Multi-scale image segmentation reconstruction method and system applied to remote sensing image

Abstract

The invention discloses a multi-scale image segmentation reconstruction method and system applied to a remote sensing image, and relates to the field of image segmentation reconstruction, and the method comprises the steps: obtaining a high-definition remote sensing image of a to-be-detected region, and building a detection region high-definition remote sensing data image set; extracting color and contour features of a to-be-detected area, and establishing a standard feature map data set; based on a dark channel prior defogging algorithm, the influence of haze on the quality of the remote sensing image is removed; segmenting the collected remote sensing image based on the color and contour features of the to-be-detected area; calculating the ambiguity of each part of the remote sensing image, and judging the quality of each part of the remote sensing image; on the basis of a neural network model, replacing and reconstructing a part with poor quality of the remote sensing image; and based on a Laplacian pyramid algorithm, performing optimization processing on the remote sensing image after replacement and reconstruction processing. The method has the advantages that the output quality of the remote sensing image is effectively improved, and the influence of haze and partial region blurring or shielding is reduced.

Description

Technical Field

[0001] The present invention relates to the field of image segmentation and reconstruction, and specifically to a multi-scale image segmentation and reconstruction method and system for remote sensing images. Background Art

[0002] Remote sensing images are image data of ground or surface objects collected through remote sensing technology. Through remote sensing pictures, the environmental changes, resource distribution, and disaster warnings of the monitored area can be monitored in real time. Therefore, it is of great significance to ensure the quality of remote sensing pictures and improve their accuracy and reliability. Since the acquisition of remote sensing pictures is affected by the environment at that time, a hazy or cloudy environment is likely to cause a decline in the quality of remote sensing pictures, which is not conducive to the analysis of the area to be detected through remote sensing pictures. Therefore, it is of great significance to improve the quality of remote sensing pictures, remove the influence of haze through image segmentation and reconstruction, and improve the clarity and accuracy of remote sensing pictures.

[0003] The existing image segmentation and reconstruction technologies improve the reduction of the influence of fog on remote sensing pictures and improve the quality of remote sensing pictures by combining the dark channel prior algorithm and U-Net for remote sensing images. For example, Chinese Patent CN113763488A discloses a "method for removing fog degree of remote sensing images by combining the dark channel prior algorithm and U-Net". This technology reconstructs one channel of the remote sensing image respectively, and then combines it with the other two channels to obtain three new fog degree images. At the same time, the U-Net network of the deep learning method is used to segment the thick fog area of the remote sensing image, estimate the atmospheric light value, obtain three transmission maps respectively according to the three new fog degree images, and finally use the atmospheric scattering model to complete defogging for the three new fog degree images respectively, and obtain the final defogging effect after weighting the image weights. Although such technologies can reduce the influence of fog on remote sensing images, they still cannot avoid the influence of fog. Even in a harsh environment, this technology is difficult to achieve the expected effect. Most importantly, it is difficult to avoid occlusion by other obstacles, resulting in local occlusion shadows, which causes the quality of the final remote sensing image to be poor. Summary of the Invention

[0004] To solve the above technical problems, a multi-scale image segmentation and reconstruction method and system for remote sensing images are provided. This technical solution solves the problems proposed in the above background art that it is always impossible to avoid the influence of fog, and even in a harsh environment, this technology is difficult to achieve the expected effect. Most importantly, it is difficult to avoid occlusion by other obstacles, resulting in local occlusion shadows, which causes the quality of the final remote sensing image to be poor.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows: A multi-scale image segmentation and reconstruction method for remote sensing images, comprising: Obtaining a high-definition remote sensing image of the area to be detected based on a remote sensing image database, and establishing a high-definition remote sensing data atlas for the detection area; Extracting the color and contour features of the area to be detected according to the type of event in the area to be detected, and establishing a standard feature map data set; Removing the influence of haze on the quality of the remote sensing image based on the dark channel prior dehazing algorithm according to the high-definition remote sensing data atlas of the detection area; Segmenting the collected remote sensing image based on the color and contour features of the area to be detected according to the high-definition remote sensing data atlas of the detection area; Obtaining the segmented remote sensing image, calculating the blurriness of each part of the remote sensing image, and judging the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; Replacing and reconstructing the parts with poor quality of the remote sensing image based on a neural network model according to the evaluation results of the quality of each part of the remote sensing image; Optimizing the processed remote sensing image after replacement and reconstruction based on the Laplacian pyramid algorithm.

