frequency components in the luminance component and reducing low-frequency components in the luminance component;

[0009] The brightness component after homomorphic filtering is combined with saturation and hue, and inversely transformed back to the RGB color space to perform image enhancement on the remote sensing image;

[0010] Obtaining an image grayscale distribution in the remote sensing image after image enhancement, and determining an information entropy of the remote sensing image after image enhancement based on the image grayscale distribution;

[0011] Obtaining the grayscale value of each pixel of the remote sensing image after image enhancement, and determining the contrast of the remote sensing image after image enhancement based on each pixel grayscale value;

[0012] Determine an enhancement score of the remote sensing image after image enhancement based on the information entropy and contrast, and compare the enhancement score with a preset enhancement score, wherein:

[0013] If the enhancement score is greater than or equal to the preset enhancement score, determining that the remote sensing image has completed image enhancement;

[0014] If the enhancement score is lower than the preset enhancement score, it is determined that the remote sensing image has not completed image enhancement, and the remote sensing image is re-processed with homomorphic filtering until the enhancement score is greater than or equal to the preset enhancement score.

[0015] Furthermore, when homomorphic filtering is performed on the luminance component based on a high-pass filter, the method includes:

[0016] A preset high-frequency gain and a preset low-frequency gain are pre-configured, and the high-pass filter is configured according to the preset high-frequency gain to perform homomorphic filtering on the luminance component, wherein the homomorphic filtering expression is as follows:

[0017]

[0018] Wherein, H(u,v) is the brightness component processed by homomorphic filtering, u is the high-frequency component, and v is the low-frequency component; R h is the preset high frequency gain, R l is the preset low-frequency gain, D(u, v) is the frequency of the current frequency point, D0 is the low-frequency cutoff frequency, and c is the preset amplitude of the gain change.

[0019] Furthermore, after the luminance component is subjected to homomorphic filtering, the following steps are included:

[0020] Dividing the brightness component after homomorphic filtering into regions according to a preset distance, and performing histogram equalization processing on the divided regional images;

[0021] Obtaining the total number of pixels in each of the regional images and the grayscale level of each of the pixels, and determining the total number of grayscale levels in the regional images based on the total number of pixels and the grayscale level of each of the pixels;

[0022] Determining a grayscale value threshold of the region according to the total number of pixels and the total number of grayscale levels in the region image;

[0023] Obtaining the total number of gray levels of each of the regional images after histogram equalization processing, and determining whether to adjust the preset distance based on the relationship between the total number of gray levels and the gray value threshold, wherein:

[0024] When the total number of gray levels is lower than or equal to the gray value threshold, determining not to adjust the preset distance;

[0025] When the total number of gray levels is higher than the gray value threshold, it is determined to adjust the preset distance.

[0026] Furthermore, when performing histogram equalization processing on the regional image, it includes:

[0027] Performing grayscale transformation on the regional image through a cumulative distribution histogram;

[0028] Obtaining a cumulative distribution histogram of each gray level and extracting a cumulative distribution probability of each gray level;

[0029] Substitute the cumulative distribution probability of each gray level into Formula I to obtain a preset gray range, and map the original gray value to the preset gray range to ensure that the distributed gray value covers the preset gray range, wherein Formula I is as follows:

[0030]

[0031] Among them, CDF(k) is the cumulative distribution value of gray value k, CDF min is the non-zero minimum cumulative distribution value, and S(k) is the preset grayscale distribution range.

[0032] Furthermore, when performing homomorphic filtering on the brightness component based on a high-pass filter, the method further includes:

[0033] The brightness component is subjected to multi-scale decomposition based on a wavelet transform formula to decompose the remote sensing image into a low-frequency part and a high-frequency part, wherein the wavelet transform formula is as follows:

[0034]

[0035] Among them, z a，b(t) is the representation of the low-frequency part and the high-frequency part after decomposition of the remote sensing image, a is the scaling factor, b is the translation factor, and t is the brightness component;

[0036] The low-frequency part is the overall illumination and structural information of the remote sensing image, and the high-frequency part is the texture and grayscale information of the remote sensing image.

[0037] Furthermore, obtaining the image grayscale distribution in the remote sensing image after image enhancement, and determining the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution, includes:

[0038] Obtain the occurrence probability of each grayscale value in the remote sensing image, and substitute the occurrence probability of each grayscale value into Formula II to obtain the information entropy of the remote sensing image, where Formula II is as follows:

[0039]

[0040] Where H is the information entropy of the remote sensing image, p i is the occurrence probability of gray value i, and n is the total number of gray levels.

[0041] Furthermore, obtaining the grayscale value of each pixel of the remote sensing image after image enhancement, and determining the contrast of the remote sensing image after image enhancement based on each pixel grayscale value, includes:

[0042] Obtaining the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image;

[0043] Substitute the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image into Formula III to obtain the contrast of the remote sensing image, where Formula III is as follows:

[0044]

[0045] Wherein, C is the contrast of the remote sensing image, f(i, j) is the grayscale value of the pixel, f is the average grayscale value of the remote sensing image, M is the width of the remote sensing image, and N is the length of the remote sensing image.

