A land consolidation boundary line division method and system based on machine vision

Abstract

The present application relates to the technical field of remote sensing image processing, in particular to a land consolidation boundary line division method and system based on machine vision. The present application obtains a plurality of superpixel blocks of a satellite remote sensing image; according to the elevation information distribution of different pixel points in each superpixel block in a digital elevation model, the terrain height variation degree of each superpixel block is obtained; according to the elevation information difference between each pixel point in each superpixel block and other pixel points in the corresponding neighborhood range in the digital elevation model, and the gradient distribution of the corresponding pixel points in different directions in the digital elevation model, the overall water flow direction value of each superpixel block is obtained; and then the merging possibility between different superpixel blocks is obtained; the superpixel optimal block is obtained, and the consolidation boundary line of the land region to be divided is divided. The present application accurately obtains the superpixel segmentation process of the land region to be divided, and improves the accuracy of the consolidation boundary line division.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image processing, and particularly relates to a land consolidation boundary line division method and system based on machine vision. BACKGROUND

[0002] By clearly dividing the land consolidation boundary line of the region, the spatial layout of cultivated land and other types of land in the mountainous area can be reasonably arranged, the range of different purpose lands can be clearly defined, the quality of cultivated land can be improved, the effective cultivated land area can be increased, and the agricultural ecological conditions and ecological environment can be improved; therefore, in order to realize the reasonable utilization and sustainable development of land resources, the land needs to be consolidated and optimally allocated.

[0003] In the prior art, the superpixel segmentation technology is used to divide the satellite remote sensing image of the mountainous area land, and the superpixel block is combined and adjusted to adjust the form and size of the land; however, due to the complex mountainous terrain, large slope change and undulating change, the land slope and land suitability are not considered by only analyzing the satellite remote sensing image, resulting in a large error in the division of the consolidation boundary. SUMMARY

[0004] In order to solve the technical problem that the land slope and land suitability are not considered by only analyzing the satellite remote sensing image, resulting in a large error in the division of the consolidation boundary, the purpose of the present application is to provide a land consolidation boundary line division method and system based on machine vision, and the technical solution is as follows:

[0005] The present application provides a land consolidation boundary line division method based on machine vision, which comprises the following steps:

[0006] Obtaining the satellite remote sensing image, the spectral image and the digital elevation model of the land region to be divided;

[0007] According to the spectral characteristics of each pixel point on the spectral image and the color characteristics on the satellite remote sensing image, a plurality of superpixel blocks of the satellite remote sensing image are obtained;

[0008] According to the elevation information distribution of different pixel points in each superpixel block in the digital elevation model, the terrain height variation degree of each superpixel block is obtained; according to the elevation information difference between each pixel point in each superpixel block and other pixel points in the corresponding neighborhood range in the digital elevation model, the local water flow direction value of each pixel point in each superpixel block is obtained; according to the local water flow direction value of different pixel points in each superpixel block and the gradient distribution of the corresponding pixel points in different directions in the digital elevation model, the overall water flow direction value of each superpixel block is obtained;

[0009] According to the terrain height variation degree and the overall water flow direction value between different superpixel blocks, a merging possibility between different superpixel blocks is obtained;

[0010] According to the merging possibility between all superpixel blocks, an optimal superpixel block is obtained, and a boundary line of a farmland region to be divided is arranged.

[0011] Further, the superpixel block obtaining method comprises:

[0012] According to the spectral feature of each pixel point in the spectral image and the color feature of each pixel point on the satellite remote sensing image, a vegetation performance degree of each pixel point is obtained.

[0013] A plurality of superpixel initial blocks of the satellite remote sensing image with equal width are obtained by using a superpixel segmentation algorithm.

[0014] A performance difference between each pixel point and a center pixel point in each superpixel initial block is obtained; if the performance difference is greater than a preset difference threshold, a relative distance between the corresponding pixel point and a center pixel point of another superpixel block is calculated, and another superpixel block corresponding to the minimum relative distance is selected as a new superpixel initial block of each pixel point, until the performance difference corresponding to all pixel points in each superpixel initial block is not greater than the preset difference threshold, and a plurality of superpixel blocks are obtained.

[0015] Further, the vegetation performance degree obtaining method comprises:

[0016] An NDVI vegetation coverage factor of each pixel point is obtained based on the spectral image; and color components of each pixel point in the remote sensing image in an RGB space are obtained.