[0006] Preferably, the obtaining a high-definition remote sensing image of the area to be detected based on a remote sensing image database and establishing a high-definition remote sensing data atlas for the detection area specifically includes: Obtaining a high-definition remote sensing image of the area to be detected based on a remote sensing image database, wherein the high-definition remote sensing image of the area to be detected mainly includes: all-round image information of the area to be detected and color changes in four seasons; Sorting out, classifying and numbering the high-definition remote sensing images of the detection area, and establishing a high-definition remote sensing data atlas for the detection area.

[0007] Preferably, the extracting the color and contour features of the area to be detected according to the type of event in the area to be detected and establishing a standard feature map data set specifically includes: Extracting the color and contour features of the area to be detected according to the type of event in the area to be detected, wherein the event types include: buildings, vegetation, rocks, rivers, crops, terrain and man-made objects; Based on the color and contour features of the area to be detected and the high-definition remote sensing data atlas of the detection area, screening out the color and contour feature pictures of the area to be detected that meet the standards, and establishing a standard feature map data set.

[0008] Preferably, the removing the influence of haze on the quality of the remote sensing image based on the dark channel prior dehazing algorithm according to the high-definition remote sensing data atlas of the detection area specifically includes: According to the high-definition remote sensing data atlas of the detection area, perform grayscale processing on the high-definition remote sensing data atlas of the detection area to obtain the grayscaled high-definition remote sensing data atlas of the detection area; According to the grayscaled high-definition remote sensing data atlas of the detection area, optimize the quality of the high-definition remote sensing data atlas of the detection area based on the Gaussian filtering algorithm; According to the optimized high-definition remote sensing data atlas of the detection area, obtain the dark channel atlas of the optimized high-definition remote sensing data atlas of the detection area; Among them, the dark channel map refers to a grayscale map obtained by calculating the minimum value of the RGB three color channels of each pixel in the local area of the image and having the same size as the original image, and this grayscale map is the dark channel map; According to the dark channel atlas of the optimized high-definition remote sensing data atlas of the detection area, obtain the atmospheric light value of the dark channel prior dehazing algorithm; According to the atmospheric scattering coefficient and the depth of field of the scene light, obtain the light transmission rate of each pixel point in the dark channel atlas of the high-definition remote sensing data atlas of the detection area through the fog; According to the dark channel prior dehazing algorithm, establish an optimization model for the high-definition remote sensing data atlas of the detection area to remove the influence of haze on the quality of remote sensing images; The expression of the dark channel prior dehazing algorithm is: In the formula, is the pixel value after Gaussian filtering at the th row and th column of the observed hazy image, is the pixel value at the th row and th column of the fog-free image to be restored, is the light transmission rate through the fog, is the atmospheric light value, is the atmospheric scattering coefficient, is the depth of field of the scene light.

[0009] Preferably, the specific steps of segmenting the collected remote sensing image based on the color and contour features of the area to be detected according to the high-definition remote sensing data atlas of the detection area include: According to the high-definition remote sensing data atlas of the detection area, mark and number different positions in the high-definition remote sensing data atlas of the detection area according to the color and contour features of the area to be detected; According to the marked and numbered high-definition remote sensing data atlas of the detection area, perform image segmentation on the real-time collected remote sensing image according to the marked and numbered positions, and confirm that the color and contour features of each segmented area are the most obvious; According to the standard feature map dataset, set the segmented atlas of the standard feature map dataset as the target standard atlas.

[0010] Preferably, the obtaining of the segmented remote sensing image, calculating the blurriness of each part of the remote sensing image, and judging the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image specifically include: According to the result of image segmentation of the remotely sensed image collected in real time according to the positions of the marks and numbers, extract the optimized pixel values of the segmented real-time remotely sensed image; According to the segmentation result of the target standard atlas, extract the optimized pixel values of the segmented standard remotely sensed image; Through the normalization formula, perform the dimension elimination optimization process on the optimized pixel values of the real-time remotely sensed image and the optimized pixel values of the standard remotely sensed image; According to the optimized pixel values of the processed real-time remotely sensed image and the optimized pixel values of the standard remotely sensed image, calculate the blurriness of each part of the remote sensing image; According to the blurriness of each part of the remote sensing image, set a threshold for judging the quality of each part of the remote sensing image; Judge whether the blurriness of the remote sensing image is greater than the threshold. If so, it means that the quality of the current part of the remote sensing image is poor and does not meet the expectation. If not, it means that the quality of the current part of the remote sensing image meets the expectation; The expression for the blurriness of each part of the remote sensing image is: In the formula, is the blurriness of each part of the remote sensing image, is the correlation factor, is the optimized pixel value of the th row and th column of the segmented real-time remotely sensed image after optimization, is the optimized pixel value of the th row and th column of the segmented standard remotely sensed image after optimization, is the number of optimized pixel values of the segmented real-time remotely sensed image, is the number of images with the same position pixel values in the target standard atlas.