[0046] Furthermore, determining the enhancement score of the remote sensing image after image enhancement based on the information entropy and contrast includes:

[0047] Obtaining an information entropy difference between the information entropy and a preconfigured preset information entropy, and determining the enhancement score based on a relationship between the information entropy difference and a preconfigured first preset information entropy difference and a second preset information entropy difference:

[0048] When the information entropy difference is less than the first preset information entropy difference and the information entropy difference is greater than zero, determining the enhancement score to be W1;

[0049] When the information entropy difference is greater than or equal to the first preset information entropy difference, and the information entropy difference is less than the second preset information entropy difference, determining the enhancement score to be W2;

[0050] When the information entropy difference is greater than or equal to the second preset information entropy difference, determining the enhancement score to be W3;

[0051] Among them, the first preset information entropy difference is smaller than the second preset information entropy difference, and the first preset information entropy difference is greater than zero; W1<W2<W3.

[0052] Furthermore, when the enhancement score is determined to be Wi, i=1, 2, 3, including:

[0053] Obtaining a contrast difference between the contrast and a preset contrast, and determining whether to adjust the enhancement score Wi according to a relationship between the contrast difference and a first preset contrast difference and a second preset contrast difference configured in a preset manner;

[0054] When the contrast difference is less than the first preset contrast difference, determining not to adjust the enhancement score Wi;

[0055] When the contrast difference is greater than or equal to the first preset pair difference and the contrast difference is less than the second preset pair difference, the adjustment coefficient is determined to be g1, and the enhancement score Wi is adjusted according to the adjustment coefficient g1;

[0056] When the contrast difference is greater than or equal to the second preset contrast difference, the adjustment coefficient is determined to be g2, and the enhancement score Wi is adjusted according to the adjustment coefficient g2;

[0057] The first preset pair-wise difference is smaller than the second preset pair-wise difference, and g1<g2<1.

[0058] Compared to existing technologies, the present invention offers the following advantages: Through a terrain correction method based on homomorphic filtering, the brightness and contrast of remote sensing images can be significantly improved. This is particularly applicable to remote sensing images that suffer from loss of detail due to uneven illumination or low contrast. In remote sensing image processing, uneven illumination and low contrast often affect image quality, reducing the visibility of image details and, in turn, the accuracy of image analysis and interpretation. By converting remote sensing images from RGB color space to HSV color space, extracting the luminance component, and applying a high-pass filter to it, the high-frequency components in the image can be enhanced, thereby improving the presentation of image details. This method effectively avoids the over- or under-enhancement of contrast found in traditional image enhancement methods and has significant practical significance, particularly in remote sensing image applications. Furthermore, by automatically calculating the image's grayscale distribution, information entropy, and contrast, the image enhancement effect is ensured to meet expected standards. Information entropy reflects the complexity of the image, while contrast directly affects the image's visual quality. By quantitatively evaluating these two metrics, it is possible to objectively determine whether the image enhancement effect has achieved the desired level. Automatic comparative evaluation during the enhancement process avoids the inconsistencies that may occur when traditional parameters are manually adjusted, thereby improving the automation and precision of image enhancement. This automatic adjustment mechanism based on standardized evaluation makes remote sensing image enhancement more intelligent, efficient, and adaptable, and can meet the needs of different application scenarios. Finally, by setting a preset enhancement score and comparing it with the actual enhancement score, the quality control of image enhancement is ensured. If the enhancement effect does not meet expectations, the system will automatically start the iterative process of homomorphic filtering until the image enhancement effect meets the standard. This adaptive adjustment mechanism avoids the tedious manual operation and can accurately optimize for different image characteristics, improving the robustness and stability of image enhancement. This innovative measure not only improves the automation level of image processing, but also enables the enhanced image to provide higher accuracy in subsequent analysis, meeting the high-precision requirements of remote sensing data analysis, change detection, and other fields.

[0059] On the other hand, the present application also provides a remote sensing image enhancement system based on homomorphic filtering terrain correction, comprising:

[0060] An acquisition module, used to acquire remote sensing images to be enhanced;

[0061] An extraction module, electrically connected to the acquisition module, configured to convert the remote sensing image from an RGB color space to an HSV color space and extract a brightness component from the remote sensing image;

[0062] a central control module electrically connected to the extraction module, the central control module being configured with a high-pass filter, and performing homomorphic filtering on the luminance component based on the high-pass filter, thereby enhancing the high-frequency components in the luminance component and reducing the low-frequency components in the luminance component; the central control module is further configured to combine the luminance component after homomorphic filtering with saturation and hue, and invert the luminance component back to the RGB color space to perform image enhancement on the remote sensing image;

[0063] An evaluation module is electrically connected to the central control module, and the evaluation module is used to obtain the image grayscale distribution in the remote sensing image after image enhancement, and determine the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution; the evaluation module is also used to obtain the grayscale value of each pixel of the remote sensing image after image enhancement, and determine the contrast of the remote sensing image after image enhancement based on each pixel grayscale value; the evaluation module is also used to determine the enhancement score of the remote sensing image after image enhancement based on the information entropy and contrast, and determine whether the remote sensing image has completed image enhancement based on the relationship between the enhancement score and the preset enhancement score.

