Vegetation distribution prediction method and system based on remote sensing image

Abstract

The invention relates to the technical field of remote sensing image processing, in particular to a vegetation distribution prediction method and system based on a remote sensing image. According to the method, local analysis is carried out on a remote sensing image, and the remote sensing image composed of a plurality of pixel blocks is obtained. In consideration of different growth characteristics of arbor plants and herbaceous plants, herbaceous plant pixel blocks and arbor plant pixel blocks can be screened based on the vegetation diffusion degree obtained. And performing prediction of different rules according to the pixel blocks of different plant types. And selecting a reference pixel point from the neighborhood of the pixel point in the arbor plant pixel block, and predicting by taking the change of the reference pixel point in time sequence as the reference of the original pixel point to obtain a predicted remote sensing image. According to the method, the plant types are identified and predicted respectively, prediction errors are eliminated, and the accurate prediction remote sensing image is obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing, and in particular to a vegetation distribution prediction method and system based on remote sensing images. Background Art

[0002] Remote sensing images can effectively analyze the plant growth characteristics of vegetation-covered areas, and prediction algorithms can be used to predict the distribution and changes of vegetation, thereby facilitating technicians in fields such as agriculture or forestry to analyze the current area.

[0003] In the prior art, in order to obtain the growth trend of vegetation, the ARIMA prediction model can be used in combination with remote sensing images at different time series for analysis, and the time series changes of pixel values ​​in different regions can be used as sample data for prediction, and the vegetation development results in the future time series can be obtained according to the change characteristics of the data. However, for vegetation-covered areas, there are obvious differences in vegetation texture between tree plants and herbaceous plants, and the vegetation growth rate is quite different. This is mainly manifested in the spatial heterogeneity of tree-type vegetation, and there are obvious gaps between vegetation. In the process of vegetation growth and diffusion, gaps and other targets are needed, resulting in unstable changes in some pixel values. Therefore, if a global prediction of the entire remote sensing image is performed based only on the changes in the time series, there will be obvious errors in the prediction due to the inconsistency of vegetation growth and diffusion. Summary of the invention

[0004] In order to solve the technical problem that the existing technology does not take into account the uneven vegetation diffusion between herbaceous plants and tree plants, resulting in prediction errors, the purpose of the present invention is to provide a vegetation distribution prediction method and system based on remote sensing images. The technical solution adopted is as follows:

[0005] The present invention proposes a vegetation distribution prediction method based on remote sensing images, the method comprising:

[0006] Obtaining multiple frames of continuous remote sensing images of the vegetation area to be predicted; the remote sensing images are composed of multiple pixel blocks; the pixel blocks of different remote sensing images have the same distribution and size;

[0007] For each pixel block at each position, the vegetation diffusion degree of the pixel block at each position is obtained according to the change characteristics of the pixel distribution uniformity of the pixel block in the remote sensing images of multiple frames and the color characteristics within the pixel block;

[0008] According to the vegetation diffusion degree, herbaceous plant pixel blocks and tree plant pixel blocks in the remote sensing image are screened out; the pixel points in the herbaceous plant pixel blocks are predicted using the ARIMA algorithm; for the pixel points in the tree plant pixel blocks, reference pixel points are selected in the neighborhood of each pixel point according to the vegetation diffusion degree of the tree plant pixel blocks, and the pixel value change results of the reference pixel points in time series are weighted and fused to obtain the reference time series change characteristics of the corresponding pixel points; the pixel points in the tree plant pixel blocks are predicted using the ARIMA algorithm according to the reference time series change characteristics; and the predicted remote sensing image is obtained.

[0009] Furthermore, after obtaining the remote sensing image, the method further includes:

[0010] The weight of the green channel is increased in the traditional weighted average grayscale algorithm, and the remote sensing image is grayscaled.

[0011] Furthermore, the pixel block division method includes:

[0012] In the multiple consecutive frames of remote sensing images, superpixel segmentation is performed on the first frame of remote sensing images to obtain pixel blocks in the first frame of remote sensing images, and the same segmentation is performed on other frames of remote sensing images according to the size and distribution of the pixel blocks in the first frame of remote sensing images to obtain pixel blocks in each frame of remote sensing images.