[0011] Preferably, the replacing and reconstructing of the parts with poor quality of the remote sensing image based on the neural network model according to the evaluation results of the quality of each part of the remote sensing image specifically include: According to the target standard atlas, obtain the remotely sensed images in different states of the area to be detected, and establish the target atlas and the training atlas respectively; Take the training atlas and the evaluation results of the quality of each part of the remote sensing image as the input matrix, and the target atlas as the verification output matrix to construct the input-output matrix of the neural network; Based on a neural network model, the input-output matrix data is trained and learned. Through the training results of the model, the quality of each part of the newly received remotely sensed image collected in real time is evaluated; Determine whether the quality of each part of the newly received remotely sensed image collected in real time meets the requirements. If so, it means that the quality of this part of the picture meets the expectations, and the original picture of this part is output. If not, it means that the quality of this part of the picture does not meet the expectations, and this part of the image is replaced and reconstructed.

[0012] Preferably, the optimization processing of the remotely sensed image after replacement and reconstruction based on the Laplacian pyramid algorithm specifically includes: According to the reconstruction result of the remotely sensed image by the neural network model, by sampling and processing the result picture, the largest image and the smallest image are separated, and an image pyramid model is constructed; According to the image pyramid model, the image of the next layer is Gaussian blurred, and the even rows and columns of the blurred image are deleted, and this is repeated in turn to obtain a Gaussian pyramid model; According to the Gaussian pyramid model, based on the Laplacian pyramid algorithm, a reconstruction optimization model of the remotely sensed image is constructed to optimize the remotely sensed image after replacement and reconstruction processing; The Laplacian pyramid algorithm is: In the formula, is the image data of the th layer of the Laplacian pyramid, is the image data of the th layer of the Gaussian pyramid, is to map the pixel value of the in the Gaussian pyramid image to the position of the target image , that is, the process of upsampling, is the convolution symbol, is Gaussian kernel of.

[0013] Furthermore, this solution proposes a multi-scale image segmentation and reconstruction system for remotely sensed images, which is used to implement the multi-scale image segmentation and reconstruction method for remotely sensed images as described above, including: A data processing module, which is used to obtain a high-definition remotely sensed image of the area to be detected based on a remotely sensed image database and establish a high-definition remotely sensed data atlas of the detection area; according to the type of event in the area to be detected, extract the color and contour features of the area to be detected and establish a standard feature map data set; A haze removal and optimization module, which is used to remove the influence of haze on the quality of remotely sensed images based on the dark channel prior haze removal algorithm according to the high-definition remotely sensed data atlas of the detection area; An image segmentation module, which is used to segment the collected remote sensing image based on the high-definition remote sensing data atlas of the detection area and the color and contour features of the area to be detected; An image reconstruction and optimization module, which is used to obtain the segmented remote sensing image, calculate the blurriness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; according to the evaluation results of the quality of each part of the remote sensing image, based on a neural network model, replace and reconstruct the parts of the remote sensing image with poor quality; based on the Laplacian pyramid algorithm, optimize the processed remote sensing image after replacement and reconstruction.