[0017] A product of the NDVI vegetation coverage factor and the color components of each pixel point is obtained as the vegetation performance degree of each pixel point.

[0018] Further, the terrain height variation degree obtaining method comprises:

[0019] A difference mean value between a center pixel point and all other pixel points in each superpixel block in elevation information of a digital elevation model is obtained as a first height coefficient.

[0020] A difference mean value between different pixel points in each superpixel block in the elevation information of the digital elevation model is obtained as a second height coefficient.

[0021] A product between the first height coefficient and the second height coefficient in each superpixel block is obtained as the terrain height variation degree of each superpixel block.

[0022] Further, the local water flow direction value obtaining method comprises:

[0023] For any superpixel patch, the elevation information difference between each pixel point and each other pixel point in the neighborhood range in the digital elevation model is obtained;

[0024] The maximum value of the elevation information difference between each pixel point and all other pixel points in the neighborhood range is selected as the local water flow direction value of each pixel point.

[0025] Further, the method for obtaining the overall water flow direction value comprises:

[0026] The elevation gradient of each pixel point in the east-west direction in the digital elevation model is obtained as a first gradient, and the elevation gradient of each pixel point in the north-south direction in the digital elevation model is obtained as a second gradient;

[0027] The ratio of the second gradient to the first gradient is obtained, and a negative correlation mapping is performed, as the elevation slope between the pixel points; the arctangent function is used to calculate the elevation slope as the first direction coefficient of each pixel point;

[0028] The accumulation value of the product of the local water flow direction value of different pixel points and the first direction coefficient of the corresponding pixel point is obtained as the overall water flow direction value in each superpixel patch.

[0029] Further, the method for obtaining the merging possibility comprises:

[0030] The terrain height variation degree and the overall water flow direction value of each superpixel patch are constructed into a two-dimensional row vector;

[0031] The cosine similarity of the corresponding row vectors between each superpixel patch and each other superpixel patch is obtained; the relative distance between the center pixel points of each superpixel patch and each other superpixel patch is obtained;

[0032] The ratio between the cosine similarity and the relative distance between each superpixel patch and each other superpixel patch is obtained as the merging possibility between each superpixel patch and each other superpixel patch.

[0033] Further, the method for obtaining the superpixel optimal patch comprises:

[0034] Whether the superpixel patches need to be merged is determined according to the merging possibility between all superpixel patches; if it is determined that the superpixel patches need to be merged, new superpixel patches are obtained until it is determined that the superpixel patches do not need to be merged, and the superpixel optimal patch is obtained.

[0035] Further, the method for determining whether the superpixel patches need to be merged according to the merging possibility between all superpixel patches comprises:

[0036] If the merging possibility between all superpixel patches is greater than a preset merging threshold, it is judged that the corresponding superpixel patch will be merged.

[0037] The application further provides a land consolidation boundary line division system based on machine vision, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and when the processor executes the computer program, the steps of any one of the land consolidation boundary line division methods based on machine vision are implemented.

[0038] The application has the following beneficial effects:

[0039] The application obtains multiple superpixel patches of the satellite remote sensing image according to the spectral features of each pixel point on the spectral image and the color features on the satellite remote sensing image, reduces the complexity of subsequent processing, and retains local details of the image; obtains the terrain height variation degree of each superpixel patch according to the elevation information distribution of different pixel points in each superpixel patch on the digital elevation model, and reflects the terrain undulation in the patch; obtains the local flow direction value of each pixel point in each superpixel patch and the local hydrological features of the ground surface according to the elevation information difference between each pixel point in each superpixel patch and other pixel points in the corresponding neighborhood range on the digital elevation model; obtains the overall flow direction value of each superpixel patch by comprehensively considering the local flow direction value and the elevation gradient, and reflects the hydrological features and terrain features in the patch; obtains the merging possibility between different superpixel patches according to the terrain height variation degree and the overall flow direction value of different superpixel patches, determines which patches can be merged into a larger region, obtains the optimal superpixel patch according to the merging possibility between all superpixel patches, and divides the consolidation boundary line of the to-be-divided soil region. The application can accurately obtain the superpixel segmentation process of the to-be-divided soil region, and improve the accuracy of the consolidation boundary line division. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0041] Figure 1 A flow chart of a land consolidation boundary line division method based on machine vision provided by an embodiment of the present application;

[0042] Figure 2 A flow chart of a method for acquiring a merging possibility is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the following describes in detail the specific implementation, structure, features and effects of a land consolidation boundary line demarcation method and system based on machine vision according to the present application, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0044] Unless otherwise defined, 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 application belongs.

