Remote Sensing Data Processing Method for Garden Design

Abstract

The present invention belongs to the field of image processing, and specifically relates to a remote sensing survey data processing method for garden design, the method comprising: acquiring multiple remote sensing images containing a target garden area in different spectral bands; constructing a texture sequence corresponding to each feature point in each remote sensing image; determining multiple co-location evaluation values ​​corresponding to each feature point; determining the image texture preservation degree corresponding to each feature point based on the multiple co-location evaluation values ​​corresponding to each feature point; uniformly dividing each remote sensing image into multiple image regions, determining a fusion weight corresponding to each image region in each remote sensing image; performing weighted fusion on the multiple remote sensing images based on the fusion weight corresponding to each image region, and post-processing the weighted fused remote sensing images to obtain remote sensing survey data processing results. The solution provided by the present invention improves the precision of the fusion weights and the accuracy and referenceability of the remote sensing survey data processing results.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a remote sensing survey data processing method for garden design. Background Art

[0002] Garden design is a comprehensive process that integrates art, ecology, engineering, and the humanities. In practical applications, garden design is often based on remote sensing survey data. Traditional remote sensing data is mostly obtained by surveying garden areas from high altitudes using satellites or drones. Due to the influence of various atmospheric conditions, remote sensing survey data contains a lot of interference information and some data is not useful for garden design. Therefore, remote sensing survey data must be processed before it can be used in garden design.

[0003] In related technologies, when processing remote sensing survey data, it is necessary to perform image fusion and cropping on multi-source remote sensing data to obtain remote sensing survey data corresponding to the garden area. Because the remote sensing images collected come from different sources and have different resolutions, accurate positioning is difficult during image fusion. Therefore, it is necessary to first extract corresponding feature information from the multiple remote sensing images, and then comprehensively analyze and fuse this feature information. Traditionally, a weighted fusion method is used to fuse multi-source remote sensing images. The fusion weights are constructed based on the weighted average method, that is, the corresponding fusion weights are constructed based on the similarity between the relevant texture features extracted from the multiple remote sensing images and the sum of the multiple texture features. Because multiple remote sensing images come from different spectral bands and spatial resolutions, and different spectral bands react differently to the same object, this leads to differences in the extracted feature information. Specifically, the texture features of two different objects combined in different spectral bands differ. This causes some non-essential features to be included in the overall texture feature summation, resulting in a reduced weight for the main garden vegetation and terrain information when calculating the fusion weights. Consequently, the fused remote sensing images are unable to provide effective data support for garden design.

[0004] It is not difficult to find that the traditional remote sensing survey data processing solution has the problem of low accuracy and insufficient reference value of data processing results due to improper determination of fusion weights. Summary of the Invention

[0005] In order to solve the technical problems of low accuracy and insufficient reference value of data processing results in traditional remote sensing survey data processing solutions, the purpose of the present invention is to provide a remote sensing survey data processing method for garden design. The technical solution adopted is as follows:

[0006] A remote sensing survey data processing method for garden design, the method comprising:

[0007] Acquire multiple remote sensing images containing the target garden area in different spectral bands;

[0008] Construct the texture sequence corresponding to each feature point in each remote sensing image;

[0009] Determining, based on the texture sequence, a collocation evaluation value between each feature point in each remote sensing image and a corresponding reference feature point in other remote sensing images, and obtaining a plurality of collocation evaluation values ​​corresponding to each feature point;

[0010] Determine the image texture preservation degree corresponding to each feature point based on multiple co-location evaluation values ​​corresponding to each feature point;

[0011] Each remote sensing image is uniformly divided into multiple image regions, and the fusion weight corresponding to each image region in each remote sensing image is determined according to the image texture preservation degree corresponding to each feature point in each image region;

[0012] According to the fusion weights corresponding to each image area, multiple remote sensing images are weightedly fused by region, and the weighted fused remote sensing images are post-processed to obtain the remote sensing survey data processing results of the target garden area.

[0013] According to a remote sensing survey data processing method for garden design provided by the present invention, a texture sequence corresponding to each feature point in each remote sensing image is constructed, comprising:

[0014] For each remote sensing image, obtain the gradient direction corresponding to each feature point, the edge points of the neighborhood sub-region where each feature point is located, and the gradient direction of the edge points;

[0015] Take the lower left corner edge point of the neighborhood sub-region where each feature point is located as the starting point, and use the gradient direction corresponding to the lower left corner edge point as the starting element value;

[0016] Starting from the starting point, traverse the gradient directions corresponding to the edge points in sequence along a preset direction to obtain a plurality of sequentially arranged intermediate element values; wherein the preset direction is first rightward to the lower right corner, and then upward to the upper right corner;

[0017] Until the gradient direction corresponding to the last edge point in the upper right corner is traversed, the end element value is obtained;

[0018] According to the starting element value, a plurality of sequentially arranged intermediate element values ​​and an end element value, a texture sequence corresponding to each feature point in each remote sensing image is constructed.

[0019] According to a remote sensing survey data processing method for garden design provided by the present invention, based on the texture sequence, a collocation evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in each other remote sensing image is determined, and multiple collocation evaluation values ​​corresponding to each feature point are obtained, including:

[0020] Determining, based on the texture sequence, a texture matching degree between each feature point in each remote sensing image and each feature point to be matched in other remote sensing images;

[0021] The feature points to be matched whose texture matching degree is greater than a preset matching threshold are used as candidate feature points;

[0022] Determining the positioning isotexture degree between each feature point in each remote sensing image and feature points in other remote sensing images based on the texture matching degree and the candidate feature points;

[0023] According to the positioning isotexture degree between each feature point and feature points in other remote sensing images, the texture consistency degree of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood sub-region and the reference degree of the neighborhood sub-region are determined respectively;

[0024] Based on the texture consistency and the reference degree of the neighborhood sub-region, calculating the co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images;

[0025] The candidate feature point with the highest co-location evaluation value in each other remote sensing image is used as the reference feature point, and the co-location evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in other remote sensing images is used as the multiple co-location evaluation values ​​corresponding to each feature point.