[0013] Furthermore, the average grayscale value of the pixel block is used as the color feature.

[0014] Furthermore, the method for obtaining the uniformity of pixel distribution includes:

[0015] In each frame of the remote sensing image, the entropy value and variance of the pixel values ​​in each pixel block are obtained, and the product of the entropy value and the variance is negatively correlated mapped to obtain the pixel distribution uniformity of each pixel block.

[0016] Furthermore, the method for obtaining the vegetation diffusion degree includes:

[0017] Take any pixel block position as the target position and any frame of remote sensing image as the target remote sensing image;

[0018] Obtaining a pixel distribution uniformity difference of a pixel block at a target position between the target remote sensing image and a next frame of remote sensing image; after negative correlation mapping the pixel distribution uniformity difference, multiplying it with the color feature and pixel distribution uniformity of the pixel block at the target position in the target remote sensing image to obtain an initial diffusion degree of the pixel block at the target position in the target remote sensing image;

[0019] The average value of the initial diffusion degrees of the pixel blocks at the target position in all frames of remote sensing images is used as the vegetation diffusion degree of the pixel blocks at the target position.

[0020] Furthermore, the method for obtaining the reference pixel includes:

[0021] Arrange the tree plant pixel blocks in ascending order according to the degree of vegetation diffusion to obtain the order of each tree plant pixel block, set the number of reference pixel points of the first tree plant pixel block to a preset value, and use the preset value as the initial value to reduce it in sequence according to a preset step size to obtain the number of reference pixel points of the tree plant pixel blocks of each order; based on the number of reference pixel points, take each pixel point as the center, obtain the reference pixel point of each pixel point in the horizontal and vertical directions.

[0022] Furthermore, the method for acquiring the reference time series variation feature includes:

[0023] Gaussian weights are used to perform weighted summation on the pixel value change results of the reference pixel points in time series to obtain the reference time series change characteristics.

[0024] Furthermore, the step of screening out the herbaceous plant pixel blocks and the tree plant pixel blocks in the remote sensing image according to the vegetation diffusion degree includes:

[0025] In the remote sensing image, if the vegetation diffusion degree of a pixel block is greater than a preset diffusion degree threshold, the pixel block is a herbaceous plant pixel block; otherwise, it is a tree plant pixel block.

[0026] The present invention also proposes a vegetation distribution prediction system based on remote sensing images, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the processor implements any one of the steps of the vegetation distribution prediction method based on remote sensing images.

[0027] The present invention has the following beneficial effects:

[0028] The embodiment of the present invention takes into account that tree plants and herbaceous plants have obvious differences in image features such as texture and color. In order to effectively distinguish the two types of preparations, it is first necessary to perform local analysis on the remote sensing image, so as to obtain a remote sensing image composed of multiple pixel blocks, and the pixel block represents a local distribution area of ​​a type of vegetation in the remote sensing image. Further considering that tree plants and herbaceous plants have different growth characteristics, the texture and change rules of pixels at different time sequences will be different. Based on this, the degree of vegetation diffusion can be obtained to screen out herbaceous plant pixel blocks and tree plant pixel blocks. Different rules can be predicted according to pixel blocks of different plant types. Because tree plants have growth characteristics such as crown and trunk height, vegetation has spatial specificity. Therefore, in order to prevent prediction errors, reference pixels are selected in the neighborhood of pixel points in the tree plant pixel block, and then the changes in the reference pixels in the time sequence are used as the reference of the original pixels for prediction, eliminating the errors caused by spatial specificity, and obtaining accurate predicted remote sensing images. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] 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.

[0030] Figure 1 A flow chart of a vegetation distribution prediction method based on remote sensing images provided by an embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a remote sensing image provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] 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 vegetation distribution prediction method and system based on remote sensing images proposed by the present invention, its specific implementation method, structure, features and effects, in combination 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.