[0045] The specific scheme of the land consolidation boundary line demarcation method and system based on machine vision provided by the present application is described in detail below with reference to the accompanying drawings.

[0046] Please refer to Figure 1 which shows a method flow chart of a land consolidation boundary line demarcation method based on machine vision provided by an embodiment of the present application, which specifically includes:

[0047] Step S1: Obtain satellite remote sensing images, spectral images and digital elevation models of the land area to be demarcated.

[0048] In the embodiment of the present application, for single-source remote sensing images of complex mountainous terrain, superpixel segmentation only uses each component in the color space, and may not be able to accurately capture the local texture features of the image when facing complex terrain, especially terrain changes such as slope and vegetation coverage. Therefore, it is necessary to collect and analyze the feature information of multiple parts of the image. First, a satellite remote sensing device is used to obtain high-resolution optical images of the land area to be demarcated, i.e. satellite remote sensing images. A drone with a high-resolution camera and a laser radar device is selected to obtain standard three-dimensional point cloud data and spectral images of the land area to be demarcated. The three-dimensional point cloud data is processed by the inverse distance weighting method to generate a high-precision digital elevation model for analysis, which reflects the topographic features of the land area. Satellite remote sensing images, spectral images and digital elevation models of the land area to be demarcated are obtained.

[0049] It should be noted that in the embodiments of the present application, the satellite remote sensing image and the spectral image are subjected to noise removal, radiation correction and the like to ensure data quality, and the image data is unified into a grid format; the DEM data is taken as one layer, the spectral image is taken as another layer, and the geometric registration method is used to fuse the spatial data with the satellite remote sensing image of a corresponding size, which is helpful to analyze the local details such as spectral information and terrain information.

[0050] Step S2: obtaining a plurality of superpixel blocks of the satellite remote sensing image according to the spectral features of each pixel point on the spectral image and the color features on the satellite remote sensing image.

[0051] In the satellite remote sensing image, green represents relatively strong vegetation coverage, and terraces usually have artificial management of landform and relatively low vegetation coverage.

[0052] Preferably, in an embodiment of the present application, the method for obtaining the superpixel blocks comprises:

[0053] obtaining the vegetation performance degree of each pixel point according to the spectral features of each pixel point in the spectral image and the color features on the satellite remote sensing image;

[0054] Preferably, in an embodiment of the present application, the method for obtaining the vegetation performance degree comprises:

[0055] obtaining the NDVI vegetation coverage factor of each pixel point based on the spectral image; and obtaining the color component of each pixel point in the RGB space of the remote sensing image;

[0056] obtaining the product of the NDVI vegetation coverage factor and the color component of each pixel point as the vegetation performance degree of each pixel point.

[0057] It should be noted that in other embodiments of the present application, other basic mathematical operations such as addition can also be used to realize the positive correlation relationship, and the specific means are well known to those skilled in the art, which will not be described here.

[0058] obtaining a plurality of superpixel initial blocks of the satellite remote sensing image in equal width by using a superpixel segmentation algorithm;

[0059] obtaining the performance difference of the vegetation performance degree between each pixel point and the center pixel point in each superpixel initial block; if the performance difference is greater than a preset difference threshold, calculating the relative distance between the corresponding pixel point and the center pixel point of other superpixel blocks, selecting the other superpixel block corresponding to the minimum relative distance as the new superpixel initial block of each pixel point, until the performance difference corresponding to all pixel points in the superpixel initial blocks is not greater than the preset difference threshold, and obtaining a plurality of superpixel blocks.

[0060] Based on this, for the performance difference corresponding to a pixel point greater than the preset difference threshold, the superpixel initial block where the pixel point is located is updated, that is, a new superpixel initial block is obtained to analyze the performance difference corresponding to the pixel point, until the performance difference corresponding to the pixel point in all superpixel initial blocks is not greater than the preset difference threshold, the adjustment of the superpixel initial block where the pixel point is located is stopped, and the final superpixel block is obtained for analysis.

