Complex background remote sensing image ground object intelligent interpretation method

Abstract

The invention discloses a complex background remote sensing image ground object intelligent interpretation method, and relates to the technical field of remote sensing image target interpretation, and the method comprises the steps: obtaining a remote sensing image ground object interpretation public data set, carrying out the preprocessing, and carrying out the dividing according to a proportion after the preprocessing, thereby obtaining a training set, a test set and a verification set; constructing a perceptron model, calculating a training data set, outputting a target ground feature complexity threshold value, selecting an improved UNet remote sensing ground feature interpretation model on the basis, performing training weight iteration solution on the selected interpretation type ground features in the public data set, and obtaining a remote sensing ground feature intelligent interpretation weight model; and importing the remote sensing ground feature intelligent interpretation weight model into the improved UNet remote sensing ground feature interpretation model, inputting a test set, and carrying out adaptive remote sensing ground feature intelligent interpretation. The problems that an existing UNet model is single in structure, complex in parameter and long in interpretation time are solved, and the application range, the interpretation speed and the flexibility degree of the UNet model under different threshold value conditions are increased.

Description

[0001] This application is a divisional application. The application number of the original application is 202210064145.2, the application date is January 20, 2022, and the name of the invention is “A method and system for intelligent interpretation of remote sensing images based on adaptive perception machine”. Technical Field

[0002] The present invention relates to the technical field of remote sensing image target interpretation, and in particular to a method for intelligent interpretation of remote sensing image objects with complex backgrounds. Background Art

[0003] Remote sensing technology is an important means of studying the global land cover status and changes. In recent years, deep learning technology has developed rapidly, which can complete the tasks of identifying, extracting, and inverting targets from massive data through iterative training. In theory, deep learning can extract specific targets from remote sensing image interpretation datasets and realize large-scale remote sensing intelligent interpretation.

[0004] At present, the most widely used model for intelligent interpretation of remote sensing objects using deep learning technology is the Fully Convolutional Networks (FCN) model proposed in 2015 and its improved model. However, the FCN model does not consider the difference between the foreground and background of the target object when modeling, which is obviously inconsistent with the interpretation of complex remote sensing objects in reality and has problems such as large amount of calculation, difficulty in model deployment, and long calculation time. In the same year, the UNet model was proposed, which made it possible to interpret complex remote sensing objects. The UNet model reduces data costs while ensuring extraction accuracy, but the interpretation speed of this method still needs to be improved. In recent years, improved models based on FCN and UNet have been applied to multiple fields such as remote sensing, mapping, image, and computer, and can solve certain tasks in a single field, but they do not consider the difference between the foreground and background of the target object, and fail to adaptively select the best scale model suitable for a specific target.

[0005] Therefore, in the intelligent interpretation of objects in remote sensing images with complex backgrounds, it is an urgent problem for technical personnel in this field to propose an intelligent interpretation method for objects in remote sensing images with complex backgrounds to solve the problems existing in the prior art. Summary of the invention

[0006] In view of this, the present invention provides a method for intelligent interpretation of objects in remote sensing images with complex backgrounds, which solves the problem that the traditional UNet model does not consider the difference between the foreground and background of the target object, and describes the interpretation of remote sensing objects as a fixed single task, which is inconsistent with reality.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A remote sensing image intelligent interpretation method based on an adaptive perceptron comprises the following steps:

[0009] S101: Acquire a public data set of remote sensing image interpretation, preprocess the public data set to obtain preprocessed image data, and divide the preprocessed image data into a training set, a test set, and a validation set according to a proportion;

[0010] S102: constructing a perceptron model, calculating the training data set, and outputting a target object complexity threshold;

[0011] S103: According to the complexity threshold of the target object, an improved UNet remote sensing object interpretation model with multiple scales and multiple network depths is selected to iteratively solve the training weights of the selected interpretation category objects in the public data set to obtain a remote sensing object intelligent interpretation weight model;

[0012] S104: Importing the remote sensing object intelligent interpretation weight model into the improved UNet remote sensing object interpretation model, inputting a test set, and performing adaptive remote sensing object intelligent interpretation.