[0033] 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.

[0034] The scenario targeted by the embodiment of the present invention is to use remote sensing technology to predict and analyze vegetation growth in a region. The collected remote sensing image is an RGB image, and the prediction algorithm used is the ARIMA prediction algorithm. The ARIMA prediction algorithm is a technical means well known to those skilled in the art, and the process of the algorithm is only briefly described here:

[0035] 1. Data cleaning and preprocessing: ensure data quality and handle missing values ​​and outliers.

[0036] 2. Stationarity test: Use methods such as the ADF test to ensure that the time series data is stationary or can be made stationary through differencing.

[0037] 3. Model parameter estimation: Determine the order of AR and MA terms by analyzing the autocorrelation function (ACF) and partial autocorrelation function (PACF) graphs. According to the identification rules of time series models, establish the corresponding model. If the stationary sequence's correlation function is truncated and the autocorrelation function is tailing, it can be concluded that the sequence is suitable for the AR model. If the stationary sequence's partial correlation function is tailing and the autocorrelation cluster number is truncated, it can be concluded that the sequence is suitable for the MA model; if the stationary sequence's partial correlation function and autocorrelation function are both tailing, the sequence is suitable for the ARMA model.

[0038] 4. Model fitting and training: Fit the ARIMA model using the determined parameter values.

[0039] 5. Model testing and validation: Evaluate the performance of the model by calculating the prediction error of the model.

[0040] The following is a detailed description of a vegetation distribution prediction method and system based on remote sensing images provided by the present invention in conjunction with the accompanying drawings.

[0041] See also Figure 1 , which shows a flow chart of a vegetation distribution prediction method based on remote sensing images provided by an embodiment of the present invention, the method comprising:

[0042] Step S101: obtaining a plurality of consecutive frames of remote sensing images of the vegetation area to be predicted; the remote sensing image is composed of a plurality of pixel blocks; the pixel blocks of different remote sensing images have the same distribution and size.

[0043] In order to effectively predict vegetation growth, it is necessary to analyze the changes in vegetation areas at different time sequences within the region, so it is necessary to obtain multiple frames of continuous remote sensing images of the vegetation area to be predicted. In one embodiment of the present invention, the prediction period of the vegetation area to be predicted is 15 days, and 20 frames of remote sensing images need to be continuously collected for prediction. Figure 2 , which shows a schematic diagram of a remote sensing image provided by an embodiment of the present invention.

[0044] Preferably, in one embodiment of the present invention, because the present invention is intended to analyze the growth of vegetation areas, vegetation areas have obvious green features, and the image is usually converted into a grayscale image during image processing. Therefore, when graying the remote sensing image, the weight of the green channel is increased in the traditional weighted average graying algorithm, and the remote sensing image is grayed. It should be noted that the weights of the three channels of red, green and blue in the traditional weighted average graying algorithm are 0.3, 0.59, and 0.11, respectively, and the adjusted weights in one embodiment of the present invention are 0.18, 0.72, and 0.1, respectively.

[0045] In some embodiments of the present invention, in order to ensure the accuracy of subsequent image processing, the remote sensing image may be processed using image enhancement algorithms such as histogram equalization to improve image quality.

[0046] Because of the large crowns of trees, there are many gaps in the crowns, and the high trunks of trees will cause obvious shadows in the image; while herbaceous plants show uniform texture characteristics in the image, and because the height and coverage density between herbaceous plants are relatively consistent, the pixel values ​​of herbaceous plants are brighter and more vivid than those of trees. That is, the two types of vegetation have obvious pixel differences. In order to accurately distinguish the plant types, we first obtain multiple pixel blocks in the remote sensing image, that is, each pixel block is composed of one type of pixels, and accurate classification results can be obtained by locally analyzing the image information in the pixel blocks. In order to effectively analyze the pixel blocks in time sequence in the subsequent analysis process, it is necessary to ensure that the distribution and size of the pixel blocks in different remote sensing images are the same.

