Drawing method and system based on soil type

Abstract

The invention discloses a soil type-based mapping method and system, and relates to the technical field of image processing. The mapping method based on the soil type comprises the following steps: acquiring an angular point array of a first spectral image and a second spectral image of a target soil area in an image overlapping area; for each angular point array, determining the observability of the angular point array by using the reflectivity of each angular point; according to the observability of each angular point array, dividing to obtain a first angular point array group and a second angular point array group; respectively comparing the target first angular point array group with each second angular point array group to obtain a target second angular point array group which meets a preset similarity condition with the target first angular point array group; obtaining a first corrected spectral image and a second corrected spectral image according to the target first angular point array group and the target second angular point array group; and splicing the first corrected spectral image and the second corrected spectral image to obtain a target soil area image.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a soil type-based mapping method and system. Background Art

[0002] In soil science, soil type classification is an important research content and component, and soil classification represents the development level of soil science. Therefore, in-depth research on soil classification has vital theoretical and practical significance.

[0003] In the traditional soil type mapping method, soil type mapping is completed by stitching soil remote sensing images taken by drones.

[0004] However, due to the offset of UAV flight, the sampling points in some areas of the captured soil images may be offset, which will reduce the accuracy of mapping in the process of stitching multiple soil images to complete soil type-based mapping. Summary of the invention

[0005] The embodiment of the present invention provides a soil type-based mapping method and system, which can improve the accuracy of mapping.

[0006] A first aspect of an embodiment of the present invention provides a soil type-based mapping method, comprising: Acquire at least one corner point array of a first spectral image and a second spectral image of a target soil region in an image overlap region, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments; For each corner point array, the observability of the corner point array is determined by using the reflectivity of each corner point in the corner point array; According to the observability of each corner point array, at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image are obtained by division, wherein the corner point array group includes a plurality of corner point arrays belonging to the same soil type; Compare the target first corner point array group with each second corner point array group respectively to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group; According to the target first corner point array group and the target second corner point array group, the array distance between the corner point arrays of the target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image; The first corrected spectral image and the second corrected spectral image are spliced ​​to obtain the target soil area image.

[0007] A second aspect of an embodiment of the present invention provides a soil type-based mapping system, comprising: An array acquisition module, used to acquire at least one corner point array of a first spectral image and a second spectral image of a target soil area in an image overlap area, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments; An array calculation module is used to determine the observability of each corner point array by using the reflectivity of each corner point in the corner point array; An array grouping module, used for obtaining at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image according to the observability of each corner point array, wherein the corner point array group includes a plurality of corner point arrays belonging to the same soil type; An array group comparison module, used to compare the target first corner point array group with each second corner point array group, to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group; A distance correction module is used to correct the array distance between the corner point arrays of the first corner point array group of the target and the second corner point array group of the target according to the first corner point array group of the target and the second corner point array group of the target, so as to obtain a first corrected spectral image and a second corrected spectral image; The image stitching module is used to stitch the first corrected spectral image with the second corrected spectral image to obtain a target soil area image.

[0008] In the soil type-based mapping method provided by an embodiment of the present invention, the observability of the corner point array is determined according to the reflectivity of each corner point in the corner point array, which is used to evaluate the offset of the corner points in the first spectral image and the second spectral image. Then, each corner point array is grouped to obtain a first corner point array group belonging to the same soil type in the first spectral image and a second corner point array group belonging to the same soil type in the second spectral image. Then, the array distance between the corner point arrays of the matching target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image. In this way, the present invention can improve the accuracy of mapping by correcting the offset of the corner points in the first spectral image and the second spectral image, and then splicing the corrected first corrected spectral image and the second corrected spectral image. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. 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 creative work.

[0010] Figure 1 A schematic flow chart of a first soil type-based mapping method provided by an embodiment of the present invention; Figure 2 A schematic diagram of a spectral image of a target soil area provided by an embodiment of the present invention; Figure 3 A schematic flow chart of a second soil type-based mapping method provided by an embodiment of the present invention; Figure 4 A schematic flow chart of a third soil type-based mapping method provided by an embodiment of the present invention; Figure 5 A schematic diagram of the structure of a soil type-based mapping system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a soil type mapping method and system proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0012] 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.

[0013] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of the present invention are in compliance with the relevant provisions of laws and regulations.

[0014] It should be noted that in the embodiments of the present invention, certain software, components, models and other existing solutions in the industry may be mentioned, which should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present invention, but it does not mean that the applicant has or will necessarily use the solution.