[0013] Optionally, the preprocessing in S101 includes: histogram correction and band combination;

[0014] The specific content of the histogram correction is as follows:

[0015] Subtract the minimum brightness value of the pixel in each band from the brightness values ​​of the M×N pixels in each band, where the image width is M pixels and the height is N pixels. The calculation is as follows:

[0016]

[0017] In the formula, f ij is the pixel brightness value of this band, f min is the minimum brightness of the pixel in this band, f g is the brightness value of the image pixel after histogram correction;

[0018] The specific content of the band combination is: setting different band combination modes to generate false color images.

[0019] Optionally, the step of constructing a perceptron model in S102 includes the following steps:

[0020] S1021: The preprocessed public dataset is stacked to form a linearly separable perceptron dataset. The formula is as follows:

[0021] T={(x 1 ,y 1 ),(x 2 ,y 2),...,(x n ,y n )} (2)

[0022] In the formula, x i ∈X=R n ,y i ∈Y=R n , i=1,2,...,N,x i ,y i All are on the maximum margin hyperplane ω * x+b * = 0; X is an arbitrary axis of the two-dimensional feature hyperplane, Y is an axis orthogonal to X in the two-dimensional feature hyperplane, ω * is the slope of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system, b * is the intercept of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system and the Z axis, n is the n-dimensional feature space, and N is an integer set.

[0023] S1022: Define the perceptron loss function as the sum of the distances from the misclassified points to the feature classification hyperplane, as shown in the following formula:

[0024]

[0025] Among them, the output result of the perceptron of the misclassified point has the opposite sign to the true result, as shown in the following formula:

[0026]

[0027] The adaptive perceptron model expression of the fusion loss function is shown as follows:

[0028]

[0029] In formulas (3)-(5), M is the set of misclassified points.

[0030] Optionally, the improved UNet remote sensing object interpretation model in S103 is to construct two groups of remote sensing object interpretation models based on the UNet model, and the specific contents are as follows:

[0031] UNet-1 model: A feature extraction network composed of 9 3×3 convolution modules and 3 maximum pooling modules; specifically, 9 deconvolution modules upsample the target feature information, and fuse the feature map containing the low-order position information of the target object extracted by the feature extraction network with the feature map retaining the high-order semantic information of the upsampled part through 3 inter-layer connections, and extract and classify the target objects of the preprocessed public data set end-to-end;

[0032] UNet-2 model: A feature extraction network composed of 6 3×3 convolution modules and 2 maximum pooling modules; specifically, an upsampling network composed of 6 deconvolution modules, and two inter-layer connections to achieve the fusion of low-order position information and high-order semantic information, and to extract and classify the scene target objects in the foreground and background areas of the preprocessed public data set.

[0033] Optionally, a remote sensing object interpretation model corresponding to the network depth is selected according to the target object complexity threshold, wherein the corresponding relationship between the target object complexity threshold and the remote sensing object interpretation model is: the target object complexity threshold>k 1 When , it corresponds to the UNet-2 model; k 2 <The target object complexity threshold ≤ k 1 When, it corresponds to the UNet-1 model; the target object complexity threshold ≤ k 2 When, for the UNet model, k 1 >k 2 , k 1 , k 2 are constants, and 0 <k 2 <k 1 <1.

[0034] Optionally, in S101, the preprocessed image data is divided into a training set, a test set, and a validation set in a ratio of 0.8:0.1:0.1; wherein the training set is a sample set used for training the improved UNet remote sensing ground object interpretation model, the test set is a sample set used for cross-validation of the improved UNet remote sensing ground object interpretation model, and the validation set is a sample used to evaluate the model performance after the training of the improved UNet remote sensing ground object interpretation model is completed.

[0035] A remote sensing image intelligent interpretation system based on an adaptive sensor, comprising an image preprocessing module, an adaptive sensor building module, a feature training module and a remote sensing intelligent interpretation module connected in sequence;

[0036] The image preprocessing module is used to obtain a public data set of remote sensing image interpretation, preprocess the public data set to obtain preprocessed image data, and divide the preprocessed image data into a training set, a test set, and a validation set according to a proportion;

[0037] The adaptive perceptron construction module is used to construct a perceptron model, calculate the training data set, and output a target object complexity threshold;

[0038] The feature training module is used to select improved UNet remote sensing object interpretation models of various scales and network depths according to the complexity threshold of the target object, iteratively solve the training weights of the selected interpretation category objects in the public data set, and obtain the remote sensing object intelligent interpretation weight model;

[0039] The remote sensing intelligent interpretation module is used to import the remote sensing object intelligent interpretation weight model into the improved UNet remote sensing object interpretation model, input the test set, and perform adaptive remote sensing object intelligent interpretation.