[0047] Preferably, in one embodiment of the present invention, the pixel block division method includes:

[0048] In a remote sensing image of multiple consecutive frames, superpixel segmentation is performed on the first frame of remote sensing image to obtain pixel blocks in the first frame of remote sensing image, that is, the pixel points in the pixel blocks all have pixel values ​​with the same characteristics. The same segmentation is performed on other frames of remote sensing images according to the size and distribution of the pixel blocks in the first frame of remote sensing image to obtain pixel blocks in each frame of remote sensing image. That is, the embodiment of the present invention performs pixel block segmentation based on the first frame of remote sensing image, thereby ensuring that the division structure of pixel blocks in different frames of remote sensing images is unified. In the embodiment of the present invention, superpixel segmentation selects the SLIC superpixel segmentation method, and the specific algorithm is a technical means well known to those skilled in the art, which will not be described in detail here.

[0049] Step S102: for each pixel block at each position, the vegetation diffusion degree of the pixel block at each position is obtained according to the variation characteristics of the pixel distribution uniformity of the pixel block in multiple frames of remote sensing images and the color characteristics within the pixel block.

[0050] For tree plants, because of their obvious spatial specificity, the gaps between the crowns lead to obvious texture features in the tree area, and as the tree grows, the gaps may be filled, that is, the tree plant area at different time sequences has obvious texture change features; while the pixel information of herbaceous plants is more uniform and the texture information is less than that of tree plants, and after growth, there will be no obvious texture change features between different time sequences. Therefore, in order to effectively distinguish between tree plants and herbaceous plants, the change characteristics of the pixel distribution uniformity of the pixel block in multiple frames of remote sensing images are important analysis factors. Further considering that tree plants will present obviously darker image features due to occlusion, shadows and other factors, while herbaceous plants are brighter and brighter than tree plants, the vegetation diffusion degree of the pixel block at each position can be further combined with the color features in the pixel block. That is, the smaller the change characteristics of the pixel distribution uniformity and the brighter the color characteristics, the greater the vegetation diffusion degree, and the more likely the pixel block is a herbaceous plant pixel block.

[0051] Preferably, in one embodiment of the present invention, the average grayscale value in the pixel block is used as the color feature, that is, the larger the average grayscale value is, the brighter the color feature is.

[0052] In other embodiments of the present invention, the mode of the grayscale values ​​in the pixel block may be selected as the color feature, or the average value of the green channel value may be selected as the color feature, which will not be described in detail here.

[0053] Preferably, in one embodiment of the present invention, a method for acquiring pixel distribution uniformity includes:

[0054] In each frame of remote sensing image, the entropy value and variance of the pixel values ​​in each pixel block are obtained. The larger the entropy value and variance, the more chaotic the pixel value categories in the pixel block. The product of the entropy value and the variance is negatively correlated to obtain the uniformity of pixel distribution in each pixel block.

[0055] As an example, in one embodiment of the present invention, the product of the entropy value and the variance is negatively correlated and mapped to obtain the pixel distribution uniformity, which is expressed as:

[0056] Where D n,t is the pixel distribution uniformity of the n-th pixel block in the t-th frame remote sensing image, H(n,t) is the entropy value of the n-th pixel block in the t-th frame remote sensing image, and δ(n,t) is the variance of the n-th pixel block in the t-th frame remote sensing image.

[0057] The pixel distribution uniformity uses the inverse form to achieve negative correlation mapping, and the denominator is added with a positive integer 1 to prevent the denominator from being 0. It should be noted that the calculation method of the entropy value and the calculation method of the variance are technical means well known to those skilled in the art, and will not be described in detail here.

[0058] Preferably, in one embodiment of the present invention, the method for obtaining the vegetation diffusion degree includes:

[0059] Any pixel block position is taken as the target position, and any frame of remote sensing image is taken as the target remote sensing image.

[0060] The pixel distribution uniformity difference of the pixel block at the target position between the target remote sensing image and the next frame remote sensing image is obtained.