[0015] In the existing method, soil type mapping is completed by stitching soil remote sensing images taken by drones. However, due to the offset of drone flight, the sampling points of some areas in the soil images taken may be offset, which will reduce the accuracy of mapping in the process of stitching multiple soil images to complete soil type mapping.

[0016] The object of the present invention is to provide a mapping method and system based on soil type. In the mapping method based on soil type provided by an embodiment of the present invention, the observability of the corner point array is determined according to the reflectivity of each corner point in the corner point array, which is used to evaluate the offset of the corner points in the first spectral image and the second spectral image. Then, each corner point array is grouped to obtain a first corner point array group belonging to the same soil type in the first spectral image and a second corner point array group belonging to the same soil type in the second spectral image. Then, the array distance between the corner point arrays of the matching target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image. In this way, the present invention can improve the accuracy of mapping by correcting the offset of the corner points in the first spectral image and the second spectral image, and then splicing the corrected first corrected spectral image and the second corrected spectral image.

[0017] The following describes a specific embodiment of a soil type-based mapping method and system provided by an embodiment of the present invention.

[0018] Figure 1 A flowchart of a soil type-based mapping method is provided. The soil type-based mapping method can be applied to a server. The soil type-based mapping method can include the following S101 to S106.

[0019] S101, obtaining at least one corner point array of a first spectral image and a second spectral image of a target soil area in an image overlapping area, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments.

[0020] In this embodiment, a multispectral image sensor can be used to collect the first spectral image and the second spectral image of the target soil area. The multispectral image sensor has four spectral channels, namely: green light (wavelength 550nm, bandwidth 40nm), red light (wavelength 660nm, bandwidth 40nm), red edge (wavelength 735nm, bandwidth 10nm) and near infrared NIR (wavelength 790nm, bandwidth 40nm). For example, Figure 2 As shown, a schematic diagram of a spectral image of a target soil area is provided.

[0021] The image overlap area is used to represent the area where the first spectral image and the second spectral image are repeatedly photographed. There are multiple corner points arranged according to a certain rule or pattern in the corner point array, and the corner points are used to represent points in the image where the grayscale value changes greatly in the horizontal and vertical directions.

[0022] As an example, the server uses a multispectral image sensor to shoot the target soil area at two adjacent moments to obtain two spectral images, namely a first spectral image and a second spectral image. The first spectral image and the second spectral image may overlap in some areas.

[0023] Then, in the image overlapping area of ​​the first spectral image and the second spectral image, image processing technology (such as Harris corner point detection, Shi-Tomasi corner point detection, etc.) is used to extract corner points, and these corner points form a corner point array.

[0024] S102, for each corner point array, using the reflectivity of each corner point in the corner point array, determining the observability of the corner point array.

[0025] In this embodiment, observability is an indicator that comprehensively reflects the clarity and stability of the corner point array and is used to evaluate the reliability of the corner point array in image stitching.

[0026] As an example, each corner point in the corner point array has its corresponding reflectance value, which reflects the spectral characteristics of the soil surface. By calculating the mean, standard deviation or other statistics of the reflectance of all corner points in the corner point array, the observability of the corner point array, that is, its stability and reliability as a feature point, is evaluated.

[0027] S103, according to the observability of each corner point array, obtaining at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image, wherein the corner point array group includes multiple corner point arrays belonging to the same soil type.

[0028] In this embodiment, each corner point array in the corner point array group belongs to the same soil type or has similar reflectivity characteristics.

[0029] As an example, the server classifies corner point arrays with similar observability into the same group according to the observability of each corner point array, so that the corner point array in the first spectral image can be divided into at least one first corner point array group, and the corner point array in the second spectral image can be divided into at least one second corner point array group. It is ensured that the corner point arrays in the same group belong to the same soil type to improve the accuracy of image stitching.

[0030] S104, comparing the target first corner point array group with each second corner point array group respectively, to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group.

[0031] In this embodiment, the target first corner point array group is any one of the first corner point array groups, and the target second corner point array group is a second corner point array group that satisfies a preset similarity condition with the target first corner point array group. The preset similarity condition may include similarity of features such as shape, size, and reflectivity distribution of the corner point arrays.

[0032] As an example, the server uses a feature matching algorithm (such as SIFT, SURF, ORB, etc.) to compare the corner point arrays in the first corner point array group and the second corner point array group, and finds corner point array pairs that are highly similar in spatial position and spectral characteristics, thereby obtaining the target first corner point array group and its corresponding target second corner point array group.

[0033] S105 , according to the target first corner point array group and the target second corner point array group, correct the array distance between the corner point arrays of the target first corner point array group and the target second corner point array group to obtain a first corrected spectral image and a second corrected spectral image.

