Geographic scene recognition method based on image processing

Abstract

The invention relates to the technical field of image processing, in particular to a geographic scene recognition method based on image processing, which comprises the following steps: acquiring a remote sensing image of an aquaculture area, and dividing the remote sensing image into a plurality of areas; and in each region, determining the reflection possibility of the pixel point according to the brightness and saturation of the pixel point, calculating a reflection comprehensive judgment coefficient according to the reflection possibility of the pixel point, the reflection consistency of the local range of the pixel point and the contrast ratio of the region where the pixel point is located, and determining the reflection pixel point based on the reflection comprehensive judgment coefficient. The light reflection possibility of the area is determined according to the characteristics of the light reflection pixel points of each area, the light reflection area is screened based on the light reflection possibility of the area, and adaptive scale parameters are dynamically set according to the density of the light reflection area, so that the problem of uneven illumination in the remote sensing image can be effectively solved; and the accuracy and reliability of subsequent geographical scene recognition are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a geographical scene recognition method based on image processing. Background Art

[0002] In the work of geographic scene recognition in aquaculture areas, optical remote sensing has become a key means of obtaining images due to its unique advantages, providing important data support for large-scale aquaculture planning, resource assessment, etc. At present, the common geographic scene recognition process is to divide the regional contours with the help of edge detection algorithms after acquiring remote sensing images, and then perform feature matching for each area to achieve geographic scene recognition.

[0003] There are large areas of water in aquaculture areas. The reflection produced by the water surface will cause uneven lighting. Uneven lighting will cause the color, texture and other features of different areas of the remote sensing image to be distorted, resulting in deviations in geographic scene recognition based on these features, and it is impossible to accurately judge the type, layout and other information of the aquaculture area.

[0004] In the prior art, the multi-scale Retinex algorithm uses Gaussian functions with different standard deviations to perform convolution operations on images to obtain multi-scale images, separate the illumination component and the reflection component in each scale image, and then adjust the illumination component through the gain coefficient, thereby achieving the effect of removing uneven illumination and enhancing image contrast.

[0005] However, in practical applications, there are obvious limitations when using the traditional multi-scale Retinex algorithm to remove uneven illumination in remote sensing images of aquaculture areas. When performing convolution processing on an image, the entire image is operated each time according to a Gaussian function with a preset fixed standard deviation to obtain the corresponding scale image, which cannot be flexibly adjusted according to the illumination characteristics of different areas of the entire image. Due to the complexity of aquaculture scenes, the illumination conditions in different areas of the image vary greatly. This method of preset fixed standard deviations makes it difficult to accurately handle the problem of uneven illumination, which in turn affects the accuracy and reliability of subsequent geographic scene recognition. Summary of the invention

[0006] In order to solve the problem that the traditional multi-scale Retinex algorithm is difficult to accurately remove the uneven illumination in remote sensing images when performing geographic scene recognition in aquaculture areas, thereby affecting the accuracy of geographic scene recognition, the present invention proposes a geographic scene recognition method based on image processing, the method comprising: Collect remote sensing images of aquaculture areas, divide them into multiple areas according to the edge detection results of the remote sensing images, and convert the remote sensing images from RGB color space to HSV color space; In each area, the reflection possibility of each pixel is determined according to the brightness and saturation of each pixel, and the reflection consistency of the local range of each pixel is evaluated; the reflection comprehensive determination coefficient of each pixel is determined in combination with the reflection possibility of each pixel, the reflection consistency of the local range of each pixel, and the contrast of the area where each pixel is located; the reflection comprehensive determination coefficient of each pixel is compared with a preset reflection comprehensive determination coefficient threshold to determine the reflective pixel; Based on the mean and discreteness of the reflection comprehensive determination coefficients of all pixels in each area, combined with the number of reflective pixels in each area and the brightness difference between each area and the surrounding areas, the reflection possibility of each area is determined; the reflection possibility of each area is compared with a preset reflection possibility threshold to determine the reflective area; The density of each reflective area is determined, and based on the density, the adaptive scale parameters of the multi-scale Retinex algorithm when processing each reflective area are dynamically set to remove the uneven illumination of the remote sensing image, and geographic scene recognition is performed based on the remote sensing image after the uneven illumination is removed.