[0026] According to a remote sensing survey data processing method for garden design provided by the present invention, based on the texture matching degree and the candidate feature points, the positioning isotexture degree between each feature point in each remote sensing image and the feature points in other remote sensing images is determined, including:

[0027] The target feature point with the largest texture matching degree in each remote sensing image is used as the first transition point, and the candidate feature point corresponding to the target feature point is used as the second transition point;

[0028] Determining, based on the first transition point and the second transition point, a positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images, thereby obtaining a plurality of positioning matching degrees corresponding to each feature point in each remote sensing image;

[0029] The maximum positioning matching degree among multiple positioning matching degrees corresponding to each feature point is extracted, and the positioning isomorphism between each feature point in each remote sensing image and the feature points in other remote sensing images is determined based on the maximum positioning matching degree.

[0030] According to a remote sensing survey data processing method for garden design provided by the present invention, based on the first transition point and the second transition point, respectively determining the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images, the method includes:

[0031] Determining a first distance value between each feature point in the same remote sensing image and the first transition point, and determining a second distance value between each feature point in the same remote sensing image and the second transition point;

[0032] Calculating an absolute value of a difference between the first distance value and the second distance value;

[0033] Inputting the absolute value of the difference into an inverse proportional normalization function to obtain an inverse proportional normalized value;

[0034] Determining a texture matching degree between a feature point in the remote sensing image where the first transition point is located and a feature point in the remote sensing image where the second transition point is located;

[0035] The inversely proportional normalized value is multiplied by the texture matching degree between the feature point in the remote sensing image where the first transition point is located and the feature point in the remote sensing image where the second transition point is located, to obtain the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images.

[0036] According to a remote sensing survey data processing method for garden design provided by the present invention, based on the positioning texture consistency between each feature point and feature points in other remote sensing images, the texture consistency of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood sub-region and the reference degree of the neighborhood sub-region are determined, including:

[0037] Determine the neighborhood feature points and the total number of neighborhood feature points contained in the neighborhood sub-region, and determine the positioning isotexture between each feature point in the remote sensing image and each neighborhood feature point in the same neighborhood sub-region in the remote sensing image where any candidate feature point is located;

[0038] Sum the localization iso-grainness corresponding to all neighborhood feature points to obtain the sum value of iso-grainness;

[0039] Determining a first intermediate value according to the sum of the same-grain degrees;

[0040] Perform a quadratic operation on the positioning isotexture corresponding to each neighborhood feature point to obtain the isotexture product value of each neighborhood feature point;

[0041] Sum the product values ​​of the same texture of all neighborhood feature points to obtain the second intermediate value;

[0042] Calculating, based on the first intermediate value and the second intermediate value, the texture consistency of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood subregion;

[0043] The reference degree of the neighborhood sub-region is obtained by dividing the sum of the isotexture values ​​by the total number of the neighborhood feature points.

[0044] According to a remote sensing survey data processing method for garden design provided by the present invention, based on the texture consistency and the reference degree of the neighborhood sub-region, a co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images is calculated, including:

[0045] Determine all neighborhood sub-regions contained in the neighborhood range corresponding to each feature point in each remote sensing image;

[0046] Multiply the texture consistency and reference degree corresponding to each neighborhood sub-region to obtain the parameter product value corresponding to each neighborhood sub-region;

[0047] Sum the parameter product values ​​of all neighborhood sub-regions to obtain the product sum value;

[0048] Sum the reference degrees corresponding to all neighborhood sub-regions to obtain the reference degree sum value;

[0049] The product sum value is divided by the reference degree sum value to obtain a co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature point in other remote sensing images.

[0050] According to a remote sensing survey data processing method for garden design provided by the present invention, the image texture preservation degree corresponding to each feature point is determined based on multiple co-location evaluation values ​​corresponding to each feature point, including:

[0051] The multiple co-evaluation values ​​corresponding to each feature point are averaged to obtain the co-evaluation mean corresponding to each feature point;

[0052] Calculate the standard deviation of multiple parity evaluation values ​​corresponding to each feature point to obtain the parity evaluation standard deviation value corresponding to each feature point;

[0053] The image texture preservation degree corresponding to each feature point is calculated based on multiple co-evaluation values, co-evaluation mean values, and co-evaluation standard deviation values ​​corresponding to each feature point.

[0054] According to a remote sensing survey data processing method for garden design provided by the present invention, a fusion weight corresponding to each image area in each remote sensing image is determined based on the image texture preservation degree corresponding to each feature point in each image area, including:

[0055] The image texture preservation degrees of all feature points in each image area of ​​each remote sensing image are averaged, and the fusion weights corresponding to each image area in each remote sensing image are calculated.

[0056] According to a remote sensing survey data processing method for garden design provided by the present invention, post-processing is performed on the weighted fused remote sensing image to obtain the remote sensing survey data processing result of the target garden area, including:

[0057] The weighted fused remote sensing image is cropped to obtain the regional image corresponding to the target garden area;

[0058] Divide the regional image into multiple image blocks, and determine the gray level co-occurrence matrix of each image block respectively;

[0059] Determine the similarity of gray level co-occurrence matrices between adjacent image blocks, and merge the image blocks based on the similarity of the gray level co-occurrence matrices between adjacent image blocks to obtain a block-merged image;

[0060] Clustering processing is performed on the block merged image to obtain remote sensing survey data processing results of the target garden area.