[0034] In this embodiment, the server side calculates the transformation matrix between the first corner point array group of the target and the second corner point array group based on the image registration algorithm to achieve accurate correction of the array distance between the first corner point array group of the target and the second corner point array group of the target, and obtains the first corrected spectral image and the second corrected spectral image after correction.

[0035] S106, splicing the first corrected spectral image and the second corrected spectral image to obtain a target soil area image.

[0036] In this embodiment, the server splices the first corrected spectral image with the second corrected spectral image to obtain the target soil area image. The splicing process can be based on an image fusion algorithm (such as weighted average, multi-band fusion, etc.), and the two corrected spectral images are smoothly transitioned in the overlapping area to eliminate splicing marks and retain image details.

[0037] In the soil type-based mapping method provided in this embodiment, the observability of the corner point array is determined according to the reflectivity of each corner point in the corner point array, which is used to evaluate the offset of the corner points in the first spectral image and the second spectral image. Then, each corner point array is grouped to obtain a first corner point array group belonging to the same soil type in the first spectral image and a second corner point array group belonging to the same soil type in the second spectral image. Then, the array distance between the corner point arrays of the matching target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image. In this way, the present invention can improve the accuracy of mapping by correcting the offset of the corner points in the first spectral image and the second spectral image, and then splicing the corrected first corrected spectral image and the second corrected spectral image.

[0038] As an optional embodiment, the corner point array includes a first corner point array and a second corner point array; like Figure 3 As shown, S101 may specifically include the following S301 to S303: S301, determining a first spectral image and a second spectral image according to shooting times corresponding to the spectral images of the target soil area; S302, comparing the first spectral image with the second spectral image to obtain an image overlap area between the first spectral image and the second spectral image; S303, obtaining a first corner point array of the first spectral image in the image overlapping area, and obtaining a second corner point array of the second spectral image in the image overlapping area.

[0039] In this embodiment, the first corner point array is used to characterize the corner point array of the first spectral image in the image overlap region, and the second corner point array is used to characterize the corner point array of the second spectral image in the image overlap region.

[0040] As an example, the server selects two representative or specific spectral images taken at adjacent times from a large number of spectral images according to the shooting times corresponding to the spectral images of the target soil area, and uses them as the first spectral image and the second spectral image respectively.

[0041] Then, by detecting the feature points (such as corner points, edge points, etc.) in the first spectral image and the second spectral image and calculating the matching relationship between them, the corresponding relationship between the first spectral image and the second spectral image can be established. And according to the matching relationship of the feature points, a suitable transformation model (such as affine transformation, projection transformation, etc.) is estimated so that the first image can be aligned with the second image in the overlapping area after transformation. Thus, using the transformation model, the first image is transformed to the same coordinate system as the second image, and then the overlapping area of ​​the two images is calculated.

[0042] Finally, a corner point detection algorithm is used to detect the corner points of the first spectral image and the second spectral image in the overlapping area of ​​the images, and the detected corner points are arranged according to their positions in the images to form a corner point array.

[0043] Through this embodiment, the first spectral image and the second spectral image of the target soil area are obtained, and then the first spectral image and the second spectral image are compared to determine the image overlap area, and the first corner point array and the second corner point array in the image overlap area are obtained. In this way, accurate comparison and analysis of the spectral image of the target soil area can be achieved, which is helpful for subsequent correction of the spectral image according to the corner point array, thereby improving the accuracy of mapping.

[0044] As an optional embodiment, S302 may specifically include: Determine a first pixel point corresponding to the first spectral image and a second pixel point corresponding to the second spectral image according to internal and external parameters of the image acquisition device and an image coverage area; Aligning the first pixel point with the second pixel point to obtain overlapping pixel points of the first spectral image and the second spectral image; The area of ​​overlapping pixels is determined as the image overlapping area of ​​the first spectral image and the second spectral image.

[0045] In this embodiment, the internal and external parameters of the image acquisition device may include internal parameters and external parameters. The internal parameters mainly include the focal length, principal point coordinates, distortion coefficients, etc. of the camera, which determine the optical properties and imaging relationship inside the camera. The external parameters describe the position and posture of the camera in three-dimensional space, including the rotation matrix and translation vector, which determine the relative position relationship between the camera and the outside world.

[0046] As an example, the server can calculate the three-dimensional spatial position corresponding to each pixel point in the spectral image based on the internal and external parameters of the image acquisition device and the image coverage area of ​​the spectral image, thereby obtaining the first pixel point corresponding to the first spectral image and the second pixel point corresponding to the second spectral image.