[0007] This technical solution first performs preliminary preprocessing on the original remote sensing image, divides the area so that different areas can be processed differently later, and the color space conversion provides a more suitable color representation for subsequent reflection possibility judgment and other operations, which is conducive to more accurate analysis of the light and color related information in the image. In addition, based on the principle that reflection usually causes changes in the brightness and saturation of pixels, the possibility of reflection is determined. The reflection consistency of the local range can be evaluated to determine whether the reflection of the pixel is consistent with the surrounding pixels, avoiding misjudgment due to individual noise points. Combining these two factors and regional contrast to determine the comprehensive reflection determination coefficient, the judgment of the reflective pixel is more accurate and comprehensive. Furthermore, the overall reflection trend of each area and the difference in the reflection of each area, as well as the difference in light between each area and the surrounding environment are comprehensively considered, and the reflection characteristics of each area are more comprehensively evaluated from multiple angles, so as to accurately determine the reflective area. Furthermore, by dynamically setting the adaptive scale parameters, the algorithm can better adapt to the lighting characteristics of different areas, more accurately remove uneven lighting, and provide more reliable image data for subsequent geographic scene recognition, thereby improving the accuracy and reliability of geographic scene recognition.

[0008] Furthermore, a method for dividing a remote sensing image into multiple regions based on an edge detection result is as follows: for any edge of the remote sensing image, a contour tracking algorithm is used, with any edge pixel point of the edge as a starting point, and tracking is performed along the edge in a clockwise or counterclockwise direction until returning to the starting point to obtain a closed contour, and the range determined by the closed contour is taken as a region.

[0009] This technical solution uses a contour tracking algorithm to track along the edges of remote sensing images, and can accurately find and define areas with clear boundaries. This method can separate different geographic features or target objects in the image, providing a basis for subsequent detailed analysis and processing of different areas.

[0010] Furthermore, the reflection probability of each pixel is determined based on the following formula: ; In the formula, For the The reflection possibility of each pixel, For the The brightness of a pixel, For the The saturation of each pixel, For the The average brightness of all pixels within the local range of pixels, is the mean brightness of all pixels in the remote sensing image.

[0011] This technical solution takes into account the local brightness information of each pixel and the global brightness information of the entire image. It can adapt to the differences in lighting conditions in different areas of the image. Regardless of whether the lighting is strong or weak, it can accurately judge the reflection possibility of the pixel through relative brightness comparison and saturation information, and will not be affected by changes in the overall light intensity.

[0012] Furthermore, the reflection consistency of the local range of each pixel point satisfies the following relationship: ; In the formula, For the The local reflective consistency of the pixels, For the The total number of pixels within the local range of pixels, and are the serial numbers of the pixels within the local range. For the The local range of the pixel The reflection possibility of each pixel, For the The local range of the pixel The reflection possibility of each pixel, For the The average of the reflection possibilities of all pixels within the local range of a pixel.

[0013] This technical solution calculates the reflection possibility of all pixels in the local range of each pixel by constructing an autocorrelation function, which can fully consider the relationship between each pixel and other pixels. It not only pays attention to the reflection possibility of a single pixel, but also captures the synergy and correlation between pixels in the local area in the form of an autocorrelation function.

[0014] Furthermore, the reflection comprehensive determination coefficient of each pixel point is determined based on the following formula: ; In the formula, For the The comprehensive reflection determination coefficient of each pixel point is: For the The contrast of the area where the pixel is located, For the The local reflective consistency of the pixels, For the The reflection possibility of each pixel.

[0015] This technical solution combines the reflection possibility of pixels, the reflection consistency of the local range and the contrast of the area where the pixels are located, comprehensively considering multiple factors that affect the reflection of pixels. It can accurately evaluate the degree of reflection of each pixel in the image, reducing the possibility of misjudgment and omission of reflective pixels.

[0016] Furthermore, before determining the comprehensive reflection determination coefficient of each pixel point, the method further includes: The reflection possibility of each pixel, the reflection consistency of the local range of each pixel, and the contrast of the area where each pixel is located are normalized so that the value is within Within the range.

[0017] Furthermore, the contrast of the area where each pixel is located is determined by: The standard deviation of the brightness of all pixels in the area where each pixel is located is used as the contrast of the area where each pixel is located; alternatively, the brightness co-occurrence matrix of the area is obtained based on the brightness of all pixels in the area where each pixel is located, and the method for obtaining the brightness co-occurrence matrix is ​​consistent with the method for obtaining the grayscale co-occurrence matrix; the contrast of the area is calculated based on the brightness co-occurrence matrix, and the calculation process is consistent with the calculation process of calculating the contrast of the area based on the grayscale co-occurrence matrix.