[0064] It can be understood that the remote sensing image enhancement method and system based on homomorphic filtering terrain correction in the above embodiments of the present invention have the same beneficial effects and will not be described in detail. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0066] Figure 1 A flowchart of a remote sensing image enhancement method based on homomorphic filtering terrain correction provided by an embodiment of the present invention;

[0067] Figure 2 This is an effect diagram of converting a remote sensing image from RGB color space to HSV space in an embodiment of the present invention;

[0068] Figure 3 This is a diagram showing the terrain enhancement effect of a remote sensing image based on homomorphic filtering and wavelet transform in an embodiment of the present invention;

[0069] Figure 4 This is a functional block diagram of a remote sensing image enhancement system based on homomorphic filtering terrain correction provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless there is a conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0071] like Figure 1-Figure 3 In some embodiments of the present application, a remote sensing image enhancement method based on homomorphic filtering terrain correction is provided, comprising:

[0072] Step S100: Acquire a remote sensing image to be enhanced, convert the remote sensing image from an RGB color space to an HSV color space, and extract a brightness component from the remote sensing image.

[0073] It can be understood that by converting remote sensing images from RGB color space to HSV color space, the brightness component (V component) in HSV space is used to represent the image's luminance information. While the RGB color space is primarily used to represent color, the HSV color space is more suitable for processing image features related to brightness, saturation, and hue. The converted brightness component (V component) can independently reflect the image's luminance characteristics, facilitating subsequent image enhancement processing. By extracting the brightness component, the brightness and color information of an image can be effectively separated.

[0074] Specifically, when performing homomorphic filtering on the brightness component based on a high-pass filter, it also includes: performing multi-scale decomposition on the brightness component based on a wavelet transform formula to decompose the remote sensing image into a low-frequency part and a high-frequency part, wherein the wavelet transform formula is as follows: Among them, z a，b (t) represents the low-frequency and high-frequency components of the decomposed remote sensing image, a is the scaling factor, b is the translation factor, and t is the brightness component. The low-frequency component represents the overall illumination and structural information of the remote sensing image, while the high-frequency component represents the texture and grayscale information of the remote sensing image.

[0075] It can be understood that, based on the wavelet transform formula, by adjusting the scaling factor (a) and the translation factor (b), images can be analyzed at different scales and locations. The wavelet transform can capture detailed variations in images at different resolutions, is highly adaptable, and effectively separates different frequency components within an image. Specifically, the low-frequency component represents the overall illumination and structural information of the remote sensing image, including macroscopic features such as large-scale illumination variations, the overall structure, and shape of the image. The high-frequency component contains detailed information within the image, such as texture, edges, and grayscale variations, reflecting the image's microscopic features. By decomposing the brightness component of a remote sensing image into low-frequency and high-frequency components, the system can perform different enhancement processes on different frequency components. Processing of the low-frequency component generally aims to enhance the image's illumination and overall structure, improving image uniformity and clarity; whereas processing of the high-frequency component focuses on enhancing texture and details, improving sharpness and contrast. This separation allows for more precise adjustment of various aspects of the image, avoiding information loss in overall image processing and improving the visual quality of the image. Furthermore, the multi-scale nature of the wavelet transform enables image analysis at different scales, offering significant advantages for processing complex remote sensing images. Remote sensing images often contain multi-layered and multi-scale features, such as macroscopic terrain structures and minute details. The wavelet transform can extract relevant information from images at different scales, allowing the processed image to better reflect both the overall morphology and local details of the terrain. This multi-scale processing approach is particularly suitable for analyzing remote sensing images, as these images are often affected by factors such as illumination variations and perspective changes. The wavelet transform can effectively reduce these interferences, ensuring accurate extraction of important features within the image. Another advantage of the wavelet transform is its localization in time and frequency. Unlike the traditional Fourier transform, which is global and cannot provide information on local features, the wavelet transform can effectively capture frequency variations within local regions, making it particularly suitable for processing images containing local texture and detail variations. In remote sensing image processing, local features such as buildings, roads, and vegetation often determine the practical application value of an image. The wavelet transform can precisely process these local regions, enhancing their detail.

[0076] Step S200: Perform homomorphic filtering on the luminance component based on a high-pass filter, enhance the high-frequency components in the luminance component, and reduce the low-frequency components in the luminance component.

[0077] Specifically, when performing homomorphic filtering on the luminance component based on the high-pass filter, the process includes: pre-configuring a preset high-frequency gain and a preset low-frequency gain, and configuring the high-pass filter according to the preset high-frequency gain to perform homomorphic filtering on the luminance component. The homomorphic filtering expression is as follows:

[0078] Among them, H(u,v) is the brightness component processed by homomorphic filtering, u is the high-frequency component, and v is the low-frequency component. h To preset high frequency gain, R l is the preset low-frequency gain, D(u,v) is the frequency of the current frequency point, D0 is the low-frequency cutoff frequency, and c is the preset amplitude of the gain change.