[0014] Preferably, the data processing module specifically includes: A high-definition remote sensing atlas unit, which is used to obtain the high-definition remote sensing image of the area to be detected based on the remote sensing image database and establish a high-definition remote sensing data atlas of the detection area; A feature extraction unit, which is used to extract the color and contour features of the area to be detected according to the type of event in the area to be detected and establish a standard feature map data set; The image reconstruction and optimization module specifically includes: A blurriness unit, which is used to obtain the segmented remote sensing image, calculate the blurriness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; A quality evaluation unit, which is used to replace and reconstruct the parts of the remote sensing image with poor quality based on a neural network model according to the evaluation results of the quality of each part of the remote sensing image; A reconstruction and optimization unit, which is used to optimize the processed remote sensing image after replacement and reconstruction based on the Laplacian pyramid algorithm.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: By obtaining high-definition remote sensing data maps of the detection area and extracting the color and contour features of the area to be detected, a high-definition remote sensing data map set and a standard feature map data set of the detection area are established. Based on this, and using the dark channel prior dehazing algorithm, haze removal operations are performed on the real-time collected remote sensing images to improve the quality of the remote sensing images. Secondly, the blurriness of each part of the remote sensing image is determined through the optimized pixel values of the real-time remote sensing image and the optimized pixel values of the standard remote sensing image. Thresholds are set according to the blurriness of each part of the remote sensing image to judge the quality of each part of the remote sensing image. Furthermore, based on the evaluation results of the quality of each part of the remote sensing image and using a neural network model, the quality of each part of the newly received real-time collected remote sensing image is evaluated, so as to replace and reconstruct the parts with poor quality of the remote sensing image. Finally, based on the Laplacian pyramid algorithm, a remote sensing image reconstruction and optimization model is constructed to optimize the remote sensing image after replacement and reconstruction processing, effectively fusing the pictures of each segmented area, thus effectively improving the output quality of the remote sensing image, minimizing the impact of haze and partial area blurring or occlusion to the greatest extent, and ensuring the reliability and accuracy of the remote sensing image. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 FIG. is a flowchart of a multi-scale image segmentation and reconstruction method for remote sensing images according to the present invention; Figure 2 FIG. is a flowchart of removing the influence of haze on the quality of remote sensing images based on the dark channel prior dehazing algorithm according to the high-definition remote sensing data map set of the detection area of the present invention; Figure 3 FIG. is a flowchart of obtaining the segmented remote sensing image, calculating the blurriness of each part of the remote sensing image, and judging the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image according to the present invention; Figure 4 FIG. is a flowchart of optimizing the remote sensing image after replacement and reconstruction processing based on the Laplacian pyramid algorithm according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations.

[0018] Referring to Figure 1 as shown, a multi-scale image segmentation and reconstruction method for remote sensing images includes: Obtaining high-definition remote sensing images of the area to be detected based on a remote sensing image database and establishing a high-definition remote sensing data map set of the detection area; Extracting the color and contour features of the area to be detected according to the type of event in the area to be detected and establishing a standard feature map data set; According to the high-definition remote sensing data atlas of the detection area, based on the dark channel prior dehazing algorithm, the influence of haze on the quality of remote sensing images is removed; According to the high-definition remote sensing data atlas of the detection area, based on the color and contour features of the area to be detected, the collected remote sensing images are segmented; Obtain the segmented remote sensing images, calculate the blurriness of each part of the remote sensing images, and judge the quality of each part of the remote sensing images through the blurriness of each part of the remote sensing images; According to the evaluation results of the quality of each part of the remote sensing images, based on the neural network model, replace and reconstruct the parts with poor quality of the remote sensing images; Based on the Laplacian pyramid algorithm, optimize the processed remote sensing images after replacement and reconstruction.

[0019] It can be explained that in this solution, by obtaining the high-definition remote sensing data map of the detection area and extracting the color and contour features of the area to be detected, a high-definition remote sensing data atlas of the detection area and a standard feature map data set are established, and on this basis, based on the dark channel prior dehazing algorithm, haze removal operations are performed on the real-time collected remote sensing images, so as to improve the quality of remote sensing images. Secondly, through the pixel values of the optimized real-time remote sensing images and the pixel values of the optimized standard remote sensing images, the blurriness of each part of the remote sensing images is determined, and thresholds are set according to the blurriness of each part of the remote sensing images to judge the quality of each part of the remote sensing images. Furthermore, according to the evaluation results of the quality of each part of the remote sensing images, based on the neural network model, the quality of each part of the newly received real-time collected remote sensing images is evaluated, so as to replace and reconstruct the parts with poor quality of the remote sensing images. Finally, based on the Laplacian pyramid algorithm, a remote sensing image reconstruction and optimization model is constructed to optimize the processed remote sensing images after replacement and reconstruction, effectively fuse the pictures of each segmented area, thereby effectively improving the output quality of remote sensing images, reducing the influence of haze and partial area blurring or occlusion with the highest efficiency, and ensuring the reliability and accuracy of remote sensing images. Among them, the remote sensing image database obtains the remote sensing image data of the detection area through satellite remote sensing, aerial remote sensing and unmanned aerial vehicle remote sensing, and records and stores the historical remote sensing image data of the detection area.