[0018] Furthermore, the reflectivity of each area is determined based on the following formula: ; In the formula, For the The comprehensive reflection determination coefficient of each area is: For the The mean value of the comprehensive reflection determination coefficient of all pixels in the area, For the The number of reflective pixels in an area, For the The discrete degree of the reflection comprehensive determination coefficient of all pixels in the area, For the The average brightness of the area, For the The average brightness of all adjacent areas of an area, is a natural exponential function; wherein the discrete degree of the reflection comprehensive determination coefficient of all pixel points is determined based on the standard deviation of the reflection comprehensive determination coefficient of all pixel points.

[0019] This technical solution comprehensively and meticulously measures the reflective possibility of an area by taking into account factors such as the mean of the comprehensive reflection determination coefficient of pixels in the area, the number of reflective pixels, the degree of discreteness, and the average brightness difference between the area and adjacent areas. It can accurately characterize the reflective characteristics of the area from multiple dimensions and avoid the limitations of single-factor judgment.

[0020] Furthermore, the adaptive scale parameter of the multi-scale Retinex algorithm when processing each reflective area is dynamically set based on the following formula: ;In the formula, For the Adaptive scale parameter of the reflective area, is the maximum value of the preset scale parameter, is the minimum value of the preset scale parameter, is the maximum value of the density of all reflective areas, is the minimum density of all reflective areas, For the The density of each reflective area is determined by the ratio of the number of reflective pixels in each reflective area to the number of all pixels in each reflective area.

[0021] This technical solution fully considers the differences in the density of different reflective areas. The obtained adaptive scale parameters enable the multi-scale Retinex algorithm to perform comprehensive illumination correction based on the reflective characteristics of different reflective areas of the image, avoiding the shortcomings of traditional methods and improving the processing effect of reflective areas of remote sensing images.

[0022] Furthermore, the method for geographic scene recognition based on remote sensing images after removing uneven illumination is as follows: The image feature extraction algorithm is used to extract features from the remote sensing image after removing the uneven illumination, and the extracted features are matched with the pre-constructed feature library of aquaculture geographical scenes to realize the recognition of the geographical scenes of the aquaculture area.

[0023] The present invention has the following effects: The present invention can accurately identify reflective pixels and reflective areas through multi-dimensional and detailed analysis of each pixel point and each area of ​​the remote sensing image of the aquaculture area, and then more accurately remove uneven lighting by dynamically setting the algorithm scale parameters, providing more reliable image data for subsequent geographic scene recognition, thereby improving the accuracy and reliability of geographic scene recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic flow chart of the method of the present invention; Figure 2 It is a schematic diagram of the method flow of step S3 of the present invention. DETAILED DESCRIPTION

[0025] Reference Figure 1 The present invention provides a method for identifying geographic scenes based on image processing, comprising steps S1 to S5: S1: Collect remote sensing images of aquaculture areas.

[0026] Drones equipped with high-resolution optical sensors are preferred as collection equipment. With their flexible maneuverability, they can collect multiple remote sensing images of the aquaculture area to ensure coverage of the entire aquaculture area.

[0027] S2: Divide the remote sensing image into multiple regions according to the edge detection result of the remote sensing image.

[0028] For any remote sensing image, a backup is first performed to obtain a backup image. After the backup image is grayed, the Canny algorithm is used for edge detection to obtain all edges in the backup image. All edges of the original remote sensing image are determined based on all edges in the backup image.

[0029] Then, a contour tracking algorithm is used, for example, the classic Suzuki contour tracking method, which starts from any edge pixel point of each edge of the remote sensing image and continues to track along the edge in a clockwise or counterclockwise direction until it returns to the starting point to obtain a closed contour. By processing each edge of the entire remote sensing image, all closed contours in the remote sensing image can be found, and each closed contour corresponds to an area in the remote sensing image.

[0030] S3: Determine the reflective area.

[0031] In remote sensing images of aquaculture areas, in order to remove the impact of light on image quality, it is necessary to determine which areas are reflective areas to improve the pertinence and accuracy of subsequent analysis. At the same time, to determine the reflective area, it is necessary to first determine which pixels are reflective pixels (pixels that may be in the reflective area). The more reflective pixels a region contains, the more likely it is that the region is a reflective area.