[0061] The present invention has the following beneficial effects:

[0062] By constructing a texture sequence corresponding to each feature point in each remote sensing image, the co-location evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in other remote sensing images is determined based on the texture sequence, and multiple co-location evaluation values ​​corresponding to each feature point are obtained. Then, the image texture preservation degree corresponding to each feature point is determined based on the multiple co-location evaluation values ​​corresponding to each feature point. Each remote sensing image is uniformly divided into multiple image regions. Then, based on the image texture preservation degree corresponding to each feature point in each image region, the corresponding fusion weight of each image region in each remote sensing image is determined. Finally, based on the fusion weight corresponding to each image region, multiple remote sensing images are weightedly fused by region, and the weighted fused remote sensing images are post-processed to obtain the remote sensing survey data processing results of the target garden area. Since the co-location evaluation values ​​between feature points in different remote sensing images and the image texture preservation degree of each image region are introduced in the process of determining the fusion weight, the corresponding fusion weight can be constructed according to the degree of texture information preservation in each image region, thereby improving the accuracy of the fusion weight and thus improving the accuracy and reference of the remote sensing survey data processing results. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0064] Figure 1 A flow chart of a remote sensing survey data processing method for garden design provided by one embodiment of the present invention;

[0065] Figure 2 A remote sensing image of a garden area;

[0066] Figure 3 It is the remote sensing image after correction and grayscale processing;

[0067] Figure 4 is the remote sensing image after edge detection;

[0068] Figure 5 is a remote sensing image marked with corner detection results;

[0069] Figure 6 It is a neighborhood structure diagram centered on a certain feature point;

[0070] Figure 7 This is a system structure diagram of a remote sensing survey data processing system for garden design provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0071] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail a remote sensing survey data processing method for landscape design, including its specific implementation, structure, features, and effectiveness. In the following description, references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0072] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0073] The specific scheme of the remote sensing survey data processing method for garden design provided by the present invention is described in detail below with reference to the accompanying drawings.

[0074] See also Figure 1 , which shows a method flow chart of a remote sensing survey data processing method for garden design provided by an embodiment of the present invention, such as Figure 1 As shown, the remote sensing survey data processing method for garden design specifically includes:

[0075] Step 110: Acquire multiple remote sensing images containing the target garden area in different spectral bands.

[0076] In practical applications, multiple remote sensing images of the target garden area in various spectral bands can be obtained on remote sensing image platforms such as the geospatial data cloud platform and the national remote sensing data application service platform according to the strip number and row number corresponding to the latitude and longitude of the target garden area. Figure 2 The remote sensing image of a certain garden area is shown as an example.

[0077] In some embodiments, remote sensing data of near-infrared bands, mid-infrared bands, and hyperspectral bands can be obtained from multiple remote sensing images. These remote sensing data can be used to monitor vegetation photosynthetic activity, vegetation moisture content, soil conditions, and other related conditions.

[0078] Step 120: Construct a texture sequence corresponding to each feature point in each remote sensing image.

[0079] It is understandable that the texture sequence can characterize the texture trend at the location of each feature point, that is, the texture sequence can present the plant texture characteristics at the location of each feature point in the form of a data sequence.

[0080] Step 130: Determine the collocation evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in other remote sensing images based on the texture sequence, and obtain multiple collocation evaluation values ​​corresponding to each feature point.

[0081] It can be understood that the co-location evaluation value can represent the possibility that each feature point and the corresponding reference feature point belong to the same location point in the target garden area. The higher the co-location evaluation value, the greater the possibility that the feature point and the corresponding reference feature point belong to the same location point in the target garden area.

[0082] Step 140 : Determine the image texture preservation degree corresponding to each feature point based on the multiple co-location evaluation values ​​corresponding to each feature point.

[0083] In this embodiment, the image texture retention degree can represent the degree to which the texture information on the image where the feature point is located needs to be retained from the perspective of each feature point. The higher the image texture retention degree, the better the image where the feature point is located matches other images, and more texture information needs to be retained.

[0084] Step 150: Divide each remote sensing image into multiple image regions, and determine the fusion weight corresponding to each image region in each remote sensing image based on the image texture preservation degree corresponding to each feature point in each image region.

[0085] It should be noted that this embodiment partitions each remote sensing image in advance to obtain multiple regions, and subsequently determines the fusion weight of each image region based on the image texture retention, thereby improving the reference and accuracy of the fusion weight.

[0086] Step 160: performing weighted fusion on the multiple remote sensing images by region based on the fusion weights corresponding to the respective image regions, and performing post-processing on the weighted fused remote sensing images to obtain the remote sensing survey data processing results of the target garden area.

[0087] In practical applications, using the fusion weights of each image region to perform weighted fusion of multiple remote sensing images in different regions can improve the weighted fusion effect of remote sensing images, and thus improve the reliability of remote sensing survey data processing results.

[0088] Understandably, since garden design primarily operates on existing terrain, the design process requires an examination of the target garden area's topographical factors and vegetation. Currently, after extracting features using algorithms such as regional contours, line intersections, the Canny operator, or the Sobel operator, these features are then compared uniformly during feature fusion. However, due to the varying absorption responses of the same object to different light waves, and the varying resolutions of different remote sensing images, the extracted feature points and feature information may differ.

[0089] However, for garden design, the main focus is on terrain information and the information of the many plants planted on it. Since the shapes of the crowns, leaves, etc. produced by different plants after planting are different, this information is roughly the same in different remote sensing images. At the same resolution and the same angle, they can basically overlap. However, at different resolutions, although the details are different, the texture trends are basically the same. Similarly, at different angles, the continuity of many inflection points is consistent. Since garden design pays more attention to the types of these plants and related information such as terrain, it is necessary to retain more texture information based on the matching of plant texture information, and then assign more fusion weights to remote sensing images containing more texture features, so as to adjust the referenceability and fusion weights when remote sensing images are fused. Accordingly, this embodiment improves the process of determining the fusion weight based on texture information.

[0090] In one embodiment, constructing a texture sequence corresponding to each feature point in each remote sensing image specifically includes:

[0091] The first step is to obtain the gradient direction corresponding to each feature point, the edge points of the neighborhood sub-region where each feature point is located, and the gradient direction of the edge points for each remote sensing image.

[0092] In the second step, the lower left corner edge point in the neighborhood sub-region where each feature point is located is taken as the starting point, and the gradient direction corresponding to the lower left corner edge point is used as the starting element value.