[0047] Then, the first pixel of the first spectral image is aligned with the second pixel of the second spectral image. Specifically, an image registration technology is used to calculate the transformation relationship between the two spectral images so that they can be aligned at the pixel level.

[0048] Among them, in the alignment process, methods such as feature point matching and gray value comparison can be used to establish the correspondence between pixels. Feature point matching is to establish the correspondence by detecting feature points (such as corner points, edge points, etc.) in the spectral image and calculating the matching relationship between them. Gray value comparison is to establish the correspondence by comparing the gray values ​​of pixels in the spectral image.

[0049] Finally, in the aligned spectral image, find the overlapping pixel points corresponding to the first pixel point and the second pixel point. According to the position of the overlapping pixel points, the image overlapping area of ​​the first spectral image and the second spectral image can be determined. Specifically, connected domain analysis, morphological processing and other methods can be used to extract the overlapping area. Connected domain analysis extracts the overlapping area by analyzing the connectivity of overlapping pixels, while morphological processing uses morphological operations (such as expansion, erosion, etc.) to optimize and extract the overlapping area.

[0050] Through this embodiment, the first spectral image is compared with the second spectral image, so as to accurately obtain the image overlap area of ​​the first spectral image and the second spectral image. This helps to subsequently stitch the images according to the image overlap area to obtain the target soil area image, thereby improving the accuracy of mapping.

[0051] As an optional embodiment, S102 may specifically include: Based on the reflectivity of each corner point in the corner point array, determine the reflectivity standard deviation and the reflectivity mean; Divide the difference between the reflectivity of each corner point and the mean reflectivity by the standard deviation of the reflectivity to obtain the observability of each corner point; The observability of the corner point array is determined using the observability of each corner point.

[0052] In this embodiment, the observability of the corner point is determined by the following formula 1: Formula 1 In the formula, Used to characterize the observability of corner point z. It is used to characterize the reflectivity of the corresponding position of the corner point z on the spectral image, where the pixel point in the multispectral image represents the reflectivity of the corresponding position, that is, the reflectivity of the corresponding position can be obtained directly by obtaining the pixel point. It is used to characterize the mean reflectivity in the corner point array where the corner point z is located. Used to characterize the standard deviation of reflectivity in the corner point array where the corner point z is located.

[0053] Among them, the greater the reflectivity of the corresponding position of the corner point z on the spectral image, the smaller the reflectivity mean or reflectivity standard deviation in the corner point array where the corner point z is located, and the greater the corresponding observability.

[0054] The observability of the corner point array is determined by the following formula 2: Formula 2 In the formula, It is used to characterize the observability of the corner point array where the corner point z is located. It is used to characterize the number of corner points in the corner point array where the corner point z is located. Used to characterize the observability of corner point z.

[0055] in, It is used to accumulate from the first corner point to the last corner point in the corner point array where the corner point z is located. The greater the observability of each corner point in the corner point array, the greater the observability of the corner point array.

[0056] Among them, the greater the observability of the corner point array, the closer the soil types of the corner points in the corner point array are, and therefore the regional consistency after splicing is better.

[0057] Through this embodiment, the observability of each corner point is accurately calculated according to the reflectivity of each corner point in the corner point array, thereby obtaining the observability of the corner point array. This helps to accurately divide the corner point arrays belonging to the same soil type into the same corner point array group according to the observability of the corner point array. This can further realize the correction of the spectral image, thereby improving the accuracy of mapping.

[0058] As an optional embodiment, S103 may specifically include: For the first spectral image and the second spectral image, the following steps are performed respectively until all corner point arrays in the first spectral image and the second spectral image are traversed: Obtaining the absolute value of the difference between the observability of the target corner point array and the observability of other corner point arrays in the target spectral image; Sort the absolute values ​​of the observable degree differences from small to large to obtain an ascending sequence of differences; Perform first-order difference calculation on the ascending sequence of differences to obtain a difference difference sequence; According to the difference difference sequence, the absolute value of the difference from the target position to the last observable degree in the difference ascending sequence is used to form the target sequence. The target position is the position corresponding to the maximum difference value in the difference difference sequence in the difference ascending sequence. The target corner point array and the corner point array corresponding to the absolute value of each observable degree difference in the target sequence form a target corner point array group.

[0059] In this embodiment, the target corner point array is used to characterize any unprocessed corner point array among the corner point arrays, and the target corner point array group is used to characterize any one of the first corner point array groups and the second corner point array groups.