[0040] It can be seen from the above technical solutions that, compared with the prior art, the present invention provides a remote sensing image intelligent interpretation method and system based on an adaptive perceptron: the remote sensing image data set is preprocessed by histogram correction and band combination to highlight the difference between the foreground and background of the target object, which is convenient for the extraction of the target object while reducing the model calculation amount and model training time; the remote sensing object interpretation technology combining an adaptive perceptron with an improved UNet is more in line with realistic scenarios, and can be applied to remote sensing interpretation of a variety of target objects, with a wide range of applications and more flexible applications; by setting the target object complexity judgment threshold size, and improving the UNet model to obtain multiple interpretation networks with different network depths, the interpretation accuracy can be guaranteed and the overall interpretation speed can be improved in the case of different target object interpretations; it is easy to implement and can better serve remote sensing interpretation, global land cover change research, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0042] Figure 1 This is a schematic diagram of the intelligent interpretation method of remote sensing images based on an adaptive sensor;

[0043] Figure 2 The interpretation network structure diagram of the remote sensing image intelligent interpretation method based on the adaptive sensor;

[0044] Figure 3 This is a flow chart of a remote sensing image intelligent interpretation method based on an adaptive sensor in the present invention;

[0045] Figure 4 This is a structural block diagram of a remote sensing image intelligent interpretation system based on an adaptive sensor in the present invention;

[0046] Figure 5These are the results of the ground feature interpretation of the present invention, where 5a is a remote sensing image of a town, and 5b is a remote sensing intelligent interpretation image based on this method. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Reference Figure 1 As shown, a schematic diagram of a remote sensing image intelligent interpretation method based on an adaptive perceptron is disclosed, which includes three aspects: impact preprocessing, adaptive perceptron and remote sensing intelligent interpretation.

[0049] Reference Figure 2 As shown in FIG. 1 , the interpretation network structure diagram of the remote sensing image intelligent interpretation method based on the adaptive perceptron includes: UNet model, UNet-1 model and UNet-2 model.

[0050] Reference Figure 3 As shown, the present invention discloses a remote sensing image intelligent interpretation method based on an adaptive perceptron, comprising the following steps:

[0051] S101: Obtain a public data set of remote sensing image interpretation, preprocess the public data set to obtain preprocessed image data, and divide the preprocessed image data into a training set, a test set, and a validation set according to a proportion;

[0052] S102: construct a perceptron model, calculate the training data set, and output the target object complexity threshold;

[0053] S103: According to the target object complexity threshold, an improved UNet remote sensing object interpretation model with multiple scales and multiple network depths is selected to iteratively solve the training weights of the selected interpretation category objects in the public data set to obtain a remote sensing object intelligent interpretation weight model;

[0054] S104: Import the remote sensing object intelligent interpretation weight model into the improved UNet remote sensing object interpretation model, input the test set, and perform adaptive remote sensing object intelligent interpretation.

[0055] Furthermore, the preprocessing in S101 includes: histogram correction and band combination;

[0056] Among them, the specific content of histogram correction is:

[0057] For pixels whose radiation brightness or reflection brightness should be 0 or close to 0 in mountain shadows and deep sea water bodies with extremely low reflectivity, their actual pixel values ​​are affected by atmospheric scattering. Based on the histogram correction method, the brightness values ​​of M×N pixels in each band are subtracted from the minimum brightness value of the pixel in this band, where the image width is M pixels and the height is N pixels. The calculation is as follows:

[0058]

[0059] In the formula, f ij is the pixel brightness value of this band, f min is the minimum brightness of the pixel in this band, f g is the brightness value of the image pixel after histogram correction;

[0060] The specific content of band combination is: set different band combination methods to generate false color images. Through preprocessing, remove the influence of atmospheric radiation on remote sensing images and highlight the difference between the foreground and background of the image. Specifically, different bands of multispectral images have different spectral reflectance characteristics for different objects. Set different 433, 456 and other band combination methods to generate false color images, highlighting the difference between the foreground and background of various target objects in the image.