[0020] Refer to Figure 2 As shown, the specific steps of removing the influence of haze on the quality of remote sensing images according to the high-definition remote sensing data atlas of the detection area and based on the dark channel prior dehazing algorithm include: According to the high-definition remote sensing data atlas of the detection area, perform grayscale processing on the high-definition remote sensing data atlas of the detection area to obtain the grayscale high-definition remote sensing data atlas of the detection area; According to the grayscale high-definition remote sensing data atlas of the detection area, based on the Gaussian filtering algorithm, optimize the quality of the high-definition remote sensing data atlas of the detection area; According to the optimized high-definition remote sensing data atlas of the detection area, obtain the dark channel atlas of the optimized high-definition remote sensing data atlas of the detection area; Among them, the dark channel map refers to a grayscale map obtained by calculating the minimum value of the RGB three color channels of each pixel in the local area of the image, and this grayscale map is the dark channel map; According to the dark channel atlas of the optimized high-definition remote sensing data atlas of the detection area, obtain the atmospheric light value of the dark channel prior defogging algorithm; According to the atmospheric scattering coefficient and the depth of field of the scene light, obtain the light transmittance of each pixel point in the dark channel atlas of the high-definition remote sensing data atlas of the detection area through the fog; According to the dark channel prior defogging algorithm, establish an optimization model for the high-definition remote sensing data atlas of the detection area to remove the influence of haze on the quality of remote sensing images; The expression of the dark channel prior defogging algorithm is: In the formula, is the pixel value after Gaussian filtering at the th row and th column of the observed hazy image, is the pixel value at the th row and th column of the fog-free image to be restored, is the light transmittance through the fog, is the atmospheric light value, is the atmospheric scattering coefficient, is the depth of field of the scene light.

[0021] It can be explained that the dark channel prior defogging algorithm is a classic and effective image defogging algorithm in the field of computer vision. This algorithm points out that in most natural images, there is at least one color channel whose pixel values are very close to zero in at least one local window. This channel, called the "dark channel", can be used to estimate the global atmospheric light in the image, and by analyzing the relationship between the distance between pixel points and the atmospheric light intensity, the haze information in the image can be inferred, and finally this information is used to remove the haze in the image. In this scheme, by introducing the dark channel prior defogging algorithm, the influence of the decline in the quality of remote sensing images caused by haze or cloudy weather is reduced. By establishing an optimization model for the high-definition remote sensing data atlas of the detection area, the influence of haze is removed, thereby improving the quality of remote sensing images and facilitating subsequent analysis.

[0022] Referring to Figure 3 as shown, the obtaining of the segmented remote sensing image, calculating the blurriness of each part of the remote sensing image, and judging the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image specifically include: According to the result of image segmentation of remotely sensed images collected in real time according to the positions of marks and numbers, extract the optimized pixel values of the real-time remotely sensed images after segmentation; According to the segmentation result of the target standard atlas, extract the optimized pixel values of the standard remotely sensed images after segmentation; Through the normalization formula, perform dimension elimination optimization processing on the optimized pixel values of the real-time remotely sensed images and the optimized pixel values of the standard remotely sensed images; According to the optimized pixel values of the real-time remotely sensed images and the optimized pixel values of the standard remotely sensed images after processing, calculate the blurriness of each part of the remotely sensed images; According to the blurriness of each part of the remotely sensed images, set a threshold value for judging the quality of each part of the remotely sensed images; Judge whether the blurriness of the remotely sensed image is greater than the threshold value. If so, it means that the quality of the current part of the remotely sensed image is poor and does not meet the expectation. If not, it means that the quality of the current part of the remotely sensed image meets the expectation; The blurriness expression of each part of the remotely sensed images is as follows: In the formula, is the blurriness of each part of the remotely sensed images, is the correlation factor, is the optimized pixel value of the rd row and th column of the real-time remotely sensed image after segmentation, is the optimized pixel value of the rd row and th column of the standard remotely sensed image after segmentation, is the number of optimized pixel values of the real-time remotely sensed image after segmentation, is the number of images with the same position pixel values in the target standard atlas.

[0023] It can be explained that when evaluating the quality of remotely sensed images, it is necessary to consider the image quality of each segmentation area of the remotely sensed image quality for subsequent reconstruction. This solution determines the blurriness of each part of the remotely sensed image by establishing the blurriness of each part of the remotely sensed image and analyzing the relationship between the optimized pixel values of the real-time remotely sensed image after segmentation and the optimized pixel values of the standard remotely sensed image after segmentation. Among them, the greater the blurriness, the worse the image quality of the segmentation area and the greater the possibility of replacement.