[0060] As an example, taking the first spectral image as an example, the following steps are performed respectively until each corner point array in the first spectral image is traversed: First, the server calculates the observability of each corner point array in the first spectral image, and then calculates the observability difference between the target corner point array and other corner point arrays in the first spectral image, and takes the absolute value. Then, a sorting algorithm (such as quick sort, merge sort, etc.) is used to sort the absolute values ​​of the observability differences in ascending order to obtain an ascending sequence of differences.

[0061] Then, traverse the ascending sequence of differences, calculate the difference between each element in the ascending sequence of differences and its previous element, and form a difference sequence of differences. Find the maximum difference in the difference sequence, and then determine the absolute value of the two corresponding observable degree differences in the ascending sequence of differences according to the maximum difference, and take the larger absolute value of the two observable degree differences as the target position. Starting from the target position to the last element of the ascending sequence of differences, the target sequence is formed.

[0062] Finally, the target corner point array and the two corner point arrays corresponding to the absolute values ​​of the observable degree differences in the target sequence together constitute a target corner point array group.

[0063] Through this embodiment, at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image are divided according to the observability of each corner point array. In this way, it is helpful to correct the first spectral image and the second spectral image according to the first corner point array group in the first spectral image and the second corner point array group in the second spectral image to obtain the first corrected spectral image and the second corrected spectral image, thereby improving the accuracy of mapping.

[0064] As an optional embodiment, Figure 4 As shown, S104 may specifically include the following S401 to S404: S401, acquiring a target first region corresponding to a target first corner point array group, and acquiring a second region corresponding to each second corner point array group; S402, determining a second region with the largest overlap with the target first region as a candidate second region; S403, determining the target similarity between the target first region and the candidate second region by using the target first corner point array group and the candidate second corner point array group corresponding to the candidate second region; S404: When the target similarity is greater than a preset similarity threshold, determine the candidate second corner point array group as the target second corner point array group.

[0065] In this embodiment, the first area of ​​the target is used to represent the closed area generated by the mutual linkage between the corresponding points of each corner point array in the first corner point array group of the target, and the second area is used to represent the closed area generated by the mutual linkage between the corresponding points of each corner point array in the second corner point array group.

[0066] As an example, the server links each point position to generate a closed area based on the point position corresponding to each corner point array in the target first corner point array group, and the closed area is the target first area. For each second corner point array group, the second area corresponding to each second corner point array group is determined in the same manner.

[0067] Then, the overlapping area between the target first region and each second region is calculated, and the second region with the largest overlapping area is selected as the candidate second region.

[0068] Then, for each corner point in the target first corner point array group and the candidate second corner point array group, feature descriptors (such as SIFT, SURF, ORB, etc.) are extracted. Feature matching algorithms (such as FLANN, BFMatcher, etc.) are used to match the feature descriptors of the two groups of corner points. The target similarity between the target first region and the candidate second region is calculated based on the number of successfully matched corner point pairs, the geometric consistency (such as distance, angle, etc.) between the matched corner points, and the similarity between the feature descriptors of the matched corner points.

[0069] Finally, a preset similarity threshold is set, and the preset similarity threshold is adjusted according to the application scenario and image quality. If the calculated target similarity is greater than the preset similarity threshold, the candidate second corner point array group is considered to be sufficiently similar to the target first corner point array group and is determined as the target second corner point array group; if the calculated target similarity is less than or equal to the preset similarity threshold, it is necessary to reselect the candidate second region or determine that there is no target second corner point array group corresponding to the target first corner point array group.

[0070] Through this embodiment, the target first corner point array group is compared with each second corner point array group to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group. This helps to correct the first spectral image and the second spectral image according to the target first corner point array group and the target second corner point array group to obtain a first corrected spectral image and a second corrected spectral image, thereby improving the accuracy of mapping.

[0071] As an optional embodiment, S403 may specifically include: Calculate the first observability mean based on the observability of each corner point array in the target first corner point array group; Calculating a second observability mean based on the observability of each corner point array in the candidate second corner point array group; The target similarity between the target first region and the candidate second region is determined by using the first observability mean and the second observability mean.

[0072] In this embodiment, the target similarity is determined by the following formula 3: Formula 3 In the formula, It is used to characterize the first observable degree mean corresponding to the first corner point array group of the target, It is used to characterize the second observability mean corresponding to the candidate second corner point array group, and B is used to characterize the target similarity between the target first region and the candidate second region.

[0073] The closer the first observability mean is to the second observability mean, the greater the target similarity between the corresponding target first region and the candidate second region.