[0032] Based on this, execute Figure 2 Steps shown: S31: Determine the reflectivity of each pixel using the brightness and color saturation of the remote sensing image.

[0033] In aquaculture scenes, reflective areas usually have the characteristics of high brightness and low saturation. When light shines on the surface of an object, the area appears to be bright. The light is mainly reflected directly from the light source and contains multiple wavelengths of light, similar to white light. The color of the object itself is determined by its absorption and reflection of specific wavelengths of light. When the light is mixed with the color of the object itself, it dilutes the purity of the original color of the object, making the color look lighter and less saturated.

[0034] Specifically, first converting the image from RGB color space to HSV color space can more directly obtain the brightness and saturation information of the pixels, which is very effective for identifying reflective areas with high brightness and low saturation. Then calculate the reflection possibility of each pixel based on the brightness and saturation of each pixel. Then, analyze the reflection consistency of the local range of each pixel. If the reflection possibility of a pixel itself is greater and the reflection consistency of the local range of the pixel is higher, it means that the pixel is more likely to be in a reflective area.

[0035] In one embodiment, for each pixel in any area, the surrounding The range of pixels is taken as the local range of the pixel. The size of this range is an empirical value and can be set by yourself.

[0036] Calculate the reflectivity of each pixel:

[0037] In the formula, For the The reflection possibility of each pixel, For the The brightness of each pixel. The greater the brightness, the more likely it is to be in the reflective area. For the The saturation of each pixel. The lower the saturation, the more likely it is to be in the reflective area. For the The average brightness of all pixels within the local range of pixels, is the mean brightness of all pixels in the remote sensing image.

[0038] In this formula, As The weight coefficient is adjusted. When the brightness mean of the local range of pixels (all pixels included) is relatively large compared to the brightness mean of the entire remote sensing image (all pixels included), that is, Greater than , explain The local area of ​​the pixel is brighter, The pixel is likely to be affected by light. will be larger to highlight the On the contrary, when the The brightness mean of the local range of pixels (all pixels included) is relatively small compared to the brightness mean of the entire remote sensing image (all pixels included), that is, Less than ,at this time, will be smaller to weaken the The potential reflective properties of each pixel.

[0039] In short, while considering the impact of light on pixels, the local range of pixels is compared with the overall image, highlighting the impact of uneven light on pixels in different areas. When the local range is brighter than the overall image, it means that the local area is greatly affected by light. Since local and global brightness information is considered, it can adapt to complex and changeable lighting conditions in the image. Whether it is an area with strong or weak overall light, the possibility of reflection of pixels can be accurately judged through relative brightness comparison and saturation information.

[0040] S32: Determine the reflective consistency of the local range of each pixel by constructing an autocorrelation function.

[0041] Since the reflective area is an area composed of continuous pixels, the pixels in the area will show similar reflective characteristics, which means that the brightness change pattern of the pixels in the area is consistent. Therefore, the stronger the consistency of the reflective possibility of the pixels in the local range of a certain pixel, the more consistent the brightness change pattern of the pixels in the local range of the pixel, which reflects that the pixel is more likely to be in the reflective area.

[0042] In one embodiment, the autocorrelation function of all pixels in the local range is constructed based on the reflection possibility of all pixels in the local range of each pixel point:

[0043] In the formula, For the The reflection consistency of the local range of pixels, that is, the function value of the autocorrelation function, The larger the The more consistent the reflection possibility of all pixels in the local range of the pixel, the brighter the characteristics of all pixels in the local range. The more pixels there are, the more likely they are to be in the reflective area. For the The total number of pixels within the local range of pixels, and are the serial numbers of the pixels within the local range. For the The local range of the pixel The reflection possibility of each pixel, For the The local range of the pixel The reflection possibility of each pixel, For the The average of the reflection possibilities of all pixels within the local range of a pixel.

[0044] This formula calculates the reflection possibility of all pixels in the local range in pairs, comprehensively considering the relationship between each pixel and other pixels. It not only pays attention to the reflection possibility of a single pixel, but also captures the coordination and correlation between pixels in the local area in the form of autocorrelation function, which can well reflect the characteristics of the reflective area.

[0045] S33: Quantify the contrast of the area where each pixel is located.

[0046] From a visual point of view, the reflective area often appears as a relatively uniform bright spot, lacking obvious contrast between light and dark, and exhibiting the characteristics of low contrast. Therefore, the contrast of the area can be determined by the brightness value of the pixels in each area.