[0093] The third step is to start from the starting point and traverse the gradient directions corresponding to the edge points passed in sequence along the preset direction to obtain multiple sequentially arranged intermediate element values; among which the preset direction is first to the right to the lower right corner, and then upward to the upper right corner.

[0094] The fourth step is to traverse the gradient direction corresponding to the last edge point in the upper right corner and obtain the end element value.

[0095] The fifth step is to construct a texture sequence corresponding to each feature point in each remote sensing image based on the starting element value, multiple sequentially arranged intermediate element values, and the end element value.

[0096] In practical applications, before obtaining the relevant information of each feature point, each remote sensing image needs to be radiometrically calibrated and atmospherically corrected, and then the corrected remote sensing image needs to be grayscale processed. Figure 3 Afterwards, edge detection can be performed using the Canny operator or the Sobel operator to obtain the corresponding remote sensing image after edge detection and the corresponding gradient information. The remote sensing image after edge detection is as follows: Figure 4 As shown. Subsequently, the remote sensing image can be corner detected to obtain the characteristic corner points in the remote sensing image, that is, the feature points. In the corner detection step, the initial threshold can be set to 0.04, the number of layers in the scale space can be set to 3, and the corner detection results can be marked in the remote sensing image. For details, please refer to Figure 5 .

[0097] Since the distribution of terrain and plants in the target garden area is naturally formed without human intervention, and there is relatively no special pattern for the distribution of naturally formed plants, and because the growth structures of different plant species are different, the trends of various plants and terrains detected in remote sensing images are different. Considering that the feature points in each remote sensing image obtained by corner detection are the intersections or related inflection points between the textures of plants, terrain, etc., this leads to the trend of the texture near the feature points being relatively complex. Therefore, the texture characteristics of the feature point can be described based on the trend of the texture where the feature point is located, that is, the gradient direction on the edge near it. This embodiment effectively characterizes the texture information of the vegetation and terrain in the area where each feature point is located through a texture sequence.

[0098] In this embodiment, the preset direction is set to start from the lower left corner of the remote sensing image, first to the right and then to the upper right corner. The edge point of the lower left corner of the texture where each feature point is located is used as the starting point of the texture sequence, and its gradient direction is used as the starting element value in the sequence. The gradient direction of each edge point on the texture is gradually filled into the texture sequence along the preset direction until the last edge point in the upper right corner of the texture is obtained.

[0099] In one embodiment, based on the texture sequence, a collocation evaluation value between each feature point in each remote sensing image and a corresponding reference feature point in each other remote sensing image is determined, and multiple collocation evaluation values ​​corresponding to each feature point are obtained, specifically including:

[0100] The first step is to determine the texture matching degree between each feature point in each remote sensing image and each feature point to be matched in other remote sensing images based on the texture sequence.

[0101] Since the contrast and brightness of vegetation and terrain are different in remote sensing images of different spectral bands, that is, some texture information in some remote sensing images is unclear, which makes the feature points and textures detected by different remote sensing images for the same texture area intermittently the same. Therefore, it is necessary to use a dynamic programming strategy to compare the similarities and differences in the texture trends of the same texture area in two remote sensing images, that is, the similarities and differences in the feature directions between the corresponding edge points of the texture area. This information can be represented by the texture matching degree. Matching feature points with those in another remote sensing image As an example, the texture matching between can be expressed as follows:

[0102] (1)

[0103] in, Represents target feature points and feature points to be matched The texture matching between Represents the target feature points in a remote sensing image The texture sequence at the texture; Represents the feature points to be matched in another remote sensing image The texture sequence corresponding to the texture; It represents the DTW (Dynamic Time Warping) distance between two texture sequences. The larger the DTW distance, the greater the texture difference between the target feature point and the feature point to be matched, and the lower the texture matching degree.

[0104] In the second step, the feature points to be matched whose texture matching degree is greater than the preset matching threshold are used as candidate feature points.

[0105] Taking into account that when performing matching fusion, or when matching the feature point positions and related feature information of many remote sensing images, the spectral bands of different remote sensing images are different, resulting in differences in the contrast of the same plant or terrain in different remote sensing images, and the gradient direction is difficult to detect or inaccurate in some textures, but the texture trends of these terrains and vegetation themselves have a certain degree of extensibility.

[0106] Since for a certain feature point, the corresponding feature point to be matched in other remote sensing images cannot be confirmed, that is, it is difficult to determine whether the two feature points are in the same garden location, so it is necessary to first screen out the alternative feature points in other remote sensing images that are most likely to be in the same garden location as the feature point. Considering that multiple remote sensing images are obtained by surveying the same garden area, the positions of other gardens around the same garden location are also fixed. Therefore, when the texture matching degree of two feature points in two remote sensing images (one feature point is extracted from each remote sensing image) is high, and the texture matching degree of the feature points at the corresponding positions in the surrounding area is also high, then it can be considered that the two feature points are feature points in the same garden location and have a matching relationship in the two remote sensing images.

[0107] In this embodiment, a preset matching threshold can be used to screen candidate feature points in other remote sensing images that have similar texture features to feature points in the current remote sensing image. Specifically, all feature points to be matched in the other remote sensing image can be traversed, and feature points to be matched whose texture matching degree with the feature points in the current remote sensing image is greater than or equal to the preset matching threshold can be extracted and recorded as candidate feature points. In practical applications, the preset matching threshold can be set to 0.7.

[0108] In the third step, based on the texture matching degree and the candidate feature points, the positioning texture degree between each feature point in each remote sensing image and the feature points in other remote sensing images is determined.

[0109] It can be understood that the positioning same-grainness can represent the possibility that each feature point belongs to the same garden location as a feature point in other remote sensing images. The higher the positioning same-grainness, the greater the possibility that the feature point belongs to the same garden location as a feature point in other remote sensing images.

[0110] The fourth step is to determine the texture consistency and reference degree of each remote sensing image in the same neighborhood sub-region with any remote sensing image where the candidate feature point is located based on the positioning texture consistency between each feature point and the feature points in other remote sensing images.