[0079] It is understandable that by presetting the high frequency gain (R h ) and low frequency gain (R l ), and configure a high-pass filter to perform homomorphic filtering on the brightness component, which can enhance the high-frequency components (details) in the image while suppressing the low-frequency components (flat areas or background information). The core idea of homomorphic filtering is to decompose the image into low-frequency and high-frequency components through frequency domain transformation, and then adjust them separately: the high-frequency part is enhanced (R h Gain), the low frequency part is attenuated (R l By adjusting the low-frequency cutoff frequency (D0) and the gain variation (c), the filtering range and intensity can be precisely controlled, resulting in optimized image processing. This method is particularly suitable for enhancing the clarity of details in remote sensing images while reducing background noise.

[0080] Specifically, after the luminance component is homomorphically filtered, the method includes: dividing the luminance component after homomorphic filtering into regions according to a preset distance, and performing histogram equalization on the divided regional images. Obtaining the total number of pixels in each regional image and the grayscale level of each pixel, and determining the total number of grayscale levels in the regional image based on the total number of pixels and the grayscale level of each pixel. Determining the grayscale value threshold of the region based on the total number of pixels and the total number of grayscale levels in the regional image. Obtaining the total number of grayscale levels in each regional image after histogram equalization, and determining whether to adjust the preset distance based on the relationship between the total number of grayscale levels and the grayscale value threshold, wherein: when the total number of grayscale levels is lower than or equal to the grayscale value threshold, determining not to adjust the preset distance. When the total number of grayscale levels is higher than the grayscale value threshold, determining to adjust the preset distance.

[0081] Specifically, when performing histogram equalization on a regional image, the process includes: performing grayscale transformation on the regional image using a cumulative distribution histogram. The cumulative distribution histogram of each grayscale level is obtained, and the cumulative distribution probability of each grayscale level is extracted. The cumulative distribution probability of each grayscale level is substituted into Formula I to obtain a preset grayscale range, and the original grayscale value is mapped to the preset grayscale range to ensure that the distributed grayscale value covers the preset grayscale range. Formula I is as follows: Among them, CDF(k) is the cumulative distribution value of gray value k, CDFmin is the non-zero minimum cumulative distribution value, and S(k) is the preset grayscale distribution range.

[0082] As you can understand, after homomorphic filtering, the luminance component is divided into multiple small regions through region partitioning, allowing for separate image enhancement processing for each region. This method not only equalizes the brightness of the entire image but also adjusts for brightness differences within local regions, thereby enhancing image details. This is especially true in complex terrain and lighting conditions, where local features may be obscured. Region partitioning allows for more precise control of the image processing effect for each region, effectively improving brightness differences between different regions. Histogram equalization calculates the cumulative distribution histogram (CDF) of each region's image and uses this histogram to transform the image's grayscale values. The CDF reflects the distribution of pixels at each grayscale level in an image and is commonly used to assess image contrast. If the grayscale distribution of an image is concentrated within a small range, the image contrast is low and details are difficult to discern. Histogram equalization evenly distributes these grayscale values across the entire grayscale range, enhancing image contrast and detail levels, making the image's brightness and details more clearly visible. In order to ensure that the image enhancement effect meets the expectations, the grayscale transformation of each area should be mapped not only based on the cumulative distribution histogram, but also based on the preset grayscale range to avoid excessive or insufficient brightness adjustment. In addition, this method further optimizes the adaptive ability of image processing by dynamically adjusting the preset distance of region division. When the grayscale distribution of a certain area is lower than the preset threshold, the system automatically identifies and decides whether to adjust the region division distance to ensure that the image processing effect in different areas is optimal. This adaptive adjustment mechanism makes the image processing process more intelligent, and can flexibly adjust the processing strategy according to the actual characteristics of the image, avoiding the shortcomings of manual intervention and static parameter setting. By adjusting the preset distance, the system can perform more detailed processing on areas with uneven grayscale distribution, ensuring that the brightness distribution of the entire image is more uniform, thereby improving the quality and readability of the image. Finally, the grayscale transformation formula of the histogram equalization is obtained by utilizing the minimum value of the cumulative distribution histogram (CDF min ) and the preset grayscale distribution range (S(k)), the original grayscale value is mapped to ensure that the distributed grayscale value covers the preset grayscale range. Through this mapping, the grayscale value of the image will be more evenly distributed in the entire grayscale space, thereby greatly improving the overall contrast of the image. The CDF(k) in the formula represents the cumulative distribution value of the current grayscale level, and by comparing it with the CDF minBy comparing the grayscale values, the system can identify the minimum grayscale value and map it to the appropriate grayscale range, preventing the image from appearing too bright or too dark. This process not only enhances the details of the image, but also reduces the interference of noise to a certain extent, improving the image processing effect.

[0083] Step S300: Combine the brightness component after homomorphic filtering with the saturation and hue, and inversely transform it back to the RGB color space to perform image enhancement on the remote sensing image.

[0084] It can be understood that in the homomorphic filtering process, the brightness component is processed by a high-pass filter, which removes the low-frequency part and enhances the high-frequency part, thereby improving the details and contrast of the image. Then, the processed brightness component is combined with the saturation and hue (i.e., color information) of the image, and converted from the HSV color space back to the RGB color space through an inverse transformation. This process enhances the visual effect of the image by reconstructing the relationship between brightness and color, especially in terms of lighting and detail presentation, thereby improving the overall quality of the remote sensing image. Through this technology, remote sensing images with low contrast or uneven lighting can be effectively improved, their visual performance can be enhanced, and the images can be made clearer and more recognizable.