[0024] Referring to Figure 4 shown, the optimization processing of the remotely sensed image after replacement and reconstruction based on the Laplacian pyramid algorithm specifically includes: According to the reconstruction result of the remotely sensed image by the neural network model, through sampling and processing the result image, separate the largest image and the smallest image, and construct an image pyramid model; According to the image pyramid model, perform Gaussian blur on the image of the next layer, and delete the even rows and columns of the blurred image. Repeat this process successively to obtain the Gaussian pyramid model; According to the Gaussian pyramid model, based on the Laplacian pyramid algorithm, construct an optimized model for remote sensing image reconstruction, and perform optimization processing on the remote sensing image after replacement and reconstruction; The Laplacian pyramid algorithm is as follows: In the formula, is the image data of the th layer of the Laplacian pyramid, is the image data of the th layer of the Gaussian pyramid, is to map the pixel value at the position in the Gaussian pyramid image to the target image , that is, the process of upsampling, is the convolution symbol, is 's Gaussian kernel.

[0025] It can be explained that by using a neural network model to replace and reconstruct remote sensing images, according to the color and contour features of the area to be detected, the optimal spliceable images are selected for splicing and reconstruction. However, it is difficult for the neural network model to effectively complete the perfect fusion of remote sensing images, and problems such as boundary blurring or dislocation may occur during image processing, resulting in disharmony in the overall remote sensing image. In this solution, by introducing the Laplacian pyramid algorithm, its characteristics of multi-scale decomposition and reconstruction enable it to better fuse image information from different modalities. Through the algorithm, the images of each segmented area can be effectively fused, ensuring the overall quality and smoothness of the remote sensing image, thereby further improving the quality of the remote sensing image.

[0026] Furthermore, based on the same inventive concept as the above multi-scale image segmentation and reconstruction method applied to remote sensing images, this solution proposes a multi-scale image segmentation and reconstruction system applied to remote sensing images, including: A data processing module, which is used to obtain high-definition remote sensing images of the area to be detected based on a remote sensing image database and establish a high-definition remote sensing data atlas of the detection area; according to the type of event in the area to be detected, extract the color and contour features of the area to be detected and establish a standard feature map data set; A haze removal and optimization module, which is used to remove the influence of haze on the quality of remote sensing images based on the dark channel prior dehazing algorithm according to the high-definition remote sensing data atlas of the detection area; An image segmentation module, which is used to segment the collected remote sensing images based on the color and contour features of the area to be detected according to the high-definition remote sensing data atlas of the detection area; An image reconstruction and optimization module, which is used to obtain the segmented remote sensing image, calculate the blurriness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; according to the evaluation results of the quality of each part of the remote sensing image, based on a neural network model, replace and reconstruct the parts with poor quality of the remote sensing image; based on the Laplacian pyramid algorithm, optimize the processed remote sensing image after replacement and reconstruction; The data processing module includes: A high-definition remote sensing atlas unit, which is used to obtain high-definition remote sensing images of the area to be detected based on a remote sensing image database and establish a high-definition remote sensing data atlas for the detection area; A feature extraction unit, which is used to extract the color and contour features of the area to be detected according to the type of event in the area to be detected and establish a standard feature map data set; The image reconstruction and optimization module includes: A blurriness unit, which is used to obtain the segmented remote sensing image, calculate the blurriness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; A quality evaluation unit, which is used to replace and reconstruct the parts with poor quality of the remote sensing image based on a neural network model according to the evaluation results of the quality of each part of the remote sensing image; A reconstruction and optimization unit, which is used to optimize the processed remote sensing image after replacement and reconstruction based on the Laplacian pyramid algorithm.

[0027] In summary, the advantages of the present invention are as follows: effectively fuse the pictures of each segmented area, thereby effectively improving the output quality of the remote sensing image, minimizing the influence of haze and partial area blurring or occlusion, and ensuring the reliability and accuracy of the remote sensing image.

[0028] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection required by the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-scale image segmentation and reconstruction method for remote sensing images, characterized in that: include: Obtain high-definition remote sensing images of the area to be inspected based on the remote sensing image database, and establish a high-definition remote sensing data atlas of the inspection area; According to the type of events in the area to be detected, the color and contour features of the area to be detected are extracted, and a standard feature map data set is established; According to the high-definition remote sensing data atlas of the detection area, the influence of haze on the quality of remote sensing images is removed based on the dark channel prior defogging algorithm; According to the high-definition remote sensing data atlas of the detection area, the collected remote sensing images are segmented based on the color and contour features of the area to be detected; Obtain the segmented remote sensing image, calculate the fuzziness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the fuzziness of each part of the remote sensing image; According to the evaluation results of the quality of each part of the remote sensing image, the poor quality parts of the remote sensing image are replaced and reconstructed based on the neural network model; Based on the Laplace pyramid algorithm, the remote sensing image after replacement and reconstruction is optimized.

2. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 1, characterized in that: The method of acquiring a high-definition remote sensing image of the area to be detected based on a remote sensing image database and establishing a high-definition remote sensing data atlas of the detection area specifically includes: Based on the remote sensing image database, a high-definition remote sensing image of the area to be detected is obtained, wherein the high-definition remote sensing image of the area to be detected mainly includes: all-round image information of the area to be detected and color changes of four seasons; Organize, classify and number the high-definition remote sensing images of the inspection area, and establish a high-definition remote sensing data atlas of the inspection area.

3. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 2, characterized in that: The method of extracting the color and contour features of the area to be detected according to the type of the event in the area to be detected and establishing a standard feature map data set specifically includes: Extracting color and contour features of the area to be detected according to the type of events in the area to be detected, wherein the event types include: buildings, plants, rocks, rivers, crops, terrain and man-made objects; According to the color and contour features of the area to be detected, based on the high-definition remote sensing data atlas of the detection area, the color and contour feature images of the area to be detected that meet the standards are screened out to establish a standard feature map dataset.

4. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 3, characterized in that: The above method, based on the high-definition remote sensing data atlas of the detection area and based on the dark channel prior defogging algorithm, removes the influence of haze on the quality of remote sensing images, specifically including: According to the high-definition remote sensing data atlas of the detection area, gray-scale processing is performed on the high-definition remote sensing data atlas of the detection area to obtain a gray-scaled high-definition remote sensing data atlas of the detection area; According to the grayscaled high-definition remote sensing data atlas of the detection area, the quality of the high-definition remote sensing data atlas of the detection area is optimized based on the Gaussian filtering algorithm; According to the optimized high-definition remote sensing data atlas of the detection area, a dark channel atlas of the optimized high-definition remote sensing data atlas of the detection area is obtained; The dark channel image is a grayscale image of the same size as the original image obtained by calculating the minimum value of the three RGB color channels in the local area of ​​each pixel in the image. This grayscale image is the dark channel image. According to the optimized dark channel atlas of the high-definition remote sensing data atlas of the detection area, the atmospheric light value of the dark channel prior defogging algorithm is obtained; According to the atmospheric scattering coefficient and the depth of field of the scene light, the transmission rate of light through fog at each pixel in the dark channel atlas of the high-definition remote sensing data atlas of the detection area is obtained; Based on the dark channel prior defogging algorithm, an optimization model for high-definition remote sensing data atlas of the detection area is established to remove the impact of haze on the quality of remote sensing images; The dark channel priori defogging algorithm expression is: In the formula, The foggy image is observed in Line The pixel values ​​after Gaussian filtering, is the haze-free image to be restored in Line The pixel value of the column, is the transmission rate of light through fog, is the atmospheric light value, is the atmospheric scattering coefficient, is the depth of field of the scene light.

5. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 4, characterized in that: The segmentation of the collected remote sensing image based on the color and contour features of the area to be detected according to the high-definition remote sensing data atlas of the detection area specifically includes: According to the high-definition remote sensing data atlas of the detection area, different positions in the high-definition remote sensing data atlas of the detection area are marked and numbered according to the color and contour characteristics of the area to be detected; According to the high-definition remote sensing data atlas of the marked and numbered detection area, the remote sensing images collected in real time are segmented according to the marked and numbered positions, and the color and contour features of each segmented area are confirmed to be the most obvious; According to the standard feature map dataset, the atlas after the standard feature map dataset is segmented is set as the target standard atlas.

6. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 5, characterized in that: The step of obtaining the segmented remote sensing image, calculating the blurriness of each part of the remote sensing image, and judging the quality of each part of the remote sensing image by the blurriness of each part of the remote sensing image specifically includes: According to the result of image segmentation of the remote sensing image collected in real time according to the positions of the marks and numbers, the pixel values ​​of the optimized real-time remote sensing image after segmentation are extracted; According to the segmentation results of the target standard atlas, the optimized pixel values ​​of the segmented standard remote sensing image are extracted; Through the normalization formula, the pixel values ​​after optimization of the real-time remote sensing image and the pixel values ​​after optimization of the standard remote sensing image are subjected to dimension elimination optimization processing; Calculate the blurriness of each part of the remote sensing image according to the optimized pixel value of the processed real-time remote sensing image and the optimized pixel value of the standard remote sensing image; According to the blurriness of each part of the remote sensing image, a threshold is set to judge the quality of each part of the remote sensing image; Determine whether the blurriness of the remote sensing image is greater than a threshold. If so, it means that the quality of the current part of the remote sensing image is poor and does not meet expectations. If not, it means that the quality of the current part of the remote sensing image meets expectations. The fuzziness expression of each part of the remote sensing image is: In the formula, is the blur of each part of the remote sensing image, is the correlation factor, After segmentation, the real-time remote sensing image is optimized. Line The pixel value of the column, After segmentation, the standard remote sensing image is optimized. Line The pixel value of the column, is the number of pixel values ​​optimized for the real-time remote sensing image after segmentation, is the number of images with the same pixel value in the target standard atlas.

7. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 6, characterized in that: The replacing and reconstructing the poor quality parts of the remote sensing image based on the neural network model according to the evaluation results of the quality of each part of the remote sensing image specifically includes: According to the target standard atlas, remote sensing images of the area to be detected in different states are obtained, and a target atlas and a training atlas are established respectively; The training atlas and the evaluation results of the quality of each part of the remote sensing image are used as the input matrix, and the target atlas is used as the verification output matrix to construct the input and output matrix of the neural network; Based on the neural network model, the input and output matrix data are trained and learned, and the quality of each part of the newly received real-time remote sensing image is evaluated through the model training results; Determine whether the quality of each part of the newly received real-time remote sensing image meets the requirements. If so, it means that the quality of this part of the image meets expectations, and the original image of this part is output. If not, it means that the quality of this part of the image does not meet expectations, and the image of this part is replaced and reconstructed.

8. The multi-scale image segmentation and reconstruction method for remote sensing images according to claim 7, characterized in that: The optimization process of the replaced and reconstructed remote sensing image based on the Laplace pyramid algorithm specifically includes: According to the reconstruction results of the remote sensing image by the neural network model, the maximum image and the minimum image are separated by sampling and processing the result image, and the image pyramid model is constructed; According to the image pyramid model, Gaussian blur is performed on the image of the next layer, and the even rows and columns of the blurred image are deleted, and the process is repeated to obtain a Gaussian pyramid model; According to the Gaussian pyramid model and based on the Laplace pyramid algorithm, a remote sensing image reconstruction optimization model is constructed to optimize the remote sensing image after replacement and reconstruction. The Laplace pyramid algorithm is: In the formula, The Laplace pyramid The image data of the layer, Gaussian pyramid The image data of the layer, To convert the first The pixel values ​​are mapped to the target image The position of the upward sampling process, is the convolution symbol, for Gaussian kernel.

9. A multi-scale image segmentation and reconstruction system for remote sensing images, characterized in that: The method for implementing the multi-scale image segmentation and reconstruction method applied to remote sensing images as claimed in any one of claims 1 to 8 comprises: A data processing module, the data processing module is used to obtain high-definition remote sensing images of the area to be detected based on the remote sensing image database, and establish a high-definition remote sensing data atlas of the detection area; according to the type of events in the area to be detected, extract the color and contour features of the area to be detected, and establish a standard feature map data set; A defogging optimization module, which is used to remove the influence of haze on the quality of remote sensing images based on a dark channel prior defogging algorithm according to a high-definition remote sensing data atlas of the detection area; An image segmentation module, wherein the image segmentation module is used to segment the collected remote sensing image based on the color and contour features of the area to be detected according to the high-definition remote sensing data atlas of the detection area; The image reconstruction and optimization module is used to obtain the segmented remote sensing image, calculate the blurriness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image through the blurriness of each part of the remote sensing image; according to the judgment results of the quality of each part of the remote sensing image, based on the neural network model, replace and reconstruct the parts of the remote sensing image with poor quality; based on the Laplace pyramid algorithm, optimize the remote sensing image after replacement and reconstruction.

10. The multi-scale image segmentation and reconstruction system for remote sensing images according to claim 9, characterized in that: The data processing module specifically includes: A high-definition remote sensing atlas unit, which is used to obtain high-definition remote sensing images of the area to be detected based on a remote sensing image database, and to establish a high-definition remote sensing data atlas of the detection area; A feature extraction unit, the feature extraction unit is used to extract the color and contour features of the area to be detected according to the type of the event in the area to be detected, and to establish a standard feature map data set; The image reconstruction and optimization module specifically includes: A fuzziness unit, the fuzziness unit is used to obtain the segmented remote sensing image, calculate the fuzziness of each part of the remote sensing image, and judge the quality of each part of the remote sensing image according to the fuzziness of each part of the remote sensing image; A quality evaluation unit, wherein the quality evaluation unit is used to replace and reconstruct the poor quality parts of the remote sensing image based on the neural network model according to the evaluation results of the quality of each part of the remote sensing image; The reconstruction optimization unit is used to optimize the remote sensing image after replacement and reconstruction based on the Laplace pyramid algorithm.

Images