[0047] In one embodiment, a method of obtaining the contrast of each region is: Get the brightness of each pixel in each area (V channel value), and use the standard deviation of the brightness of all pixels as the contrast of the area. Standard deviation is an indicator used in statistics to measure the degree of dispersion of a set of data. The essence of contrast reflects the brightness difference between different areas in the image. When the standard deviation of the brightness values ​​of all pixels in a certain area is large, it means that the distribution of the brightness values ​​of all pixels in the area is more dispersed, that is, the greater the possibility of the existence of pixels of different brightness, the higher the contrast.

[0048] In one embodiment, another method of obtaining the contrast of each region is: Get the brightness of each pixel in each area, and calculate the brightness co-occurrence matrix of each area in the same way as the grayscale co-occurrence matrix. The only difference between the brightness co-occurrence matrix and the grayscale co-occurrence matrix is ​​that one is based on the grayscale value of the pixel and the other is based on the brightness of the pixel.

[0049] The contrast of each region is calculated based on the brightness co-occurrence matrix of the region (the method is consistent with the method of calculating the contrast based on the grayscale co-occurrence matrix):

[0050] In this formula, For the The contrast of the area where the pixel is located, Indicates that at a distance of , angle is Under the condition of The brightness of the area where the pixel is located is and The frequency of pixel combinations, and All are brightness (V channel value).

[0051] For areas with texture features, methods similar to the grayscale co-occurrence matrix can better capture the contrast changes caused by texture. In aquaculture scenes, the texture on the surface of the aquaculture facilities and the ripples on the water surface will cause the brightness of the pixels to show a specific distribution pattern. The co-occurrence matrix can capture this change pattern and accurately calculate the contrast of each area.

[0052] S34: Calculate the reflection comprehensive determination coefficient of each pixel point, and determine the reflection pixel point according to the size of the reflection comprehensive determination coefficient.

[0053] First, normalize the contrast of each pixel area, the reflection possibility of each pixel, and the reflection consistency of each pixel in the local range so that the values ​​are within the range of The range is convenient for subsequent comparison and calculation.

[0054] In one embodiment, the reflection comprehensive determination coefficient of each pixel is calculated based on the following formula:

[0055] In this formula, For the The comprehensive reflection determination coefficient of each pixel point is: For the The contrast of the area where the pixel is located, For the The local reflective consistency of the pixels, For the The reflection possibility of each pixel.

[0056] If a pixel is more likely to reflect and the reflection consistency between the pixel and the pixels in its local range is stronger, it means that the pixel is more likely to be in the reflective area. At the same time, if the contrast of the area where the pixel is located is smaller, the possibility of the pixel being located in the reflective area is further increased. These indicators are combined to obtain the comprehensive reflection determination coefficient of each pixel.

[0057] In one embodiment, the method for determining the reflective pixel point according to the size of the reflective comprehensive determination coefficient of each pixel point is: directly setting the threshold of the reflective comprehensive determination coefficient to 0.8 (empirical value), and taking the pixel points greater than 0.8 as reflective pixel points.

[0058] In one embodiment, another method for determining the reflective pixel point according to the size of the reflective comprehensive determination coefficient of each pixel point is: Based on the maximum inter-class variance method, a threshold is automatically found to divide the data in the image into two categories, where the data in one category are all smaller than the threshold, and the data in the other category are all larger than the threshold.

[0059] Specifically, a histogram is constructed based on the reflection comprehensive determination coefficients of all pixels in each area. The horizontal axis of the histogram is the reflection comprehensive determination coefficient value range of all pixels, and the vertical axis is the number of pixels corresponding to each reflection comprehensive determination coefficient. The maximum inter-class variance method is used to segment the histogram to obtain the optimal segmentation threshold of the reflection comprehensive determination coefficient, and the pixels whose reflection comprehensive determination coefficient is greater than the optimal segmentation threshold are regarded as reflection pixels.

[0060] In summary, by preliminarily calculating the reflection possibility of each pixel and then using the autocorrelation function to further verify the reflection consistency of other pixels within the local range of each pixel, while taking into account the contrast of the area where each pixel is located, the reflective pixels can be determined more accurately, eliminating some misjudgments caused by isolated outliers or noise, thereby making the final determined reflective pixels more reasonable and rigorous.

[0061] S35: Determine the reflective area according to the characteristics of the reflective pixels contained in each area.