[0061] It should be noted that in the embodiments of the present application, the number of divisions is set to 1500, and the superpixel segmentation algorithm is used to obtain a plurality of size-equal-width superpixel initial blocks of the satellite remote sensing image, and the relative distance between the center pixel points of the superpixel initial blocks is S; wherein the relative distance between the pixel points can be obtained by calculating the Euclidean distance or Manhattan distance between the position coordinates of the pixel points, and the specific means is a technology familiar to those skilled in the art, which will not be repeated here.

[0062] It should be noted that in an embodiment of the present application, the size of the preset difference threshold is 0.75; in other embodiments of the present application, the size of the preset difference threshold is a technology familiar to those skilled in the art, which will not be limited and repeated here.

[0063] Step S3: obtaining the terrain height variation degree of each superpixel block according to the elevation information distribution of different pixel points in the digital elevation model in each superpixel block; obtaining the local water flow direction value of each pixel point in each superpixel block according to the elevation information difference between each pixel point and other pixel points in the corresponding neighborhood range in the digital elevation model; obtaining the overall water flow direction value of each superpixel block according to the local water flow direction value of different pixel points in each superpixel block and the gradient distribution of the corresponding pixel points in different directions in the digital elevation model.

[0064] Because the terrain of the mountainous area has large difference in fluctuation change, different regions are usually cultivated in the areas with relatively gentle change in altitude to prevent a large amount of water and soil loss in the process of cultivation, and there is relatively uniform change in height information between different plots, and the original vegetation is in the form of chaotic terrain height distribution, so the terrain height variation degree of each block region is analyzed; the terrain height variation degree of each superpixel block is obtained according to the elevation information distribution of different pixel points in the digital elevation model in each superpixel block.

[0065] Preferably, in an embodiment of the present application, the method for obtaining the terrain height variation degree comprises:

[0066] obtaining the average difference in elevation information between the center pixel point and all other pixel points in each superpixel block in the digital elevation model as a first height coefficient;

[0067] obtaining the mean value of the difference between the elevation information of different pixel points in each superpixel patch in the digital elevation model as a second height coefficient;

[0068] obtaining the product between the first height coefficient and the second height coefficient in each superpixel patch as a terrain height variation degree of each superpixel patch.

[0069] In an embodiment of the present application, the formula of the terrain height variation degree is:

[0070]

[0071] wherein h i represents the terrain height variation degree of the i-th superpixel patch; Z i represents the elevation information of the center pixel point in the i-th superpixel patch in the digital elevation model; Z i,j represents the elevation information of the j-th other pixel point in the i-th superpixel patch in the digital elevation model; represents the mean value of the difference between the elevation information of different pixel points in the i-th superpixel patch in the digital elevation model, i.e. the second height coefficient; n represents the number of pixel points in the i-th superpixel patch.

[0072] In the formula of the terrain height variation degree, represents the mean value of the difference between the elevation information of the center pixel point and all other pixel points in the i-th superpixel patch in the digital elevation model, i.e. the first height coefficient. The greater the difference in elevation information, the greater the degree of fluctuation between pixel points, and the greater the second height coefficient. The greater the difference in elevation information between different pixel points, the greater the degree of terrain change.

[0073] The water flow path determines the flow direction of water on the ground surface. In the design process of converting a slope into a terrace, the terrace is usually on a flat slope surface, and the area unsuitable for planting has more significant slope changes. If the water flow path is ignored when merging superpixel patches, it may cause poor water flow in the merged area, which cannot improve the production efficiency of agriculture, and even cause problems such as soil erosion. Therefore, according to the difference in elevation information between each pixel point and other pixel points in the corresponding neighborhood range in each superpixel patch in the digital elevation model, a local water flow direction value of each pixel point in each superpixel patch is obtained.

[0074] Preferably, in an embodiment of the present application, the method for obtaining the local water flow direction value comprises:

[0075] For any superpixel patch, the difference in elevation information between each pixel point and each other pixel point in the neighborhood range in the digital elevation model is obtained;

[0076] The maximum value of the height information difference between each pixel point and all other pixel points in the neighborhood range is selected as the local water flow direction value of each pixel point.

[0077] It should be noted that in an embodiment of the present application, the neighborhood range is a range formed by taking each pixel point as the center and the pixel points in the adjacent 8 directions, and in other embodiments of the present application, the size of the neighborhood range can be set according to specific conditions, which is not limited or described here.