[0061] Furthermore, constructing the perceptron model in S102 includes the following steps:

[0062] S1021: The preprocessed public dataset is stacked to form a linearly separable perceptron dataset. The formula is as follows

[0063] T={(x 1 ,y 1 ),(x 2 ,y 2 ),...,(x n ,y n )} (2)

[0064] In the formula, x i ∈X=R n ,y i ∈Y=R n , i=1,2,...,N,x i ,y i All are on the maximum margin hyperplane ω * x+b * = 0; X is an arbitrary axis of the two-dimensional feature hyperplane, Y is an axis orthogonal to X in the two-dimensional feature hyperplane, ω * is the slope of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system, b *is the intercept of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system and the Z axis, n is the n-dimensional feature space, and N is an integer set.

[0065] Among them, the classification function expression of the adaptive perceptron model is:

[0066] f(x)=sign(ω * x+b * ) (6)

[0067] Where x is the input feature vector, specifically the color difference and boundary contour between the front and background areas of the target object in the input training set image; ω * x+b * is the feature classification hyperplane;

[0068] S1022: Define the perceptron loss function as the sum of the distances from the misclassified points to the feature classification hyperplane, as shown in the following formula:

[0069]

[0070] Among them, the output result of the perceptron of the misclassified point has the opposite sign to the true result, as shown in the following formula:

[0071]

[0072] The adaptive perceptron model expression of the fusion loss function is shown as follows:

[0073]

[0074] In formulas (3)-(5), M is the set of misclassified points.

[0075] Among them, the output of the perceptron model for the target object complexity judgment results of different false color data sets and different target objects is converted into the target object complexity threshold and input into the improved UNet network.

[0076] Furthermore, the improved UNet remote sensing object interpretation model in S103 is to construct two groups of remote sensing object interpretation models based on the UNet model, and the specific contents are as follows:

[0077] UNet-1 model: A feature extraction network composed of 9 3×3 convolution modules and 3 maximum pooling modules; specifically, 9 deconvolution modules upsample the target feature information, and fuse the feature map containing the low-order position information of the target object extracted by the feature extraction network with the feature map retaining the high-order semantic information of the upsampled part through 3 inter-layer connections, and extract and classify the target objects of the preprocessed public data set end-to-end;

[0078] UNet-2 model: A feature extraction network composed of 6 3×3 convolution modules and 2 maximum pooling modules; specifically, an upsampling network composed of 6 deconvolution modules, and two inter-layer connections to achieve the fusion of low-order position information and high-order semantic information, and to extract and classify the scene target objects in the foreground and background areas of the preprocessed public data set.

[0079] Furthermore, the remote sensing object interpretation model corresponding to the network depth is selected according to the target object complexity threshold, wherein the corresponding relationship between the target object complexity threshold and the remote sensing object interpretation model is: target object complexity threshold>k 1 When , it corresponds to the UNet-2 model; k 2 <Target object complexity threshold ≤ k 1 When , it corresponds to the UNet-1 model; the target object complexity threshold ≤ k 2 When, for the UNet model, k 1 >k 2 , k 1 , k 2 are constants, and 0 <k 2 <k 1 <1.

[0080] According to the output results of the perceptron model, the objects judged as simple targets based on the above threshold are input into the UNet-1 and UNet-2 interpretation networks with fewer layers and simpler models to reduce the amount of model calculation and increase the calculation speed. The complex target objects judged as complex targets based on the above threshold are input into the deep convolutional network because their foreground and background areas have a high degree of similarity in color, texture, contour, etc., and are easily affected by background interference. While extracting the target position information in the low-order feature map, the target semantic information in the high-order feature map is also extracted, thereby achieving the purpose of accurate target extraction.

[0081] Furthermore, in S101, the preprocessed image data is divided into a training set, a test set, and a validation set in a ratio of 0.8:0.1:0.1; wherein the training set is a sample set used for training the improved UNet remote sensing land object interpretation model, the test set is a sample set used for cross-validation of the improved UNet remote sensing land object interpretation model, and the validation set is a sample used to evaluate the model performance after the training of the improved UNet remote sensing land object interpretation model is completed.