[0062] First, based on the mean and discreteness of the reflection comprehensive determination coefficient of all pixels in each area, combined with the number of reflective pixels in each area and the brightness difference between each area and the surrounding area, the reflection possibility of each area is determined. The reflection possibility of each area is compared with the preset reflection possibility threshold to determine the reflective area.

[0063] In one embodiment, the reflectivity of each region is determined based on the following formula:

[0064] In the formula, For the The reflection comprehensive determination coefficient of each area, the larger the value, the more likely the area is a reflection area. For the The mean value of the reflection comprehensive determination coefficient of all pixels in the area reflects the average level of the reflection comprehensive determination coefficient of all pixels in the area. The larger it is, the more likely the area is a reflective area. For the The more reflective pixels an area has, the more likely it is that the area is an actual reflective area rather than a small area formed by scattered noise points. For the The degree of dispersion of the reflection comprehensive determination coefficient of all pixels in the area, that is, the standard deviation of the reflection comprehensive determination coefficient of all pixels, The smaller the The more consistent the reflective characteristics of all pixels in a region are, the more likely it is a real reflective region. When it is 0, it means The comprehensive reflection coefficients of all pixels in the area are exactly the same, which is the ideal reflection area. In order to avoid the denominator being 0, Add 1. For the The average brightness of the area, that is, The average brightness of all pixels in the region, For the The average brightness of all adjacent areas of an area, is a natural exponential function.

[0065] In this formula, when Greater than , No. The brighter an area is than the surrounding area, If the value is greater than 1, it will greatly increase the The possibility that the current area is a reflective area. For example, if the current area is a reflective area, its average brightness is usually higher than the average brightness of the surrounding non-reflective areas. This index term can effectively highlight the positive impact of this brightness difference on reflective judgment. Less than At that time, The less bright an area is compared to the surrounding areas, A value less than 1 will reduce the The possibility that the area is a reflective area.

[0066] If the average level of the reflection comprehensive determination coefficient of all pixels in a certain area is higher and the standard deviation of the reflection comprehensive determination coefficient is small, and if there are more reflective pixels in the area and the area is brighter relative to the surrounding area, then the area is more likely to be a reflective area.

[0067] In one embodiment, the method of comparing the reflection possibility of each area with a preset reflection possibility threshold to determine the reflective area is: setting the threshold of the comprehensive reflection determination coefficient of all areas to 0.9, and treating areas greater than 0.9 as reflective areas, and treating areas less than or equal to 0.9 as non-reflective areas.

[0068] S4: Dynamically set the scale parameter of each reflective area when executing the Retinex algorithm according to the density of different reflective areas to remove the uneven illumination of the remote sensing image.

[0069] The traditional Retinex algorithm convolves the image with a Gaussian function with a preset fixed standard deviation to simulate the visual perception of the human eye and analyze the image to separate the illumination and reflection components. ), the result obtained by Gaussian function convolution retains rich high-frequency detail information in the image, can reflect fine features such as object texture and edge, and corresponds to the detail scale of the image; when the preset standard deviation is larger (such as ), the convolution result of the Gaussian function can retain low-frequency information and pay more attention to the overall brightness distribution of the image, which is used to capture large-scale illumination changes. Then, specific operations are performed on the images of each scale obtained after convolution to separate the illumination component and the reflection component of the image. The illumination component is then adjusted, such as gain, offset, etc., to improve the uneven illumination of the image and enhance the contrast.

[0070] Therefore, the value range of the scale parameter of the Retinex algorithm is obtained in advance, and the ratio of the number of reflective pixels in each area to the number of all pixels in the reflective area is used as the density of the area.

[0071] Then, based on the density, the adaptive scale parameters of the multi-scale Retinex algorithm are dynamically set when processing each reflective area:

[0072] In the formula, For the Adaptive scale parameter of the reflective area, is the maximum value of the preset scale parameter, is the minimum value of the preset scale parameter, is the maximum value of the density of all reflective areas, is the minimum density of all reflective areas, For the The density of the reflective area.

[0073] In this formula, This part means that The density of the reflective areas Perform normalization.

[0074] when At that time, The density of the reflective area reaches its maximum value. , which means that for the most densely populated reflective areas, a smaller scale parameter is selected, because the more densely populated reflective areas may have more complex lighting anomalies, and it is necessary to finely capture local lighting changes. A smaller scale parameter can better preserve the high-frequency details of the image and can accurately process complex lighting.