[0085] Step S400: Obtain the image grayscale distribution in the remote sensing image after image enhancement, and determine the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution.

[0086] Specifically, obtaining the image grayscale distribution in the remote sensing image after image enhancement and determining the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution includes: obtaining the occurrence probability of each grayscale value in the remote sensing image, and substituting the occurrence probability of each grayscale value into Formula II to obtain the information entropy of the remote sensing image, wherein Formula II is as follows: Among them, H is the information entropy of the remote sensing image, p i is the occurrence probability of gray value i, and n is the total number of gray levels.

[0087] It can be understood that the grayscale distribution of the remote sensing image after image enhancement is obtained by counting the frequency of occurrence of pixels at each grayscale level in the image. The probability of occurrence of each grayscale value (p i) reflects the proportion of that grayscale in the image. Then, by substituting these probabilities into Information Entropy Formula II, the overall information entropy of the image is calculated. Information entropy, a key concept in information theory, is a tool for measuring image complexity and information content. Higher values indicate greater information content and richer details, while lower values indicate a more monotonous or lacking image detail. The calculation of information entropy is essentially an in-depth analysis of the image's grayscale distribution. By considering the distribution of each grayscale level within the image, it can reflect the image's diversity and complexity. Specifically, images with higher information entropy typically exhibit richer texture, detail, and higher discernibility. For example, in remote sensing images, details such as land types, buildings, and vegetation often require higher contrast and detail. Images with higher information entropy can better render these details, resulting in stronger image expression. Evaluating information entropy provides an objective criterion for image enhancement, helping to determine whether the enhancement results meet expectations. This process is not only applicable to remote sensing image quality assessment but also has broad application in other image processing fields, such as medical imaging and satellite imagery. In remote sensing image processing, we often face problems such as uneven illumination and low contrast. Information entropy, as an objective quantitative indicator, can effectively help determine whether the image has been improved and ensure the quality of the enhancement effect.

[0088] Step S500: obtaining the grayscale value of each pixel of the remote sensing image after image enhancement, and determining the contrast of the remote sensing image after image enhancement based on the grayscale value of each pixel.

[0089] Specifically, obtaining the grayscale value of each pixel of the remote sensing image after image enhancement and determining the contrast of the remote sensing image after image enhancement based on the grayscale value of each pixel includes: obtaining the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image. Substituting the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image into Formula III to obtain the contrast of the remote sensing image, where Formula III is as follows: Among them, C is the contrast of the remote sensing image, f(i,j) is the gray value of the pixel, is the average gray value of the remote sensing image, M is the width of the remote sensing image, and N is the length of the remote sensing image.

[0090] As you can understand, image contrast is calculated by calculating the difference between the grayscale value of each pixel and the image's average grayscale value. The grayscale value of each pixel in an image reflects the brightness of that pixel, while the image's average grayscale value represents the overall brightness level of the image. By combining these grayscale values with the image's size information, the image's contrast can be quantified. Each term in Formula III represents an important factor that needs to be considered during contrast calculation. For example, the grayscale value f(i, j) of a pixel reflects the brightness at each location, while the image's average grayscale value serves as a benchmark, measuring the degree to which each pixel deviates from the overall brightness. M and N represent the image's width and length, respectively, providing image size information, allowing the contrast calculation to take into account the overall scale of the image. The contrast value calculated in this way not only provides a quantitative basis for image quality assessment but also helps optimize the image enhancement process. Higher contrast means richer image details and clearer visuals. This is particularly important in remote sensing image processing, as remote sensing images often feature complex background objects. Low contrast images can blur or confuse different features. Therefore, enhancing image contrast can effectively improve the visual quality of the image, making the features of the ground objects more prominent and facilitating subsequent analysis and application. Furthermore, the results of contrast calculations can serve as an evaluation criterion for whether image enhancement has achieved the desired effect. A low contrast value may indicate that the image enhancement process is insufficient and may require further processing or adjustment. A high contrast value, on the other hand, indicates that the image details have been effectively enhanced, improving image quality. Using contrast as a criterion for image quality assessment not only enables more accurate image processing but also reduces manual intervention, improving the automation and efficiency of remote sensing image enhancement processing.

[0091] Step S600: determining an enhancement score of the remote sensing image after image enhancement based on information entropy and contrast, and comparing the enhancement score with a preset enhancement score.

[0092] Specifically, when determining the enhancement score of a remote sensing image after image enhancement based on information entropy and contrast, the method includes: obtaining an information entropy difference between the information entropy and a pre-configured preset information entropy, and determining the enhancement score based on the relationship between the information entropy difference and the pre-configured first and second preset information entropy differences: when the information entropy difference is less than the first preset information entropy difference and the information entropy difference is greater than zero, the enhancement score is determined to be W1. When the information entropy difference is greater than or equal to the first preset information entropy difference and the information entropy difference is less than the second preset information entropy difference, the enhancement score is determined to be W2. When the information entropy difference is greater than or equal to the second preset information entropy difference, the enhancement score is determined to be W3. The first preset information entropy difference is less than the second preset information entropy difference and the first preset information entropy difference is greater than zero. W1<W2<W3.