[0074] Through this embodiment, the target first corner point array group and the candidate second corner point array group corresponding to the candidate second area are used to determine the target similarity between the target first area and the candidate second area. In this way, it is possible to accurately determine whether the target first corner point array group matches the candidate second corner point array group based on the target similarity, thereby facilitating the correction of the first spectral image and the second spectral image to obtain the first corrected spectral image and the second corrected spectral image, and improving the accuracy of mapping.

[0075] As an optional embodiment, S105 may specifically include: Linearly normalizing the observability of each corner point array in the first target corner point array group and the second target corner point array group to obtain corresponding normalized observability; Using each normalized observable degree, correcting the array distance between the corner point arrays of the first target corner point array group and the second target corner point array group to obtain a corrected array distance; The first spectral image and the second spectral image are corrected according to the corrected array distance to obtain a first corrected spectral image and a second corrected spectral image.

[0076] In this embodiment, the server first extracts the respective corner point arrays from the target first corner point array group and the target second corner point array group. Then, the observability of each corner point array is calculated, and these observabilities are linearly normalized. Specifically, linear normalization is usually to scale all observability values ​​to a specific range (such as 0 to 1) for subsequent processing, which can be achieved by subtracting the minimum value from each observability value and dividing it by the difference between the maximum value and the minimum value.

[0077] Then, based on the normalized observability, the corrected array distance is determined by the following formula 4: Formula 4 In the formula, It is used to represent the corrected array distance, and d is used to represent the original array distance. Used to characterize the normalized observability of the corner point array in the first corner point array group of the target, Used to characterize the normalized observability of the corner point array in the second corner point array group of the target.

[0078] Finally, after obtaining the corrected array distance, the corrected array distance is allocated according to the ratio of the normalized observability of the center point of the target first corner point array group and the center point of the target second corner point array group, with the mid-range point in the original array distance as the center. Thus, according to the allocated corrected array distance, the first spectral image and the second spectral image are respectively transformed to align the same features in the two images, and the first corrected spectral image and the second corrected spectral image are obtained.

[0079] Through this embodiment, according to the target first corner point array group and the target second corner point array group, the array distance between the corner point arrays of the target first corner point array group and the target second corner point array group is corrected, thereby obtaining the first corrected spectral image and the second corrected spectral image. In this way, by obtaining the first corrected spectral image and the second corrected spectral image, the corrected first corrected spectral image and the second corrected spectral image are directly spliced, which can improve the accuracy of mapping.

[0080] As an optional embodiment, S106 may specifically include: Acquire a first homography matrix of corner points in the first corrected spectral image and a second homography matrix of corner points in the second corrected spectral image; Determine, according to the first homography matrix and the second homography matrix, a target corrected spectral image having a larger corner point movement distance between the first corrected spectral image and the second corrected spectral image; Performing perspective transformation on the target corrected spectral image to determine the target image size corresponding to the target soil area image; According to the target image size, the first corrected spectral image and the second corrected spectral image are spliced ​​to obtain the target soil area image.

[0081] In this embodiment, the homography matrix is ​​used to represent the projection transformation relationship from one plane to another plane.

[0082] As an example, for each corner point in each corrected spectral image, the server uses a feature matching algorithm (such as SIFT, SURF, ORB, etc.) to find the corresponding matching point in another corrected spectral image. Then, using these matching point pairs, two homography matrices are calculated by the least squares method or other optimization algorithms: the first homography matrix (for the corner points in the first corrected spectral image) and the second homography matrix (for the corner points in the second corrected spectral image).

[0083] Then, the corner points in each image are mapped to the other image using the calculated homography matrix, and the Euclidean distance or other distance measure between the mapped corner points and the true matching corner points is calculated as the movement distance of the corner points. The average movement distance of the corner points in the first corrected spectral image is compared with the average movement distance of the corner points in the second corrected spectral image, and the image with a larger average movement distance is selected as the target corrected spectral image. This usually means that the image has undergone a larger deformation or displacement relative to the other image.

[0084] Then, a perspective transformation is applied to the target corrected spectral image to correct image distortion caused by shooting angle, lens distortion, etc. The parameters of the perspective transformation can be adjusted by the previously calculated homography matrix. After the perspective transformation, the target image size of the stitched image is determined based on the size of the target soil area in the target corrected spectral image.

[0085] Finally, based on the previously calculated homography matrix, the first corrected spectral image and the second corrected spectral image are registered to make them consistent in alignment. Based on the registration, the two images are seamlessly spliced ​​together using an image fusion algorithm (such as weighted average, Laplacian pyramid fusion, etc.) to form the target soil area image.