[0075] when At that time, The density of the reflective area is the minimum. This means that for the reflective area with the lowest density, a larger scale parameter is selected. The illumination change of the reflective area with low density is relatively simple. A larger scale parameter can ensure the removal of uneven illumination while paying attention to the overall brightness distribution of the image, thereby improving the execution efficiency of the algorithm because there is no need to process the details too finely.

[0076] along with from Gradually increase to When, that is The bigger, The smaller the value (from 1 to 0), The value will also be smaller (from Reduce to ) realizes the inverse correlation between the scale parameter and the density of the reflective area, which meets the requirement of dynamically selecting the scale parameters of different areas according to the density of different reflective areas, and realizes the adaptive scale parameters of the Retinex algorithm when processing different reflective areas, that is, for reflective areas with high density, a smaller scale parameter is selected, because the lighting anomaly may be more complicated, so as to capture the local lighting changes more finely; for reflective areas with low density, a relatively large scale parameter is selected, which improves the execution efficiency of the algorithm while ensuring the effect of removing uneven lighting.

[0077] Through this operation, the appropriate scale parameters can be dynamically and scientifically selected according to the specific conditions of each reflective area to better execute the multi-scale Retinex algorithm. This dynamic adaptive method enables the algorithm to automatically select the appropriate processing method according to the characteristics of different areas in the image, rather than using fixed parameters to operate the entire image, which improves the algorithm's adaptability and robustness to various lighting conditions and image content, and can more accurately remove uneven lighting and enhance image contrast and quality.

[0078] S5: Geographic scene recognition based on remote sensing images after removing uneven illumination.

[0079] Pre-built feature library for aquaculture geographic scenarios: Multi-source data acquisition: collect remote sensing images of the current aquaculture area in different seasons and weather conditions in advance, and record the geometric characteristics of each land feature, such as the size, circumference, shape, type and other information of the breeding facilities, as well as the spatial characteristics of each land feature, such as the distribution location of the breeding facilities.

[0080] Feature extraction: The reflectance values ​​of each land object in multiple bands are extracted and normalized as the spectral characteristics of each land object. The gray level co-occurrence matrix (GLCM) is used to calculate the texture parameters of the corresponding area of ​​each land object, such as the entropy value of the gray level co-occurrence matrix, as the surface texture characteristics of each land object.

[0081] Feature vector construction and storage: The geometric features, spatial features, spectral features and surface texture features of each feature are integrated into a feature vector, and each feature vector corresponds to a feature type.

[0082] Extract the features of each area in the remote sensing image to be identified and compare them with the feature library: In the remote sensing image to be identified, each region corresponds to a possible type of ground object. For each region, the spectral features, spatial features, spectral features and surface texture features of the region are extracted according to the method used to construct the feature library, and integrated into the feature vector of the region. Feature matching to complete geographic scene recognition: Calculate the Euclidean distance between the feature vector of each region and the feature vector of each feature in the feature library. The smaller the Euclidean distance, the higher the feature similarity. The feature vector of the region is compared with the feature vector of the The feature vector of the object with the smallest Euclidean distance is determined. The type of land feature corresponding to the area is This operation is performed on each area of ​​the remote sensing image to be identified, thus realizing geographic scene recognition based on remote sensing images.

Claims

1. A geographic scene recognition method based on image processing, characterized in that: include: Collect remote sensing images of aquaculture areas, divide them into multiple areas according to the edge detection results of the remote sensing images, and convert the remote sensing images from RGB color space to HSV color space; In each area, the reflection possibility of each pixel is determined according to the brightness and saturation of each pixel, and the reflection consistency of the local range of each pixel is evaluated; the reflection comprehensive determination coefficient of each pixel is determined in combination with the reflection possibility of each pixel, the reflection consistency of the local range of each pixel, and the contrast of the area where each pixel is located; the reflection comprehensive determination coefficient of each pixel is compared with a preset reflection comprehensive determination coefficient threshold to determine the reflective pixel; Based on the mean and discreteness of the reflection comprehensive determination coefficients of all pixels in each area, combined with the number of reflective pixels in each area and the brightness difference between each area and the surrounding areas, the reflection possibility of each area is determined; the reflection possibility of each area is compared with a preset reflection possibility threshold to determine the reflective area; The density of each reflective area is determined, and based on the density, the adaptive scale parameters of the multi-scale Retinex algorithm when processing each reflective area are dynamically set to remove the uneven illumination of the remote sensing image, and geographic scene recognition is performed based on the remote sensing image after the uneven illumination is removed.