[0061] The greater the difference in pixel distribution uniformity, the more likely the pixel block is to be a tree plant pixel block; the smaller the difference in pixel distribution uniformity, the greater the vegetation diffusion of the pixel block, and the more likely it is to be a herbaceous plant pixel block. Further considering that the more obvious the color feature, the more likely it is to be a herbaceous plant pixel block, and the greater the pixel distribution uniformity, the more uniform the texture information in the pixel block, the more likely it is to be a herbaceous plant pixel block. Therefore, after the pixel distribution uniformity difference is negatively correlated and mapped, it is multiplied by the color feature and pixel distribution uniformity of the pixel block at the target position in the target remote sensing image to obtain the initial diffusion degree of the pixel block at the target position in the target remote sensing image. As an example, in one embodiment of the present invention, the initial diffusion degree is expressed by the formula: Among them A n,t Indicates the initial diffusion degree of the pixel block at the nth position in the tth frame remote sensing image, represents the average gray value of the pixel block at the nth position in the tth frame remote sensing image, D n,t Indicates the pixel distribution uniformity of the pixel block at the nth position in the tth frame remote sensing image, ΔD n,t Represents the pixel distribution uniformity difference of the pixel block at the nth position in the tth frame remote sensing image.

[0062] That is, in the initial diffusion degree formula, the pixel distribution uniformity difference is negatively mapped in the reciprocal form, and in order to prevent the denominator from being 0, a positive integer 1 is added to the denominator.

[0063] Because each pixel block at each position corresponds to a vegetation diffusion degree in multiple frames of remote sensing images, it is necessary to combine the vegetation diffusion degrees of all frames of remote sensing images to obtain the final result. The average value of the initial diffusion degree of the pixel block at the target position in all frames of remote sensing images is taken as the vegetation diffusion degree of the pixel block at the target position.

[0064] Step S103: Filter out herbaceous plant pixel blocks and tree plant pixel blocks in the remote sensing image according to the degree of vegetation diffusion; use the ARIMA algorithm to predict the pixel points in the herbaceous plant pixel blocks; for the pixel points in the tree plant pixel blocks, select reference pixel points in the neighborhood of each pixel point according to the degree of vegetation diffusion of the tree plant pixel blocks, perform weighted fusion on the pixel value change results of the reference pixel points in time series, and obtain the reference time series change characteristics of the corresponding pixel points; use the ARIMA algorithm to predict the pixel points in the tree plant pixel blocks according to the reference time series change characteristics; and obtain a predicted remote sensing image.

[0065] The vegetation growth characteristics of tree plants and herbaceous plants are different, so the extracted vegetation diffusion degrees are also quite different. The vegetation diffusion degree of herbaceous plant pixel blocks is significantly greater than that of tree plant pixel blocks. Therefore, the herbaceous plant pixel blocks and tree plant pixel blocks in the remote sensing image can be screened out according to the vegetation diffusion degree.

[0066] Preferably, in one embodiment of the present invention, the herbaceous plant pixel blocks and the tree plant pixel blocks in the remote sensing image are screened out according to the vegetation diffusion degree, including:

[0067] In the remote sensing image, if the vegetation diffusion degree of a pixel block is greater than a preset diffusion degree threshold, the pixel block is a herbaceous plant pixel block; otherwise, it is a tree plant pixel block. In the embodiment of the present invention, after the vegetation diffusion degree is normalized, the diffusion degree threshold is set to 0.9.

[0068] Because the growth of herbaceous plants changes steadily, the ARIMA algorithm can be directly used to predict the pixels in the herbaceous plant pixel block. However, due to the certain spatial heterogeneity of arbor plants, the volatility of pixel changes in time series is high, which does not meet the ARIMA model's requirements for data stability, so direct prediction will have large errors. Therefore, the embodiment of the present invention is based on the idea of ​​data smoothing. For the pixels in the arbor plant pixel block, reference pixels are selected in the neighborhood of each pixel according to the vegetation diffusion degree of the arbor plant pixel block, that is, each pixel in the arbor plant pixel block corresponds to a group of reference pixels, and the number of reference pixels is related to the vegetation diffusion degree. The smaller the vegetation diffusion degree, the more unstable the pixels in the arbor plant pixel block, and the more reference pixels are needed.