[0086] Through this embodiment, according to the first homography matrix of the corner points in the first corrected spectral image and the second homography matrix of the corner points in the second corrected spectral image, the first corrected spectral image and the second corrected spectral image are spliced ​​to obtain the target soil area image. In this way, the image splicing is realized by the first homography matrix and the second homography matrix, which can improve the accuracy of image splicing.

[0087] According to the mapping method based on soil type, accordingly, the present invention also provides a specific embodiment of a mapping system based on soil type.

[0088] Figure 5 A structural schematic diagram of a soil type-based mapping system provided in an embodiment of the present invention is shown. The soil type-based mapping system 500 may include an array acquisition module 510, an array calculation module 520, an array grouping module 530, an array group comparison module 540, a distance correction module 550 and an image stitching module 560.

[0089] An array acquisition module 510 is used to acquire at least one corner point array of a first spectral image and a second spectral image of a target soil region in an image overlap region, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments; An array calculation module 520 is used to determine the observability of each corner point array by using the reflectivity of each corner point in the corner point array; An array grouping module 530 is used to divide the first spectral image into at least one first corner point array group and the second spectral image into at least one second corner point array group according to the observability of each corner point array, wherein the corner point array group includes multiple corner point arrays belonging to the same soil type; An array group comparison module 540 is used to compare the target first corner point array group with each second corner point array group, and obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group; The distance correction module 550 is used to correct the array distance between the corner point arrays of the first target corner point array group and the second target corner point array group according to the first target corner point array group and the second target corner point array group, so as to obtain a first corrected spectral image and a second corrected spectral image; The image stitching module 560 is used to stitch the first corrected spectral image with the second corrected spectral image to obtain a target soil area image.

[0090] In the soil type-based mapping system provided in the present embodiment, the observability of the corner point array is determined according to the reflectivity of each corner point in the corner point array, which is used to evaluate the offset of the corner points in the first spectral image and the second spectral image. Then, each corner point array is grouped to obtain a first corner point array group belonging to the same soil type in the first spectral image and a second corner point array group belonging to the same soil type in the second spectral image. Then, the array distance between the corner point arrays of the matching target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image. In this way, the present invention can improve the accuracy of mapping by correcting the offset of the corner points in the first spectral image and the second spectral image, and then splicing the corrected first corrected spectral image and the second corrected spectral image.

[0091] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0092] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0093] The above is only a specific implementation of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.

Claims

1. A soil type-based mapping method, characterized in that: The method comprises: Acquire at least one corner point array of a first spectral image and a second spectral image of a target soil region in an image overlap region, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments; For each of the corner point arrays, the observability of the corner point array is determined by using the reflectivity of each corner point in the corner point array; According to the observability of each of the corner point arrays, at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image are obtained by division, wherein the corner point array group includes a plurality of corner point arrays belonging to the same soil type; Compare the target first corner point array group with each of the second corner point array groups respectively to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group; According to the target first corner point array group and the target second corner point array group, an array distance between the corner point arrays of the target first corner point array group and the target second corner point array group is corrected to obtain a first corrected spectral image and a second corrected spectral image; The first corrected spectral image and the second corrected spectral image are spliced ​​to obtain a target soil area image.

2. The soil type-based mapping method according to claim 1, characterized in that: The corner point array includes a first corner point array and a second corner point array; The step of acquiring at least one corner point array of the first spectral image and the second spectral image of the target soil region in the image overlap region comprises: Determining the first spectral image and the second spectral image according to shooting times corresponding to the spectral images of the target soil area; Comparing the first spectral image with the second spectral image to obtain an image overlap area between the first spectral image and the second spectral image; The first corner point array of the first spectral image in the image overlapping area is acquired, and the second corner point array of the second spectral image in the image overlapping area is acquired.

3. The soil type-based mapping method according to claim 2, characterized in that: The comparing the first spectral image with the second spectral image to obtain an image overlap area between the first spectral image and the second spectral image includes: Determine a first pixel point corresponding to the first spectral image and a second pixel point corresponding to the second spectral image according to internal and external parameters of the image acquisition device and an image coverage area; Aligning the first pixel point with the second pixel point to obtain overlapping pixel points of the first spectral image and the second spectral image; The region of the overlapping pixels is determined as an image overlapping region of the first spectral image and the second spectral image.

4. The soil type-based mapping method according to claim 1, characterized in that: The step of determining the observability of each corner point array by using the reflectivity of each corner point in the corner point array comprises: Determine a reflectivity standard deviation and a reflectivity mean based on the reflectivity of each corner point in the corner point array; Divide the difference between the reflectivity of each corner point and the mean reflectivity by the reflectivity standard deviation to obtain the observability of each corner point; The observability of the corner point array is determined by using the observability of each corner point.