[0082] Reference Figure 4 As shown, the present invention also discloses a remote sensing image intelligent interpretation system based on an adaptive sensor, which applies the above-mentioned remote sensing image intelligent interpretation method based on an adaptive sensor, including an image preprocessing module, an adaptive sensor building module, a feature training module and a remote sensing intelligent interpretation module connected in sequence;

[0083] The image preprocessing module is used to obtain the public data set of remote sensing image interpretation, preprocess the public data set to obtain preprocessed image data, and divide the preprocessed image data into training set, test set and validation set according to proportion;

[0084] The adaptive perceptron construction module is used to build a perceptron model, calculate the training data set, and output the target object complexity threshold;

[0085] The feature training module is used to select the improved UNet remote sensing object interpretation model with multiple scales and network depths according to the complexity threshold of the target object, iteratively solve the training weights of the selected interpretation category objects in the public data set, and obtain the remote sensing object intelligent interpretation weight model;

[0086] The remote sensing intelligent interpretation module is used to import the remote sensing object intelligent interpretation weight model into the improved UNet remote sensing object interpretation model, input the test set, and perform adaptive remote sensing object intelligent interpretation.

[0087] In order to further clarify the invention's remote sensing image intelligent interpretation method based on adaptive perceptron and improved UNet model, a Sentinel image of Shenyang, Liaoning provided by the European Global Environment and Security Monitoring System project and the AID Dataset remote sensing image classification public data set are used to verify the patent model and algorithm. Note that this is only for illustration, and the invention does not limit the data source.

[0088] The experimental data uses the AID Dataset remote sensing image classification public data set, and the ENVI software released by Exelis Visual Information Solutions of the United States is used for data preprocessing.

[0089] The specific steps are as follows:

[0090] S1, remote sensing image preprocessing: obtain public data sets on remote sensing image interpretation, perform histogram correction and band combination, remove the impact of atmospheric radiation on remote sensing images, and highlight the difference between the foreground and background of the image. The specific methods are as follows:

[0091] 1) Histogram correction:

[0092] For pixels in the shadow of the mountain, the radiance should be 0 or close to 0. The actual pixel value is affected by atmospheric scattering. The histogram correction method of ENVI software is used to subtract the minimum brightness value of the pixel in each band from the brightness values ​​of MN pixels in each band (the image width is M pixels and the height is N pixels), as shown in formula (1).

[0093] 2) Image band combination:

[0094] Different bands of multispectral images have different spectral reflectance characteristics for different objects. By studying the false color images of three band combinations, namely, 543 band combination, 652 band combination and 432 band combination, the differences between the foreground and background of various target objects in the images are highlighted. The preprocessed image data is divided into training set, test set and validation set in a ratio of 0.8:0.1:0.1.

[0095] S2, perceptron model construction:

[0096] The perceptron model is constructed to judge the complexity of the target object in the training set data, and the color difference between the foreground and background of the data and the outline of the target object are weighted and analyzed to output the judgment threshold. The classification function expression of the perceptron model is shown in formula (6). The classification function of the perceptron model is determined by calculating the maximum distance from each feature vector to the feature hyperplane in the feature space. The specific method is as follows:

[0097] 1) Data is linearly separable:

[0098] Based on the aforementioned multiple false color images with different band combinations, the AID Dataset remote sensing image classification public data set obtained by this patent is expanded, and different remote sensing image object interpretation data sets are used as input parameters of the adaptive perceptron, as shown in formula (2).

[0099] 2) Perceptron Learning:

[0100] The perceptron loss function is defined as the sum of the distances from the misclassified points to the feature classification hyperplane, as shown in formula (3). Among them, the perceptron output result of the misclassified point has the opposite sign to the true result, as shown in formula (4). The adaptive perceptron model expression of the fusion loss function is shown in (5), where M is the set of misclassified points. After learning, the perceptron outputs the target object complexity threshold (P).

[0101] The output of the perceptron model for the target object complexity judgment results of different false color data sets and different target objects is converted into the target object complexity threshold and input into the improved UNet network.