[0111] In this embodiment, texture consistency can represent the similarity of texture features in the area surrounding a feature point. A higher texture consistency indicates a higher similarity of texture features in the area surrounding the feature point. The referenceability of a neighborhood sub-region is mainly used to represent the reference value of a neighborhood sub-region of a feature point to the overall area. A higher referenceability indicates a greater reference value of the neighborhood sub-region to the overall area.

[0112] The fifth step is to calculate the co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images based on the texture consistency and the reference degree of the neighborhood sub-region.

[0113] In the sixth step, the candidate feature point with the highest co-location evaluation value in each other remote sensing image is used as the reference feature point, and the co-location evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in other remote sensing images is used as the multiple co-location evaluation values ​​corresponding to each feature point.

[0114] In one embodiment, based on the texture matching degree and the candidate feature points, the positioning isotexture degree between each feature point in each remote sensing image and the feature points in other remote sensing images is determined, specifically including:

[0115] First, the target feature point with the highest texture matching degree in each remote sensing image is taken as the first transition point, and the candidate feature point corresponding to the target feature point is taken as the second transition point.

[0116] Then, based on the first transition point and the second transition point, the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images is determined respectively, and multiple positioning matching degrees corresponding to each feature point in each remote sensing image are obtained.

[0117] In remote sensing images, for the detected feature points, due to the different contrast and clarity in many remote sensing images, the feature points detected in the same plant area or terrain area in many remote sensing images may not be in the same garden location. Therefore, it is necessary to first match the feature points in the current remote sensing image with those in other remote sensing images to determine the positioning matching degree between the feature points in different remote sensing images.

[0118] In a specific implementation, determining the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images based on the first transition point and the second transition point includes:

[0119] In the first step, a first distance value between each feature point and a first transition point in the same remote sensing image is determined, and a second distance value between each feature point and a second transition point in the same remote sensing image is determined.

[0120] The second step is to calculate the absolute value of the difference between the first distance value and the second distance value.

[0121] The third step is to input the absolute value of the difference into the inverse proportional normalization function to obtain the inverse proportional normalized value.

[0122] The fourth step is to determine the texture matching degree between the feature point in the remote sensing image where the first transition point is located and the feature point in the remote sensing image where the second transition point is located.

[0123] In the fifth step, the inverse normalized value is multiplied by the texture matching degree between the feature point in the remote sensing image where the first transition point is located and the feature point in the remote sensing image where the second transition point is located, so as to obtain the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images.

[0124] In this embodiment, the target remote sensing image Feature points in With other remote sensing images Middle feature point As an example, the positioning matching degree between can be expressed as follows:

[0125] (2)

[0126] Where, Representing feature points With feature points The positioning match between Representing feature points With feature points Texture matching between Representing feature points To the candidate feature point The second distance value of the second transition point; Representing feature points To the target feature point The first distance value of the first transition point; represents the inverse normalization function.

[0127] Finally, the maximum positioning matching degree among the multiple positioning matching degrees corresponding to each feature point is extracted, and the positioning isomorphism between each feature point in each remote sensing image and the feature points in other remote sensing images is determined based on the maximum positioning matching degree.

[0128] In practical applications, the first and second transition points determined above are still used as the basis to traverse other remote sensing images. All feature points in the target remote sensing image are calculated and each feature point is compared with the target remote sensing image. Middle feature point The positioning matching degree between them is calculated and the maximum value is obtained, that is, the maximum positioning matching degree. If the maximum positioning matching degree is greater than or equal to the preset matching upper limit threshold, then the feature point is considered Possess remote sensing images of the target Feature points in The same position, and record the feature point as the feature point The same texture point between the two feature points can be taken as the maximum positioning matching degree. On the contrary, if the maximum positioning matching degree is less than the preset matching upper limit threshold, it means that in other remote sensing images There is no such feature point in For isotexture points, the positioning isotexture degree between two feature points can be 0. In some embodiments, the preset matching upper limit threshold can be 0.6.

[0129] In one embodiment, based on the positioning isotexture degree between each feature point and feature points in other remote sensing images, the texture consistency and the reference degree of each remote sensing image in the same neighborhood sub-region as the remote sensing image where any candidate feature point is located are determined, specifically including:

[0130] The first step is to determine the neighborhood feature points and the total number of neighborhood feature points contained in the neighborhood sub-region, and determine the positioning isotexture between each feature point in the remote sensing image and each neighborhood feature point in the same neighborhood sub-region in the remote sensing image where any candidate feature point is located.

[0131] In this embodiment, the target feature points can be used in two remote sensing images. or alternative feature points A circular area with a preset radius is constructed as the center as the neighborhood. In practical applications, the preset radius can be set to 20 pixels in length. The neighborhood is then divided evenly to obtain k neighborhood sub-areas. In some embodiments, the value of k can be 8, that is, 8 neighborhood sub-areas are divided. Figure 6 As an example, a circular neighborhood is constructed with a certain feature point 220 as the center, and eight neighborhood sub-regions 210 are divided out, and a plurality of neighborhood feature points 230 are distributed in the neighborhood sub-regions.

[0132] In the second step, the localization isotope values ​​corresponding to all neighborhood feature points are summed to obtain the summed isotope value.

[0133] The third step is to determine the first intermediate value based on the sum of the same grain.

[0134] The fourth step is to perform a quadratic operation on the positioning isotexture corresponding to each neighborhood feature point to obtain the isotexture product value of each neighborhood feature point.

[0135] The fifth step is to sum the isotexture product values ​​of all neighborhood feature points to obtain the second intermediate value.

[0136] In the sixth step, based on the first intermediate value and the second intermediate value, the texture consistency of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood sub-region is calculated.