[0078] The gradient can reflect the height change rate in different directions, and is used to describe the water flow direction value or the terrain feature. According to the local water flow direction value of each pixel point in each superpixel block and the gradient distribution of the corresponding pixel point in different directions in the digital elevation model, the overall water flow direction value of each superpixel block is obtained.

[0079] Preferably, in an embodiment of the present application, the method for obtaining the overall water flow direction value comprises:

[0080] The height gradient of each pixel point in the east-west direction in the digital elevation model is obtained as the first gradient, and the height gradient of each pixel point in the north-south direction in the digital elevation model is obtained as the second gradient.

[0081] The ratio of the second gradient to the first gradient is obtained, and a negative correlation mapping is performed, as the height slope between the pixel points. The height slope is calculated by using the arctangent function as the first direction coefficient of each pixel point.

[0082] The accumulated value of the product of the local water flow direction value of different pixel points and the first direction coefficient of the corresponding pixel point is obtained as the overall water flow direction value in each superpixel block.

[0083] In an embodiment of the present application, the formula of the overall water flow direction value is:

[0084]

[0085] Wherein, γ i represents the overall water flow direction value in the i-th superpixel block; γ i,j is the water flow direction value of the j-th pixel point in the i-th superpixel block; represents the height gradient of the j-th pixel point in the east-west direction in the digital elevation model; represents the height gradient of the j-th pixel point in the north-south direction in the digital elevation model; arctan() represents the arctangent function.

[0086] In the formula of the water flow direction value, The difference between the elevation gradients of each pixel point in different directions in the digital elevation model is represented, the elevation gradient of each pixel point in the east-west direction in the digital elevation model is obtained as a first gradient, the elevation gradient of each pixel point in the south-north direction in the digital elevation model is obtained as a second gradient, and the ratio of the second gradient to the first gradient is obtained and negatively correlated, that is, the elevation slope between the pixel points; The arctangent function is used to calculate the elevation slope as a first direction coefficient of each pixel point, which reflects the slope surface orientation of the pixel point. The larger the first direction coefficient, the larger the slope surface, the larger the local water flow direction value, and the larger the overall water flow direction value.

[0087] Step S4: Obtain the merging possibility between different superpixel blocks according to the terrain height variation degree and the overall water flow direction value between the different superpixel blocks.

[0088] Adjacent superpixel blocks may have similar terrain, hydrological or spectral features, but are divided into different blocks. The greater the similarity, the closer the distance, and the more they need to be merged.

[0089] Preferably, in an embodiment of the present application, the method for obtaining the merging possibility can refer to Figure 2 which shows a flow chart of a method for obtaining a merging possibility, comprising:

[0090] Step S201: Construct a two-dimensional row vector with the terrain height variation degree and the overall water flow direction value of each superpixel block.

[0091] The terrain height variation degree is an important parameter for describing the degree of terrain undulation, which can reflect the complexity and diversity of the terrain. The water flow path determines the flow direction of water on the ground. The combination of the water flow direction value parameter and the terrain height variation degree can construct a more complete terrain-water system feature description system, which helps to deeply understand the terrain and water flow characteristics of the region.

[0092] Step S202: Obtain the cosine similarity of the corresponding row vectors between each superpixel block and each other superpixel block; obtain the relative distance between the center pixel points of each superpixel block and each other superpixel block.

[0093] The cosine similarity measures the similarity of the row vectors between the superpixel blocks, that is, the similarity of the terrain and water flow of the soil region. The greater the cosine similarity, the more similar the terrain and water flow, and the more likely they are to be merged. Superpixel blocks with a relatively short distance are closer in space, so they have a higher possibility of merging. By analyzing the cosine similarity and the relative distance, it is helpful to merge the regions that are adjacent in space and similar in vision together, thereby maintaining the local coherence of the image.

[0094] It should be noted that in an embodiment of the present application, the relative distance can be obtained according to the existing distance calculation method such as the Euclidean distance or Manhattan distance between the calculated position coordinates, and the specific means is the technology familiar to those skilled in the art, which is not described here.

[0095] Step S203: obtaining the ratio between the cosine similarity and the relative distance between each superpixel patch and each other superpixel patch as the merging possibility between each superpixel patch and each other superpixel patch.