[0069] Preferably, in one embodiment of the present invention, the method for obtaining reference pixels includes:

[0070] Arrange the tree and plant pixel blocks in ascending order according to the degree of vegetation diffusion to obtain the order of each tree and plant pixel block, set the number of reference pixel points of the first tree and plant pixel block to a preset value, and use the preset value as the initial value to reduce it in sequence according to the preset step length to obtain the number of reference pixel points of the tree and plant pixel blocks of each order. In one embodiment of the present invention, the preset value is set to 20, and the preset step length is set to 1. It should be noted that if it is reduced to 1 in sequence according to the preset step length, it will no longer be reduced, and the number of reference pixel points of the pixels in the subsequent tree and plant pixel blocks will all be set to 1.

[0071] According to the number of reference pixels, each pixel is taken as the center to obtain the reference pixel of each pixel in the horizontal and vertical directions. For example, if the number of reference pixels of a certain pixel is 20, then the pixel is taken as the center, and 20 neighboring pixels are found on the left side and 20 neighboring pixels are found on the right side in the horizontal and vertical directions as reference pixels.

[0072] After obtaining the reference pixel points, for each pixel point on the tree plant pixel block, the pixel value change results of the reference pixel point in time series can be weighted and fused to obtain the reference time series change characteristics of the corresponding pixel point. For example, the pixel values ​​of each reference pixel point in time series are combined into a sequence, and the sequence lengths of different reference pixels are the same. For the elements at the same position in the sequence, the elements at the same position of all sequences are weighted and fused to form a new sequence, which can characterize the reference time series change characteristics of the corresponding pixel point. According to the reference time series change characteristics, the ARIMA algorithm can be used to predict the pixels in the tree plant pixel block. In this way, the two types of plants are predicted separately, and the predicted remote sensing images are obtained. The vegetation growth trend in the current area can be analyzed based on the predicted remote sensing images.

[0073] Preferably, in one embodiment of the present invention, the pixel value change results of the reference pixel points in time series are weighted and summed using Gaussian weights to obtain the reference time series change characteristics. It should be noted that Gaussian weights are a technical means well known to those skilled in the art, that is, the closer the pixel point is to the center pixel point, the greater the weight. Specifically, the weight of each reference pixel point can be obtained according to the existing formula, which will not be repeated here.

[0074] In summary, the embodiment of the present invention performs local analysis on the remote sensing image to obtain a remote sensing image composed of multiple pixel blocks. Taking into account the differences in growth characteristics between tree plants and herbaceous plants, the herbaceous plant pixel blocks and tree plant pixel blocks can be screened out based on the vegetation diffusion degree. Predictions are made according to different rules for pixel blocks of different plant types. Reference pixels are selected in the neighborhood of the pixel points in the tree plant pixel block, and then the changes in the reference pixels in time series are used as a reference for the original pixels to perform predictions to obtain a predicted remote sensing image. The present invention eliminates prediction errors by identifying plant types and making predictions separately, thereby obtaining an accurate predicted remote sensing image.

[0075] The present invention also proposes a vegetation distribution prediction system based on remote sensing images, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any step of a vegetation distribution prediction method based on remote sensing images.

[0076] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A vegetation distribution prediction method based on remote sensing images, characterized in that: The method comprises: Obtaining multiple frames of continuous remote sensing images of the vegetation area to be predicted; the remote sensing images are composed of multiple pixel blocks; the pixel blocks of different remote sensing images have the same distribution and size; For each pixel block at each position, the vegetation diffusion degree of the pixel block at each position is obtained according to the change characteristics of the pixel distribution uniformity of the pixel block in the remote sensing images of multiple frames and the color characteristics within the pixel block; According to the vegetation diffusion degree, herbaceous plant pixel blocks and tree plant pixel blocks in the remote sensing image are screened out; the pixel points in the herbaceous plant pixel blocks are predicted using the ARIMA algorithm; for the pixel points in the tree plant pixel blocks, reference pixel points are selected in the neighborhood of each pixel point according to the vegetation diffusion degree of the tree plant pixel blocks, and the pixel value change results of the reference pixel points in time series are weighted and fused to obtain the reference time series change characteristics of the corresponding pixel points; the pixel points in the tree plant pixel blocks are predicted using the ARIMA algorithm according to the reference time series change characteristics; and the predicted remote sensing image is obtained.