[0137] For two remote sensing areas, if the textures of each feature point in the surrounding area of ​​the target feature point and the reference feature point are similar, or most of the textures are similar, then it can be considered that the two feature points are remote sensing surveys of the same garden area. Target remote sensing image and alternative feature points Corresponding to other remote sensing images The texture consistency in the same neighborhood sub-region can be expressed as follows:

[0138] (3)

[0139] Where, Represents target feature points Target remote sensing image and alternative feature points Other remote sensing images The texture consistency of the kth neighborhood sub-region is, Represents target feature points Target remote sensing image Any feature point and the sth neighborhood feature point in the kth neighborhood subregion Positioning homography between them; Represents the total number of neighborhood feature points contained in the kth neighborhood sub-region, y represents a constant that is not 0, Represents target feature points Target remote sensing image Any feature point The weighted sum of all neighborhood feature points in the kth neighborhood subregion, the first of which The reason why it is a weight value here is that in two remote sensing images, the more likely the feature points are to belong to the same area in terms of positioning, the higher their texture consistency and the greater their reference degree. Therefore, the texture consistency of the core position is described by the central trend of the overall positioning texture consistency of a certain neighborhood sub-region.

[0140] In addition, the sum of the same texture values ​​is divided by the total number of neighborhood feature points to obtain the reference degree of the neighborhood sub-region.

[0141] For each neighborhood sub-region, the higher the similarity between the neighborhood feature points and the target feature points, and the higher the texture consistency, the greater the reference value of this neighborhood sub-region to the entire neighborhood. Accordingly, this embodiment characterizes the reference value of a neighborhood sub-region to the entire region through the reference value. Specifically, the reference value of the kth neighborhood sub-region to the entire neighborhood can be expressed as follows:

[0142] (4)

[0143] in, represents the reference degree of the kth neighborhood sub-region, Representing feature points and the sth neighborhood feature point in the kth domain sub-region Positioning homography between them; Indicates the total number of neighborhood feature points contained in the kth neighborhood sub-region.

[0144] In one embodiment, based on texture consistency and the reference degree of the neighborhood sub-region, a co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature point in each other remote sensing image is calculated, specifically including:

[0145] The first step is to determine all neighborhood sub-regions contained in the neighborhood range corresponding to each feature point in each remote sensing image.

[0146] In the second step, the texture consistency and reference degree corresponding to each neighborhood sub-region are multiplied to obtain the parameter product value corresponding to each neighborhood sub-region.

[0147] The third step is to sum the parameter product values ​​of all neighborhood sub-regions to obtain the product sum value.

[0148] The fourth step is to sum up the reference degrees corresponding to all neighborhood sub-regions to obtain the reference degree sum value.

[0149] The fifth step is to divide the sum of the products by the sum of the reference degrees to obtain the co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images.

[0150] In this embodiment, the target feature point and alternative feature points The parity evaluation values ​​between can be expressed as follows:

[0151] (5)

[0152] Where, Represents target feature points and alternative feature points The same evaluation value between Represents the target remote sensing image and alternative feature points Other remote sensing images Texture consistency in the kth neighborhood sub-region; Indicates the reference degree of the kth neighborhood sub-region; Represents the total number of neighborhood feature points contained in the kth neighborhood sub-region, Represents the weighted sum of the texture consistency corresponding to all neighborhood feature points in the kth neighborhood subregion, where Here, the weight value is expressed, thereby describing the co-location evaluation value of the core position by the central tendency of the overall texture consistency of a certain neighborhood sub-region.

[0153] Subsequently, all candidate feature points in each remote sensing image can be traversed, the corresponding co-location evaluation values ​​can be calculated, and the candidate feature point with the largest co-location evaluation value can be recorded as the reference feature point of the target feature point in another remote sensing image. At this point, the reference feature point of the target feature point in each other remote sensing image can be obtained.

[0154] In one embodiment, the image texture preservation degree corresponding to each feature point is determined based on multiple co-location evaluation values ​​corresponding to each feature point, specifically including:

[0155] First, the multiple co-evaluation values ​​corresponding to each feature point are averaged to obtain the co-evaluation mean value corresponding to each feature point.

[0156] Then, the standard deviation of the multiple collocation evaluation values ​​corresponding to each feature point is calculated to obtain the collocation evaluation standard deviation value corresponding to each feature point.

[0157] Finally, the image texture preservation degree corresponding to each feature point is calculated based on multiple co-evaluation values, co-evaluation mean values, and co-evaluation standard deviation values ​​corresponding to each feature point.

[0158] During post-processing fusion, landscape design requires preserving a wide range of plant-related information, such as vegetation species, vegetation coverage, and terrain information. This, in turn, requires preserving the texture information within multiple remote sensing images. For remote sensing images, the more accurate their texture information, the better they match other remote sensing images, making the preservation of their texture information even more important. Therefore, this embodiment uses the degree to which each feature point retains texture information relative to the corresponding remote sensing image to characterize the degree to which texture information needs to be preserved.

[0159] In this embodiment, the target feature point The image texture preservation of the remote sensing image can be expressed as follows:

[0160] (6)

[0161] Where, Represents target feature points The image texture preservation of the remote sensing image, Indicates reference feature points and target feature points The same evaluation value between them; Represents target feature points The average of the same-position evaluation values ​​with all other reference feature points is the same-position evaluation mean; Represents target feature points The standard deviation of the same-position evaluation values ​​with all other reference feature points, that is, the standard deviation of the same-position evaluation; z represents a non-zero constant with a value range of (0, 0.1) to avoid the denominator being 0. ( ) represents the hyperbolic tangent function, which is used for direct proportion normalization.

[0162] In one embodiment, the fusion weights corresponding to the respective image regions in each remote sensing image are determined based on the image texture preservation corresponding to the respective feature points in each image region, specifically including:

[0163] The image texture preservation degrees of all feature points in each image area of ​​each remote sensing image are averaged, and the fusion weights corresponding to each image area in each remote sensing image are calculated.