[0102] S3, remote sensing intelligent interpretation:

[0103] This step mainly involves selecting the corresponding interpretation network after the output of the perceptron to extract the target object features and interpret the remote sensing objects, including:

[0104] 1) Target feature extraction:

[0105] According to the output results of the adaptive perceptron, the objects judged as simple targets based on the aforementioned threshold are input into the UNet-1 and UNet-2 interpretation networks with fewer layers and simpler models to reduce the amount of model calculation and increase the calculation speed. The complex target objects judged as complex targets based on the aforementioned threshold are input into the deep convolutional network because their foreground and background areas have a high degree of similarity in color, texture, contour, etc., and are easily affected by background interference. While extracting the target position information in the low-order feature map, the target semantic information in the high-order feature map is also extracted, thereby achieving the purpose of accurate target extraction.

[0106] 2) Remote sensing ground feature interpretation:

[0107] The false color data sets of the aforementioned different band combinations are used to iteratively solve the training weights on the improved UNet model, and the optimal training weights are obtained by adjusting the parameters, and the target objects are intelligently interpreted. The adaptive perceptron semantics under different band combinations are judged as simple objects, and the extraction results are stacked to achieve intelligent interpretation of multiple types of remote sensing objects.

[0108] Reference Figure 3 As shown, the remote sensing object interpretation results obtained by using this embodiment are given. For quantitative analysis, the present invention uses expert visual interpretation as a reference, constructs a mathematical model, and evaluates the interpretation accuracy of the method. The interpreted remote sensing image result map (interpretation map) and the remote sensing image visually interpreted by the expert (i.e., the expert map) are binarized. The pixels of some target objects are set to 255, and the pixels of some background are set to 0, and the pixel-level comparison results of the two images are used as evaluation indicators. This paper uses three performance indicators to evaluate the remote sensing intelligent interpretation method, namely, precision, recall, and accuracy, which are defined as:

[0109]

[0110] In formulas (7)-(9): TP is the number of pixels whose pixel values ​​in the interpretation map and the expert map are both 255, that is, the number of pixels correctly identified by the algorithm; FP is the number of pixels whose pixel value in the interpretation map is 255 but whose pixel value in the expert map is 0, that is, the number of pixels misidentified by the algorithm; FN is the number of pixels whose pixel value in the interpretation map is 0 but whose pixel value in the expert map is 255, that is, the number of pixels occupied by the target objects missed by the algorithm; TN is the number of pixels whose pixel values ​​in the interpretation map and the expert map are both 0, that is, the number of pixels correctly judged as the background area by the detection algorithm. The vegetation height result obtained by the algorithm of the present invention has an accuracy rate of 91.3%, a recall rate of 87.7%, and a precision rate of 90.3%, which is higher than the expert annotation result in the prior art.

[0111] The above description of the disclosed embodiments is provided in a progressive manner to enable those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent interpretation of remote sensing images with complex backgrounds, characterized in that: The following steps are involved: S101: Obtain a public dataset of remote sensing image interpretation, preprocess the public dataset, and divide the preprocessed image data into a training set, a test set, and a validation set according to a proportion; S102: construct a perceptron model, perform calculations on the training set, and output a threshold value of the complexity of the target object; S103: According to the complexity threshold of the target object, the improved UNet remote sensing object interpretation model is selected to iteratively solve the training weights of the selected interpretation category objects in the public data set to obtain the remote sensing object intelligent interpretation weight model; S104: importing the remote sensing object intelligent interpretation weight model into the improved UNet remote sensing object interpretation model, inputting the test set, and performing adaptive remote sensing object intelligent interpretation; The improved UNet remote sensing object interpretation model in S103 is to build two sets of remote sensing object interpretation models based on the UNet model, including: UNet-1 model: It consists of 9 3×3 convolution modules and 3 maximum pooling modules to form a feature extraction network. Specifically, 9 deconvolution modules upsample the target feature information, and fuse the feature map containing the low-order position information of the target object extracted by the feature extraction network with the feature map of the upsampled part that retains the high-order semantic information through 3 inter-layer connections, so as to extract and classify the target objects in the preprocessed public data set end-to-end. UNet-2 model: It consists of 6 3×3 convolution modules and 2 maximum pooling modules to form a feature extraction network. Specifically, it consists of 6 deconvolution modules to form an upsampling network. Two inter-layer connections are used to achieve the fusion of low-order position information and high-order semantic information, and to extract and classify the target objects in the foreground and background areas of the preprocessed public data set.

2. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: The preprocessing in S101 includes: histogram correction and band combination; Among them, the specific content of histogram correction is: Subtract the minimum brightness value of the pixel in each band from the brightness values ​​of the M×N pixels in each band, where the image width is M pixels and the height is N pixels. The calculation is as follows: In the formula, f ij is the pixel brightness value of this band, f min is the minimum brightness of the pixel in this band, f g is the brightness value of the image pixel after histogram correction; The specific content of band combination is: setting different band combinations to generate false color images.

3. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: Constructing the perceptron model in S102 includes the following steps: S1021: The preprocessed public dataset is stacked to form a linearly separable perceptron dataset. The formula is as follows T={(x1,y1),(x2,y2),...,(x n ,y n )} (2) In the formula, x i ∈X=R n ,y i ∈Y=R n , i=1,2,...,N,x i ,y i All are on the maximum margin hyperplane ω * x+b * = 0; X is an arbitrary axis of the two-dimensional feature hyperplane, Y is an axis orthogonal to X in the two-dimensional feature hyperplane, ω * is the slope of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system, b * is the intercept of the projection of the two-dimensional feature hyperplane on the XOZ plane in the spatial rectangular coordinate system and the Z axis, n is the n-dimensional feature space, and N is an integer set; Among them, the classification function expression of the adaptive perceptron model is: f(x)=sign(ω * x+b * ) (6) Where x is the input feature vector, specifically the color difference and boundary contour between the front and background areas of the target object in the input training set image; ω * x+b * is the feature classification hyperplane; S1022: Define the perceptron loss function as the sum of the distances from the misclassified points to the feature classification hyperplane, as shown in the following formula: Among them, the output result of the perceptron of the misclassified point has the opposite sign to the true result, as shown in the following formula: The adaptive perceptron model expression of the fusion loss function is shown as follows: In formulas (3)-(5), M is the set of misclassified points; Among them, the perceptron model outputs the target object complexity judgment results for different false color data sets and different target objects and converts them into target object complexity thresholds, which are input into the improved UNet network.

4. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: According to the target object complexity threshold, the remote sensing object interpretation model corresponding to the network depth is selected. The corresponding relationship between the target object complexity threshold and the remote sensing object interpretation model is as follows: when the target object complexity threshold>k1, it corresponds to the UNet-2 model; when k2<target object complexity threshold≤k1, it corresponds to the UNet-1 model; when the target object complexity threshold≤k2, it corresponds to the UNet model, where k1>k2, k1 and k2 are both constants, and 0 <k2<k1<1。 5. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: In S101, the preprocessed image data is divided into a training set, a test set, and a validation set in a ratio of 0.8:0.1:0.1; among them, the training set is a sample set used for training the improved UNet remote sensing land object interpretation model, the test set is a sample set used for cross-validation of the improved UNet remote sensing land object interpretation model, and the validation set is a sample used to evaluate the model performance after the training of the improved UNet remote sensing land object interpretation model is completed.

6. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: Also includes: According to the output results of the perceptron model, the objects judged as simple targets based on the target object complexity threshold are input into the UNet-1 and UNet-2 models; the objects judged as complex targets based on the target object complexity threshold are input into the deep convolutional network to extract the target semantic information in the high-order feature map while extracting the target position information in the low-order feature map.

7. The method for intelligent interpretation of complex background remote sensing image objects according to claim 1, characterized in that: Also includes: The remote sensing image result map after adaptive remote sensing object intelligent interpretation is regarded as the interpretation map, and the remote sensing image visually interpreted by experts is regarded as the expert map. The two are binarized to make the target object part pixel 255 and the background part pixel 0. The pixel-level comparison results of the two images are used as evaluation indicators. The precision, recall and accuracy are used to evaluate the intelligent interpretation method of remote sensing image objects with complex backgrounds. Specifically: Where TP is the number of pixels whose pixel values ​​in both the interpretation map and the expert map are 255, that is, the number of pixels correctly identified by the algorithm; FP is the number of pixels whose pixel value in the interpretation map is 255 but whose pixel value in the expert map is 0, that is, the number of pixels misidentified by the algorithm; FN is the number of pixels whose pixel value in the interpretation map is 0 but whose pixel value in the expert map is 255, that is, the number of pixels occupied by target objects missed by the algorithm; TN is the number of pixels whose pixel values ​​in both the interpretation map and the expert map are 0, that is, the number of pixels correctly judged as background areas by the detection algorithm.

Images