[0096] In an embodiment of the present application, the merging possibility is expressed as:

[0097]

[0098] wherein S i,w represents the merging possibility between the wth superpixel patch and the ith superpixel patch; D i represents the row vector corresponding to the ith superpixel patch; l i represents the center coordinate of the wth superpixel patch; D w represents the row vector corresponding to the wth superpixel patch; l w represents the two-dimensional coordinate in the wth superpixel patch; and cos() represents the cosine function.

[0099] In the formula of the merging possibility, cos(D i ,D w ) represents calculating the cosine similarity of the corresponding row vectors between the wth superpixel patch and the ith superpixel patch, and the greater the cosine similarity, the more similar the corresponding row vectors, the more consistent the terrain and water flow, and the more likely to be merged; the closer the relative distance, the more adjacent superpixel patches, and the greater the merging possibility.

[0100] Step S5: obtaining the superpixel optimal patch according to the merging possibility between all superpixel patches, and dividing the arrangement boundary line of the land area to be divided.

[0101] The merging possibility is obtained by analyzing the terrain height parameters and the overall water flow direction values between the patches to determine whether they should be merged into a larger area, and the greater the merging possibility, the more likely to be merged, which can better reflect the actual land use unit and terrain characteristics and better support the land consolidation and planning work.

[0102] Preferably, in an embodiment of the present application, the method for obtaining the superpixel optimal patch comprises:

[0103] According to the merging possibility between all superpixel blocks, it is judged whether the superpixel blocks need to be merged; if it is judged that the superpixel blocks need to be merged, new superpixel blocks are obtained until it is judged that the superpixel blocks do not need to be merged, and the optimal superpixel blocks are obtained.

[0104] Preferably, in one embodiment of the present application, the obtaining method of judging whether the superpixel blocks need to be merged according to the merging possibility between all superpixel blocks comprises:

[0105] If the merging possibility between all superpixel blocks is greater than a preset merging threshold, it is judged that the corresponding superpixel blocks will be merged.

[0106] It should be noted that, in one embodiment of the present application, the size of the preset merging threshold is 0.8; in other embodiments of the present application, the size of the preset merging threshold can be set according to specific conditions, which is not limited and described herein.

[0107] It should be noted that, in another embodiment of the present application, after the optimal superpixel blocks are obtained, the boundary line division of the land region to be divided comprises: the edge line of the optimal superpixel blocks is extracted, if there is an unclosed region, a polygon filling algorithm is used to detect the boundary of the unclosed region, the boundary line is automatically expanded to be closed, the topological consistency of the closed region is calculated to verify the integrity of the divided region, and finally a more accurate region boundary is obtained, the boundary line of the land region to be divided must satisfy the geometric integrity and logical correctness, including boundary closure, no self-intersection, no overlap, no hanging line and region integrity, and the specific means is a technology known to those skilled in the art, which is not described herein.

[0108] In summary, according to the spectral characteristics of each pixel point on the spectral image and the color characteristics on the satellite remote sensing image, the present application obtains a plurality of superpixel blocks of the satellite remote sensing image; according to the elevation information distribution of different pixel points in each superpixel block on the digital elevation model, the terrain height variation degree of each superpixel block is obtained; according to the elevation information difference between each pixel point in each superpixel block and other pixel points in the corresponding neighborhood range on the digital elevation model, and the gradient distribution of the corresponding pixel points in different directions on the digital elevation model, the overall water flow direction value of each superpixel block is obtained; and then the merging possibility between different superpixel blocks is obtained; the optimal superpixel blocks are obtained, and the boundary line division of the land region to be divided is obtained. The present application accurately obtains the superpixel segmentation process of the land region to be divided, and improves the accuracy of the boundary line division.

[0109] The application further provides a land consolidation boundary line division system based on machine vision, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor.

[0110] It should be noted that the above-mentioned embodiment sequence is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0111] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