2. The geographic scene recognition method based on image processing according to claim 1, characterized in that: The method of dividing the remote sensing image into multiple regions based on the edge detection results is: For any edge of the remote sensing image, the contour tracking algorithm is used, with any edge pixel point of the edge as the starting point, and the edge is tracked in a clockwise or counterclockwise direction until it returns to the starting point to obtain a closed contour, and the range determined by the closed contour is taken as a region.

3. The geographic scene recognition method based on image processing according to claim 1, characterized in that: The reflectivity of each pixel is determined based on the following formula: ； In the formula, For the The reflection possibility of each pixel, For the The brightness of a pixel, For the The saturation of each pixel, For the The average brightness of all pixels in the local range of pixels, is the mean brightness of all pixels in the remote sensing image.

4. The method for geographic scene recognition based on image processing according to claim 1, characterized in that: The reflective consistency of the local range of each pixel satisfies the following relationship: ； In the formula, For the The local reflective consistency of the pixels, For the The total number of pixels within the local range of pixels, and are the serial numbers of the pixels within the local range. For the The local range of the pixel The reflection possibility of each pixel, For the The local range of the pixel The reflection possibility of each pixel, For the The average of the reflection possibilities of all pixels within the local range of pixels.

5. The method for geographic scene recognition based on image processing according to claim 1, characterized in that: The comprehensive reflection determination coefficient of each pixel is determined based on the following formula: ； In the formula, For the The comprehensive reflection determination coefficient of each pixel point is: For the The contrast of the area where the pixel is located, For the The local reflective consistency of the pixels, For the The reflection possibility of each pixel.

6. The method for geographic scene recognition based on image processing according to claim 1, characterized in that: Before determining the comprehensive reflection determination coefficient of each pixel point, it also includes: The reflection possibility of each pixel, the reflection consistency of the local range of each pixel, and the contrast of the area where each pixel is located are normalized so that the value is within Within the range.

7. The method for geographic scene recognition based on image processing according to claim 5, characterized in that: The contrast of each pixel area is determined as follows: The standard deviation of the brightness of all pixels in the area where each pixel is located is used as the contrast of the area where each pixel is located; alternatively, the brightness co-occurrence matrix of the area is obtained based on the brightness of all pixels in the area where each pixel is located, and the method for obtaining the brightness co-occurrence matrix is ​​consistent with the method for obtaining the grayscale co-occurrence matrix; the contrast of the area is calculated based on the brightness co-occurrence matrix, and the calculation process is consistent with the calculation process of calculating the contrast of the area based on the grayscale co-occurrence matrix.

8. The method for geographic scene recognition based on image processing according to claim 5, characterized in that: The reflectivity of each area is determined based on the following formula: ； In the formula, For the The comprehensive reflection determination coefficient of each area is: For the The mean value of the comprehensive reflection determination coefficient of all pixels in the area, For the The number of reflective pixels in an area, For the The discrete degree of the reflection comprehensive determination coefficient of all pixels in the area, For the The average brightness of the area, For the The average brightness of all adjacent areas of an area, is the natural exponential function; The degree of dispersion of the comprehensive reflection determination coefficients of all pixel points is determined based on the standard deviation of the comprehensive reflection determination coefficients of all pixel points.

9. The method for geographic scene recognition based on image processing according to claim 1, characterized in that: The dynamic setting of the adaptive scale parameters of the multi-scale Retinex algorithm when processing each reflective area is based on the following formula: ； In the formula, For the Adaptive scale parameter of the reflective area, is the maximum value of the preset scale parameter, is the minimum value of the preset scale parameter, is the maximum value of the density of all reflective areas, is the minimum density of all reflective areas, For the The density of the reflective areas; The density of each reflective area is determined by the ratio of the number of reflective pixels in each reflective area to the number of all pixels in each reflective area.

10. The method for geographic scene recognition based on image processing according to claim 1, characterized in that: The method for geographic scene recognition based on remote sensing images after removing uneven illumination is: The image feature extraction algorithm is used to extract features from the remote sensing image after removing the uneven illumination, and the extracted features are matched with the pre-constructed feature library of aquaculture geographical scenes to realize the recognition of the geographical scenes of the aquaculture area.

Images