[0093] Specifically, when the enhancement score Wi is determined, i = 1, 2, or 3, the method includes: obtaining a contrast difference between the contrast and a preset contrast ratio, and determining whether to adjust the enhancement score Wi based on the relationship between the contrast difference and a first preset contrast difference and a second preset contrast difference. When the contrast difference is less than the first preset contrast difference, it is determined that no adjustment is made to the enhancement score Wi. When the contrast difference is greater than or equal to the first preset contrast difference and less than the second preset contrast difference, an adjustment coefficient g1 is determined, and the enhancement score Wi is adjusted based on the adjustment coefficient g1. When the contrast difference is greater than or equal to the second preset contrast difference, an adjustment coefficient g2 is determined, and the enhancement score Wi is adjusted based on the adjustment coefficient g2. The first preset contrast difference is less than the second preset contrast difference, and g1 < g2 < 1.

[0094] Specifically, when comparing the enhancement score with the preset enhancement score, if the enhancement score is greater than or equal to the preset enhancement score, it is determined that the remote sensing image has completed image enhancement. If the enhancement score is lower than the preset enhancement score, it is determined that the remote sensing image has not completed image enhancement, and the remote sensing image is re-processed with homomorphic filtering until the enhancement score is greater than or equal to the preset enhancement score.

[0095] As you can understand, by calculating the image enhancement score, image quality is quantified into three levels, W1, W2, and W3, based on changes in information entropy, representing different levels of image quality. This scoring mechanism compares the difference in information entropy to ensure that the image enhancement process effectively enhances contrast and detail. Specifically, when the information entropy difference is less than a preset threshold, the image is considered to be in a low-enhancement state. If the information entropy difference is larger, the image may be in a high-enhancement state, and is then classified using the scores W1, W2, and W3. Furthermore, contrast plays a crucial role in image quality assessment. By calculating the image contrast difference and comparing it with a preset contrast difference, the image enhancement effect is further refined. When the contrast difference falls within a certain threshold, the enhancement score is fine-tuned using the adjustment factor g1 or g2 to ensure a more vivid image and fully display details. This adjustment mechanism avoids issues such as over-enhancement or ineffective enhancement due to excessively high or low contrast. Based on this, the image enhancement score is compared with the preset target enhancement score to ultimately determine the degree of image enhancement completion. If the enhancement score does not meet the preset standard, the image enhancement process will continue until the quality requirements are met. This iterative process ensures that image enhancement can be continuously optimized, avoiding over-enhancement or under-enhancement in a single process, and improving the accuracy of image processing. Through repeated adjustment and optimization, high-quality remote sensing image enhancement effects are ultimately achieved. Through this comprehensive scoring mechanism, the quantitative relationship between information entropy and contrast is utilized, combined with multiple iterative adjustments, the contrast and detail performance of the image are significantly improved. This method not only effectively avoids the lack of manual adjustment in traditional methods, but also ensures the quality stability of image enhancement during automated processing, adapting to the needs of remote sensing image enhancement in different scenarios.

[0096] In the above-described embodiment, the terrain correction method based on homomorphic filtering can significantly improve the brightness and contrast of remote sensing images. This is particularly applicable to remote sensing images that suffer from loss of detail due to uneven illumination or low contrast. In remote sensing image processing, uneven illumination and low contrast often affect image quality, reducing the visibility of image details, and thus affecting the accuracy of image analysis and interpretation. By converting remote sensing images from RGB color space to HSV color space, extracting the luminance component, and applying high-pass filtering to it, the high-frequency components in the image can be enhanced, thereby improving the presentation of image details. This method effectively avoids the over-enhancement or under-enhancement of contrast found in traditional image enhancement methods, and has important practical significance, especially in remote sensing image applications. Furthermore, by automatically calculating the image's grayscale distribution, information entropy, and contrast, the image enhancement effect is ensured to meet the expected standards. Information entropy reflects the complexity of the image, while contrast directly affects the image's visual quality. By quantitatively evaluating these two metrics, it is possible to objectively determine whether the image enhancement effect has achieved the desired level. Automated contrast evaluation during the enhancement process avoids the inconsistencies that can occur with traditional manual parameter adjustment, improving the automation and accuracy of image enhancement. This automatic adjustment mechanism based on standardized evaluation makes remote sensing image enhancement more intelligent, efficient, and adaptable, and can meet the needs of different application scenarios. Finally, by setting a preset enhancement score and comparing it with the actual enhancement score, the quality control of the image enhancement is ensured. If the enhancement effect does not meet expectations, the system will automatically start the iterative process of homomorphic filtering until the image enhancement effect meets the standard. This adaptive adjustment mechanism avoids the tedious manual operation and can accurately optimize for different image characteristics, improving the robustness and stability of image enhancement. This innovative measure not only improves the level of automation in image processing, but also enables the enhanced image to provide higher accuracy in subsequent analysis, meeting the high-precision requirements of remote sensing data analysis, change detection, and other fields.