2. The vegetation distribution prediction method based on remote sensing images according to claim 1 is characterized in that: After obtaining the remote sensing image, the method further includes: The weight of the green channel is increased in the traditional weighted average grayscale algorithm, and the remote sensing image is grayscaled.

3. The vegetation distribution prediction method based on remote sensing images according to claim 1 is characterized in that: The pixel block division method includes: In the multiple consecutive frames of remote sensing images, superpixel segmentation is performed on the first frame of remote sensing images to obtain pixel blocks in the first frame of remote sensing images, and the same segmentation is performed on other frames of remote sensing images according to the size and distribution of the pixel blocks in the first frame of remote sensing images to obtain pixel blocks in each frame of remote sensing images.

4. The vegetation distribution prediction method based on remote sensing images according to claim 1, characterized in that: The average grayscale value of the pixel block is used as the color feature.

5. The vegetation distribution prediction method based on remote sensing images according to claim 1 is characterized in that: The method for obtaining pixel distribution uniformity comprises: In each frame of the remote sensing image, the entropy value and variance of the pixel values ​​in each pixel block are obtained, and the product of the entropy value and the variance is negatively correlated mapped to obtain the pixel distribution uniformity of each pixel block.

6. The vegetation distribution prediction method based on remote sensing images according to claim 4 is characterized in that: The method for obtaining the vegetation diffusion degree includes: Take any pixel block position as the target position and any frame of remote sensing image as the target remote sensing image; Obtaining a pixel distribution uniformity difference of a pixel block at a target position between the target remote sensing image and a next frame of remote sensing image; after negative correlation mapping the pixel distribution uniformity difference, multiplying it with the color feature and pixel distribution uniformity of the pixel block at the target position in the target remote sensing image to obtain an initial diffusion degree of the pixel block at the target position in the target remote sensing image; The average value of the initial diffusion degrees of the pixel blocks at the target position in all frames of remote sensing images is used as the vegetation diffusion degree of the pixel blocks at the target position.

7. The vegetation distribution prediction method based on remote sensing images according to claim 1, characterized in that: The method for obtaining the reference pixel point includes: Arrange the tree plant pixel blocks in ascending order according to the degree of vegetation diffusion to obtain the order of each tree plant pixel block, set the number of reference pixel points of the first tree plant pixel block to a preset value, and use the preset value as the initial value to reduce it in sequence according to a preset step size to obtain the number of reference pixel points of the tree plant pixel blocks of each order; based on the number of reference pixel points, take each pixel point as the center, obtain the reference pixel point of each pixel point in the horizontal and vertical directions.

8. The vegetation distribution prediction method based on remote sensing images according to claim 1, characterized in that: The method for acquiring the reference time series variation feature includes: Gaussian weights are used to perform weighted summation on the pixel value change results of the reference pixel points in time series to obtain the reference time series change characteristics.

9. The vegetation distribution prediction method based on remote sensing images according to claim 1, characterized in that: The step of selecting herbaceous plant pixel blocks and tree plant pixel blocks in the remote sensing image according to the vegetation diffusion degree includes: In the remote sensing image, if the vegetation diffusion degree of a pixel block is greater than a preset diffusion degree threshold, the pixel block is a herbaceous plant pixel block; otherwise, it is a tree plant pixel block.

10. A vegetation distribution prediction system based on remote sensing images, 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, the steps of the vegetation distribution prediction method based on remote sensing images as claimed in any one of claims 1 to 9 are implemented.

Images