5. The soil type-based mapping method according to claim 1, characterized in that: The step of obtaining at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image according to the observability of each corner point array comprises: For the first spectral image and the second spectral image, the following steps are performed respectively until all corner point arrays in the first spectral image and the second spectral image are traversed: Obtaining the absolute value of the difference between the observability of the target corner point array and the observability of other corner point arrays in the target spectral image; Sort the absolute values ​​of the observability differences in order from small to large to obtain an ascending sequence of differences; Performing first-order difference calculation on the ascending sequence of difference values ​​to obtain a difference difference sequence; According to the difference difference sequence, the absolute values ​​of the observability difference from the target position to the last one in the difference ascending sequence are used to form a target sequence, wherein the target position is the position corresponding to the maximum difference value in the difference difference sequence in the difference ascending sequence; The target corner point array and the corner point arrays corresponding to the absolute values ​​of the observability differences in the target sequence form a target corner point array group.

6. The soil type-based mapping method according to claim 1, characterized in that: The step of comparing the target first corner point array group with each of the second corner point array groups to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group includes: Acquire a target first area corresponding to the target first corner point array group, and acquire a second area corresponding to each of the second corner point array groups; Determine the second region with the largest overlap with the target first region as a candidate second region; Determine the target similarity between the target first region and the candidate second region by using the target first corner point array group and the candidate second corner point array group corresponding to the candidate second region; When the target similarity is greater than a preset similarity threshold, the candidate second corner point array group is determined as a target second corner point array group.

7. The soil type-based mapping method according to claim 6, characterized in that: The determining the target similarity between the target first region and the candidate second region by using the target first corner point array group and the candidate second corner point array group corresponding to the candidate second region includes: Calculating a first observability mean based on the observability of each of the corner point arrays in the target first corner point array group; Calculating a second observability mean based on the observability of each of the corner point arrays in the candidate second corner point array group; The target similarity between the target first region and the candidate second region is determined by using the first observability mean value and the second observability mean value.

8. The soil type-based mapping method according to claim 1, characterized in that: The method of correcting the array distance between the corner point arrays of the first target corner point array group and the second target corner point array group according to the first target corner point array group and the second target corner point array group to obtain a first corrected spectral image and a second corrected spectral image includes: Linearly normalizing the observability of each of the corner point arrays in the target first corner point array group and the target second corner point array group to obtain corresponding normalized observability; Using each of the normalized observable degrees, correcting the array distance between the corner point arrays of the first target corner point array group and the second target corner point array group to obtain a corrected array distance; The first spectral image and the second spectral image are corrected according to the corrected array distance to obtain the first corrected spectral image and the second corrected spectral image.

9. The soil type-based mapping method according to claim 1, characterized in that: The step of splicing the first corrected spectral image with the second corrected spectral image to obtain a target soil area image includes: Acquire a first homography matrix of corner points in the first corrected spectral image, and a second homography matrix of corner points in the second corrected spectral image; Determine, according to the first homography matrix and the second homography matrix, a target corrected spectral image having a larger corner point movement distance between the first corrected spectral image and the second corrected spectral image; Performing perspective transformation on the target corrected spectral image to determine the target image size corresponding to the target soil region image; According to the target image size, the first corrected spectral image and the second corrected spectral image are spliced ​​to obtain the target soil area image.

10. A soil type-based mapping system, characterized in that: The system comprises: An array acquisition module, used to acquire at least one corner point array of a first spectral image and a second spectral image of a target soil area in an image overlap area, wherein the first spectral image and the second spectral image are two spectral images taken at adjacent moments; An array calculation module, for determining the observability of each corner point array by using the reflectivity of each corner point in the corner point array; an array grouping module, configured to obtain at least one first corner point array group in the first spectral image and at least one second corner point array group in the second spectral image according to the observability of each corner point array, wherein the corner point array group includes a plurality of corner point arrays belonging to the same soil type; An array group comparison module, used to compare the target first corner point array group with each of the second corner point array groups, respectively, to obtain a target second corner point array group that meets a preset similarity condition with the target first corner point array group; A distance correction module is used to correct the array distance between the corner point arrays of the first target corner point array group and the second target corner point array group according to the first target corner point array group and the second target corner point array group, so as to obtain a first corrected spectral image and a second corrected spectral image; An image stitching module is used to stitch the first corrected spectral image with the second corrected spectral image to obtain a target soil area image.

Images