[0097] In another preferred embodiment based on the above embodiment, Figure 4 As shown, this embodiment provides a remote sensing image enhancement system based on homomorphic filtering terrain correction, including: an acquisition module, an extraction module, a central control module and an evaluation module.

[0098] Specifically, the acquisition module is used to acquire the remote sensing image to be enhanced. The extraction module is electrically connected to the acquisition module and is used to convert the remote sensing image from RGB color space to HSV color space and extract the luminance component within the remote sensing image. The central control module is electrically connected to the extraction module and is equipped with a high-pass filter. The central control module performs homomorphic filtering on the luminance component based on the high-pass filter, enhancing the high-frequency components within the luminance component and reducing the low-frequency components within the luminance component. The central control module is also used to combine the saturation and hue of the homomorphic filtered luminance component and invert it back to RGB color space to enhance the remote sensing image. The evaluation module is electrically connected to the central control module and is used to obtain the grayscale distribution of the enhanced remote sensing image and determine the information entropy of the enhanced remote sensing image based on the grayscale distribution. The evaluation module is also used to obtain the grayscale value of each pixel in the enhanced remote sensing image and determine the contrast of the enhanced remote sensing image based on each pixel grayscale value. The evaluation module is also used to determine the enhancement score of the remote sensing image after image enhancement based on information entropy and contrast, and to determine whether the remote sensing image has completed image enhancement based on the relationship between the enhancement score and the preset enhancement score.

[0099] It can be understood that the remote sensing image enhancement method and system based on homomorphic filtering terrain correction in the above embodiments of the present invention have the same beneficial effects and will not be described in detail.

[0100] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or a combination of software and hardware embodiments. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0101] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0102] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A remote sensing image enhancement method based on homomorphic filtering terrain correction, characterized in that: include: Acquire a remote sensing image to be enhanced, convert the remote sensing image from an RGB color space to an HSV color space, and extract a brightness component from the remote sensing image; Performing homomorphic filtering on the luminance component based on a high-pass filter, and enhancing high-frequency components in the luminance component and reducing low-frequency components in the luminance component; The brightness component after homomorphic filtering is combined with saturation and hue, and inversely transformed back to the RGB color space to perform image enhancement on the remote sensing image; Obtaining an image grayscale distribution in the remote sensing image after image enhancement, and determining an information entropy of the remote sensing image after image enhancement based on the image grayscale distribution; Obtaining the grayscale value of each pixel of the remote sensing image after image enhancement, and determining the contrast of the remote sensing image after image enhancement based on each pixel grayscale value; Determine an enhancement score of the remote sensing image after image enhancement based on the information entropy and contrast, and compare the enhancement score with a preset enhancement score, wherein: If the enhancement score is greater than or equal to the preset enhancement score, determining that the remote sensing image has completed image enhancement; If the enhancement score is lower than the preset enhancement score, it is determined that the remote sensing image has not completed image enhancement, and the remote sensing image is re-processed with homomorphic filtering until the enhancement score is greater than or equal to the preset enhancement score.

2. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 1, characterized in that: When homomorphic filtering is performed on the brightness component based on a high-pass filter, the method includes: A preset high-frequency gain and a preset low-frequency gain are pre-configured, and the high-pass filter is configured according to the preset high-frequency gain to perform homomorphic filtering on the luminance component, wherein the homomorphic filtering expression is as follows: Wherein, H(u,v) is the brightness component processed by homomorphic filtering, u is the high-frequency component, and v is the low-frequency component; R h is the preset high frequency gain, R l is the preset low-frequency gain, D(u, v) is the frequency of the current frequency point, D0 is the low-frequency cutoff frequency, and c is the preset amplitude of the gain change.

3. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 2, characterized in that: After the luminance component is subjected to homomorphic filtering, the following steps are performed: Dividing the brightness component after homomorphic filtering into regions according to a preset distance, and performing histogram equalization processing on the divided regional images; Obtaining the total number of pixels in each of the regional images and the grayscale level of each of the pixels, and determining the total number of grayscale levels in the regional images based on the total number of pixels and the grayscale level of each of the pixels; Determining a grayscale value threshold of the region according to the total number of pixels and the total number of grayscale levels in the region image; Obtaining the total number of gray levels of each of the regional images after histogram equalization processing, and determining whether to adjust the preset distance based on the relationship between the total number of gray levels and the gray value threshold, wherein: When the total number of gray levels is lower than or equal to the gray value threshold, determining not to adjust the preset distance; When the total number of gray levels is higher than the gray value threshold, it is determined to adjust the preset distance.

4. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 3, characterized in that: When performing histogram equalization on a regional image, it includes: Performing grayscale transformation on the regional image through a cumulative distribution histogram; Obtaining a cumulative distribution histogram of each gray level and extracting a cumulative distribution probability of each gray level; Substitute the cumulative distribution probability of each gray level into Formula I to obtain a preset gray range, and map the original gray value to the preset gray range to ensure that the distributed gray value covers the preset gray range, wherein Formula I is as follows: Among them, CDF(k) is the cumulative distribution value of gray value k, CDF min is the non-zero minimum cumulative distribution value, and S(k) is the preset grayscale distribution range.

5. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 4, characterized in that: When performing homomorphic filtering on the brightness component based on a high-pass filter, the method further includes: The brightness component is subjected to multi-scale decomposition based on a wavelet transform formula to decompose the remote sensing image into a low-frequency part and a high-frequency part, wherein the wavelet transform formula is as follows: Among them, Z a，b (t) is the representation of the low-frequency part and the high-frequency part after decomposition of the remote sensing image, a is the scaling factor, b is the translation factor, and t is the brightness component; The low-frequency part is the overall illumination and structural information of the remote sensing image, and the high-frequency part is the texture and grayscale information of the remote sensing image.

6. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 1, characterized in that: Obtaining the image grayscale distribution in the remote sensing image after image enhancement, and determining the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution, including: Obtain the occurrence probability of each grayscale value in the remote sensing image, and substitute the occurrence probability of each grayscale value into Formula II to obtain the information entropy of the remote sensing image, where Formula II is as follows: Where H is the information entropy of the remote sensing image, p i is the occurrence probability of gray value i, and n is the total number of gray levels.

7. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 1, characterized in that: Obtaining the grayscale value of each pixel of the remote sensing image after image enhancement, and determining the contrast of the remote sensing image after image enhancement based on each pixel grayscale value, includes: Obtaining the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image; Substitute the width and length of the remote sensing image, the grayscale value of each pixel in the remote sensing image, and the average grayscale value of the remote sensing image into Formula III to obtain the contrast of the remote sensing image, where Formula III is as follows: Wherein, C is the contrast of the remote sensing image, f(i,j) is the grayscale value of the pixel, is the average grayscale value of the remote sensing image, M is the width of the remote sensing image, and N is the length of the remote sensing image.

8. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 1, characterized in that: Determining the enhancement score of the remote sensing image after image enhancement based on the information entropy and the contrast includes: Obtaining an information entropy difference between the information entropy and a preconfigured preset information entropy, and determining the enhancement score based on a relationship between the information entropy difference and a preconfigured first preset information entropy difference and a second preset information entropy difference: When the information entropy difference is less than the first preset information entropy difference and the information entropy difference is greater than zero, determining the enhancement score to be W1; When the information entropy difference is greater than or equal to the first preset information entropy difference, and the information entropy difference is less than the second preset information entropy difference, determining the enhancement score to be W2; When the information entropy difference is greater than or equal to the second preset information entropy difference, determining the enhancement score to be W3; Among them, the first preset information entropy difference is smaller than the second preset information entropy difference, and the first preset information entropy difference is greater than zero; W1<W2<W3.

9. The remote sensing image enhancement method based on homomorphic filtering terrain correction according to claim 8, characterized in that: When the enhancement score is determined to be Wi, i=1, 2, 3, including: Obtaining a contrast difference between the contrast and a preset contrast, and determining whether to adjust the enhancement score Wi according to a relationship between the contrast difference and a first preset contrast difference and a second preset contrast difference configured in a preset manner; When the contrast difference is less than the first preset contrast difference, determining not to adjust the enhancement score Wi; When the contrast difference is greater than or equal to the first preset contrast difference and less than the second preset contrast difference, the adjustment coefficient is determined to be g1, and the enhancement score Wi is adjusted according to the adjustment coefficient g1; When the contrast difference is greater than or equal to the second preset contrast difference, the adjustment coefficient is determined to be g2, and the enhancement score Wi is adjusted according to the adjustment coefficient g2; The first preset contrast difference is smaller than the second preset contrast difference, and g1<g2<1.

10. A remote sensing image enhancement system based on homomorphic filtering terrain correction, using a remote sensing image enhancement method based on homomorphic filtering terrain correction according to any one of claims 1 to 9, characterized in that: include: An acquisition module, used to acquire remote sensing images to be enhanced; An extraction module, electrically connected to the acquisition module, configured to convert the remote sensing image from an RGB color space to an HSV color space and extract a brightness component from the remote sensing image; a central control module electrically connected to the extraction module, the central control module being configured with a high-pass filter, and performing homomorphic filtering on the luminance component based on the high-pass filter, thereby enhancing the high-frequency components in the luminance component and reducing the low-frequency components in the luminance component; the central control module is further configured to combine the luminance component after homomorphic filtering with saturation and hue, and invert the luminance component back to the RGB color space to perform image enhancement on the remote sensing image; An evaluation module is electrically connected to the central control module, and the evaluation module is used to obtain the image grayscale distribution in the remote sensing image after image enhancement, and determine the information entropy of the remote sensing image after image enhancement based on the image grayscale distribution; the evaluation module is also used to obtain the grayscale value of each pixel of the remote sensing image after image enhancement, and determine the contrast of the remote sensing image after image enhancement based on each pixel grayscale value; the evaluation module is also used to determine the enhancement score of the remote sensing image after image enhancement based on the information entropy and contrast, and determine whether the remote sensing image has completed image enhancement based on the relationship between the enhancement score and the preset enhancement score.

Images