Claims

1. A method for delineating land consolidation boundaries based on machine vision, characterized in that, The method includes: Acquire satellite remote sensing images, spectral images, and digital elevation models of the land area to be delineated; Based on the spectral features of each pixel in the spectral image and the color features in the satellite remote sensing image, multiple superpixel blocks of the satellite remote sensing image are obtained. Based on the distribution of elevation information of different pixels in the digital elevation model within each superpixel block, the terrain elevation variation of each superpixel block is obtained, including: obtaining the average difference in elevation information between the central pixel and all other pixels in the digital elevation model within each superpixel block as a first elevation coefficient; obtaining the average difference in elevation information between different pixels in the digital elevation model within each superpixel block as a second elevation coefficient; and obtaining the product between the first elevation coefficient and the second elevation coefficient within each superpixel block as the terrain elevation variation of each superpixel block. Based on the difference in elevation information of each pixel in each superpixel block with other pixels in the corresponding neighborhood range in the digital elevation model, the local water flow direction value of each pixel in each superpixel block is obtained. Based on the local water flow direction values ​​of different pixels within each superpixel block and the gradient distribution of the corresponding pixels in different directions in the digital elevation model, the overall water flow direction value of each superpixel block is obtained, including: obtaining the elevation gradient of each pixel in the east-west direction in the digital elevation model as the first gradient; obtaining the elevation gradient of each pixel in the north-south direction in the digital elevation model as the second gradient; obtaining the ratio of the second gradient to the first gradient and performing a negative correlation mapping as the elevation slope between pixels; calculating the elevation slope using the arctangent function as the first direction coefficient of each pixel; and obtaining the cumulative value of the product between the local water flow direction values ​​of different pixels and the first direction coefficient of the corresponding pixels as the overall water flow direction value within each superpixel block. Based on the terrain height variation and overall water flow direction values ​​between different superpixel blocks, the merging probability between different superpixel blocks is obtained; Based on the merging probability among all superpixel blocks, the optimal superpixel block is obtained, and the boundary line for the land area to be divided is delineated.

2. The land consolidation boundary delineation method based on machine vision according to claim 1, characterized in that, Methods for obtaining superpixel blocks include: The vegetation performance of each pixel is obtained based on the spectral features of each pixel in the spectral image and the color features of each pixel in the satellite remote sensing image. A superpixel segmentation algorithm is used to obtain multiple initial superpixel blocks of equal size and width in the satellite remote sensing image; The vegetation performance difference between each pixel and the center pixel in each superpixel initial block is obtained. If the performance difference is greater than a preset difference threshold, the relative distance between the corresponding pixel and the center pixel of other superpixel blocks is calculated. The other superpixel block with the smallest relative distance is selected as the new superpixel initial block for each pixel. This process continues until the performance difference of the pixels in all superpixel initial blocks is no greater than the preset difference threshold, thus obtaining multiple superpixel blocks.

3. The land consolidation boundary delineation method based on machine vision according to claim 2, characterized in that, Methods for obtaining vegetation performance include: The NDVI vegetation cover factor of each pixel is obtained from the spectral image; the color components of each pixel in the RGB space are obtained from the remote sensing image. The product of the NDVI vegetation cover factor and color component of each pixel is obtained as the vegetation representation of each pixel.

4. The land consolidation boundary delineation method based on machine vision according to claim 1, characterized in that, Methods for obtaining local water flow direction values ​​include: For any superpixel block, obtain the difference in elevation information between each pixel and every other pixel in the neighborhood in the digital elevation model; The maximum value of the elevation information difference between each pixel and all other pixels in its neighborhood is selected as the local water flow direction value for each pixel.

5. The land consolidation boundary delineation method based on machine vision according to claim 1, characterized in that, Methods for obtaining the possibility of merging include: The terrain height variation and overall water flow direction value of each superpixel block are used to construct a two-dimensional row vector; Obtain the cosine similarity of the corresponding row vectors between each superpixel block and each other superpixel block; obtain the relative distance between the center pixels of each superpixel block and each other superpixel block; The ratio between the cosine similarity and the relative distance between each superpixel block and each other superpixel block is obtained as the merging probability between each superpixel block and each other superpixel block.

6. The land consolidation boundary delineation method based on machine vision according to claim 1, characterized in that, Methods for obtaining the optimal superpixel block include: Determine whether a superpixel block needs to be merged based on the merging probability among all superpixel blocks; if it is determined that merging is necessary, obtain a new superpixel block, until it is determined that merging is not necessary, and obtain the optimal superpixel block.

7. The land consolidation boundary delineation method based on machine vision according to claim 6, characterized in that, Methods for determining whether a superpixel block needs to be merged based on the merging probability among all superpixel blocks include: If the probability of merging all superpixel blocks is greater than the preset merging threshold, then the corresponding superpixel blocks will be merged.

8. A land consolidation boundary delineation system based on machine vision, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a machine vision-based land consolidation boundary line delineation method as described in any one of claims 1-7.

Images