[0164] In this embodiment, the remote sensing image Image area in The fusion weight can be expressed as follows:

[0165] (7)

[0166] in, Represents the image area The fusion weight of Represents remote sensing images Middle image area Neidi The image texture preservation degree corresponding to the feature points; Represents remote sensing images Middle image area The total number of feature points in the.

[0167] In one embodiment, the weighted fused remote sensing image is post-processed to obtain the remote sensing survey data processing results of the target garden area, specifically including:

[0168] First, the weighted fused remote sensing image is cropped to obtain the regional image corresponding to the target garden area.

[0169] In this embodiment, with the help of a vector diagram of the target garden area, a regional image of the target garden area can be cropped out from the weighted fused remote sensing image using an image cropping function through remote sensing image processing software, such as ENVI (The Environment for Visualizing Images).

[0170] Then, the regional image is divided into a plurality of image blocks, and the gray level co-occurrence matrix of each image block is determined respectively.

[0171] It can be understood that the gray level co-occurrence matrix can describe the texture characteristics of an image block by calculating the frequency of occurrence of gray values ​​of two pixels having a specific spatial relationship in the image block.

[0172] Subsequently, the similarity of the gray level co-occurrence matrices between adjacent image blocks is determined, and the image blocks are merged according to the similarity of the gray level co-occurrence matrices between adjacent image blocks to obtain a block-merged image.

[0173] It can be understood that since different plants have different texture characteristics in the block image, such as roughness, contrast, etc., this embodiment randomly establishes multiple different initial points and uses the region growing method to merge the image blocks according to the similarity of the grayscale co-occurrence matrix between adjacent image blocks.

[0174] Finally, clustering processing is performed on the block merged image to obtain the remote sensing survey data processing results of the target garden area.

[0175] In this embodiment, clustering processing can be used to obtain the distribution of various plants within the target garden area, i.e., the processing results of remote sensing survey data. In practical applications, after obtaining the corresponding distribution areas of various plants, the light and water requirements of the plant species in the target garden area are examined, and plant species that complement the target garden area can be matched with their needs. The location of new plants can also be allocated based on the distribution of existing plants. For example, if the existing plants in a certain garden area are mostly light-loving plants that are tall, lush, and sparsely distributed, then some short, shade-tolerant plants, such as green radish and monstera, can be arranged in the vacant positions. This not only fills the empty spaces in the garden but also forms a spatial layout with different heights.

[0176] Based on the same inventive concept, the present invention also protects a remote sensing survey data processing system for garden design. The remote sensing survey data processing system for garden design provided by the present invention is described below. The remote sensing survey data processing system for garden design described below and the remote sensing survey data processing method for garden design described above can be referenced to each other.

[0177] like Figure 7As shown, the remote sensing survey data processing system for garden design provided by the embodiment of the present invention specifically includes:

[0178] The acquisition module 310 is used to acquire multiple remote sensing images containing the target garden area in different spectral bands.

[0179] The construction module 320 is used to construct a texture sequence corresponding to each feature point in each remote sensing image.

[0180] The evaluation value calculation module 330 is used to determine the collocation evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in other remote sensing images based on the texture sequence, and obtain multiple collocation evaluation values ​​corresponding to each feature point.

[0181] The preservation degree calculation module 340 is used to determine the image texture preservation degree corresponding to each feature point based on multiple co-location evaluation values ​​corresponding to each feature point.

[0182] The weight calculation module 350 is used to uniformly divide each remote sensing image into multiple image regions, and determine the fusion weight corresponding to each image region in each remote sensing image based on the image texture preservation corresponding to each feature point in each image region.

[0183] The processing module 360 ​​is used to perform weighted fusion on multiple remote sensing images according to the fusion weights corresponding to each image area, and post-process the weighted fused remote sensing images to obtain the remote sensing survey data processing results of the target garden area.

[0184] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated again here.

[0185] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0186] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A remote sensing survey data processing method for garden design, characterized in that: The method comprises: Acquire multiple remote sensing images containing the target garden area in different spectral bands; Construct the texture sequence corresponding to each feature point in each remote sensing image; Determining, based on the texture sequence, a collocation evaluation value between each feature point in each remote sensing image and a corresponding reference feature point in other remote sensing images, and obtaining a plurality of collocation evaluation values ​​corresponding to each feature point; Determine the image texture preservation degree corresponding to each feature point based on multiple co-location evaluation values ​​corresponding to each feature point; Each remote sensing image is uniformly divided into multiple image regions, and the fusion weight corresponding to each image region in each remote sensing image is determined according to the image texture preservation degree corresponding to each feature point in each image region; According to the fusion weights corresponding to each image area, multiple remote sensing images are weighted fused by region, and the weighted fused remote sensing images are post-processed to obtain the remote sensing survey data processing results of the target garden area; The method for obtaining the multiple co-location evaluation values ​​corresponding to each feature point includes: Determining, based on the texture sequence, a texture matching degree between each feature point in each remote sensing image and each feature point to be matched in other remote sensing images; The feature points to be matched whose texture matching degree is greater than a preset matching threshold are used as candidate feature points; Determining the positioning isotexture degree between each feature point in each remote sensing image and feature points in other remote sensing images based on the texture matching degree and the candidate feature points; According to the positioning isotexture degree between each feature point and feature points in other remote sensing images, the texture consistency degree of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood sub-region and the reference degree of the neighborhood sub-region are determined respectively; Based on the texture consistency and the reference degree of the neighborhood sub-region, calculating the co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images; The candidate feature point with the highest co-location evaluation value in each other remote sensing image is used as the reference feature point, and the co-location evaluation value between each feature point in each remote sensing image and the corresponding reference feature point in each other remote sensing image is used as the multiple co-location evaluation values ​​corresponding to each feature point; The formula for calculating texture consistency is: Where, Represents target feature points Target remote sensing image and alternative feature points Other remote sensing images The texture consistency of the kth neighborhood sub-region is, Represents target feature points Target remote sensing image Any feature point and the sth neighborhood feature point in the kth neighborhood subregion Positioning homography between them; represents the total number of neighborhood feature points contained in the kth neighborhood sub-region, and y represents a non-zero constant. Represents target feature points Target remote sensing image Any feature point And the weighted sum of all neighborhood feature points in the kth neighborhood subregion; The method for obtaining the image texture preservation degree corresponding to each feature point includes: The multiple co-evaluation values ​​corresponding to each feature point are averaged to obtain the co-evaluation mean corresponding to each feature point; Calculate the standard deviation of multiple parity evaluation values ​​corresponding to each feature point to obtain the parity evaluation standard deviation value corresponding to each feature point; The image texture preservation degree corresponding to each feature point is calculated based on multiple co-evaluation values, co-evaluation mean values, and co-evaluation standard deviation values ​​corresponding to each feature point.

2. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: Construct the texture sequence corresponding to each feature point in each remote sensing image, including: For each remote sensing image, obtain the gradient direction corresponding to each feature point, the edge points of the neighborhood sub-region where each feature point is located, and the gradient direction of the edge points; Take the lower left corner edge point of the neighborhood sub-region where each feature point is located as the starting point, and use the gradient direction corresponding to the lower left corner edge point as the starting element value; Starting from the starting point, traverse the gradient directions corresponding to the edge points in sequence along a preset direction to obtain a plurality of sequentially arranged intermediate element values; wherein the preset direction is first rightward to the lower right corner, and then upward to the upper right corner; Until the gradient direction corresponding to the last edge point in the upper right corner is traversed, the end element value is obtained; According to the starting element value, a plurality of sequentially arranged intermediate element values ​​and an end element value, a texture sequence corresponding to each feature point in each remote sensing image is constructed.

3. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: Determining the positioning isotexture degree between each feature point in each remote sensing image and feature points in other remote sensing images based on the texture matching degree and the candidate feature points, including: The target feature point with the largest texture matching degree in each remote sensing image is used as the first transition point, and the candidate feature point corresponding to the target feature point is used as the second transition point; Determining, based on the first transition point and the second transition point, a positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images, thereby obtaining a plurality of positioning matching degrees corresponding to each feature point in each remote sensing image; The maximum positioning matching degree among multiple positioning matching degrees corresponding to each feature point is extracted, and the positioning isomorphism between each feature point in each remote sensing image and the feature points in other remote sensing images is determined based on the maximum positioning matching degree.

4. The remote sensing survey data processing method for garden design according to claim 3, characterized in that: Determining the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images based on the first transition point and the second transition point, including: Determining a first distance value between each feature point in the same remote sensing image and the first transition point, and determining a second distance value between each feature point in the same remote sensing image and the second transition point; Calculating an absolute value of a difference between the first distance value and the second distance value; Inputting the absolute value of the difference into an inverse proportional normalization function to obtain an inverse proportional normalized value; Determining a texture matching degree between a feature point in the remote sensing image where the first transition point is located and a feature point in the remote sensing image where the second transition point is located; The inversely proportional normalized value is multiplied by the texture matching degree between the feature point in the remote sensing image where the first transition point is located and the feature point in the remote sensing image where the second transition point is located, to obtain the positioning matching degree between each feature point in each remote sensing image and all feature points in other remote sensing images.

5. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: Based on the positioning isotexture degree between each feature point and feature points in other remote sensing images, the texture consistency of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood sub-region and the reference degree of the neighborhood sub-region are determined, including: Determine the neighborhood feature points and the total number of neighborhood feature points contained in the neighborhood sub-region, and determine the positioning isotexture between each feature point in the remote sensing image and each neighborhood feature point in the same neighborhood sub-region of the remote sensing image where any candidate feature point is located; Sum the localization iso-grainness corresponding to all neighborhood feature points to obtain the sum value of iso-grainness; Determining a first intermediate value according to the sum of the same-grain degrees; Perform a quadratic operation on the positioning isotexture corresponding to each neighborhood feature point to obtain the isotexture product value of each neighborhood feature point; Sum the product values ​​of the same texture of all neighborhood feature points to obtain the second intermediate value; Calculating, based on the first intermediate value and the second intermediate value, the texture consistency of each remote sensing image and the remote sensing image where any candidate feature point is located in the same neighborhood subregion; The reference degree of the neighborhood sub-region is obtained by dividing the sum of the isotexture values ​​by the total number of the neighborhood feature points.

6. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: Based on the texture consistency and the reference degree of the neighborhood sub-region, the co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature points in other remote sensing images is calculated, including: Determine all neighborhood sub-regions contained in the neighborhood range corresponding to each feature point in each remote sensing image; Multiply the texture consistency and reference degree corresponding to each neighborhood sub-region to obtain the parameter product value corresponding to each neighborhood sub-region; Sum the parameter product values ​​of all neighborhood sub-regions to obtain the product sum value; Sum the reference degrees corresponding to all neighborhood sub-regions to obtain the reference degree sum value; The product sum value is divided by the reference degree sum value to obtain a co-location evaluation value between each feature point in each remote sensing image and the corresponding candidate feature point in other remote sensing images.

7. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: According to the image texture preservation degree corresponding to each feature point in each image area, the fusion weight corresponding to each image area in each remote sensing image is determined respectively, including: The image texture preservation degrees of all feature points in each image area of ​​each remote sensing image are averaged, and the fusion weights corresponding to each image area in each remote sensing image are calculated.

8. The remote sensing survey data processing method for garden design according to claim 1, characterized in that: Post-process the weighted fused remote sensing images to obtain the remote sensing survey data processing results of the target garden area, including: The weighted fused remote sensing image is cropped to obtain the regional image corresponding to the target garden area; Divide the regional image into multiple image blocks, and determine the gray level co-occurrence matrix of each image block respectively; Determine the similarity of gray level co-occurrence matrices between adjacent image blocks, and merge the image blocks based on the similarity of the gray level co-occurrence matrices between adjacent image blocks to obtain a block-merged image; Clustering processing is performed on the block merged image to obtain remote sensing survey data processing results of the target garden area.

Images