Geographic information image extraction feature and generation system and method

Abstract

The present invention relates to the field of image processing, and in particular to a system and method for extracting features and generating geographic information images. First, the collected geographic information images are standardized, and multi-scale features are extracted from the standardized geographic information images using convolution kernels. A dynamic feature fusion mechanism is designed, and the weight coefficient of the feature map is adjusted to obtain a fused feature map. Further, the information density of the fused feature map is evaluated, the scale of the feature map is dynamically adjusted, the geographic information image is generated, and the generation strategy and noise vector are adjusted according to feedback information. The present invention solves the problems that the prior art is insufficient in accuracy in extracting complex terrain features, and the quality of analysis and decision-making is limited; the computing resources are consumed too much when processing high-dimensional data, affecting the real-time performance of the system; and the lack of an effective dynamic weight allocation mechanism causes important geographic information features to be not fully utilized, thereby reducing the accuracy of the final analysis results.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a system and method for extracting features and generating geographic information images. Background Art

[0002] With the rapid development of geographic information systems (GIS) and remote sensing technology, geographic information image processing has become a key technical means to obtain, analyze and understand geographic environmental information. These technologies are widely used in many fields such as urban planning, environmental monitoring, disaster management and agricultural assessment. With the increasing amount of geographic data and the increasing complexity of geographic information, traditional image processing methods face severe challenges in feature extraction accuracy, processing efficiency, computing resource consumption and model generalization ability. Especially when dealing with complex terrain feature extraction, existing technologies often find it difficult to accurately capture multi-scale spatial features, resulting in insufficient information extraction or information loss. At the same time, the computational burden brought by high-dimensional data processing also greatly limits the processing efficiency. Therefore, developing a new image processing method that can efficiently and accurately process geographic information images, optimize computing resource consumption, reduce information loss and prevent data overfitting has become a technical problem that urgently needs to be solved in this field.

[0003] my country's patent application number: CN202010294101.X, publication date: 2020.09.01, discloses a method for extracting features and generating geographic information images, including a geographic information image extraction module and a geographic information image generation module; the geographic information image extraction module includes satellite remote sensing image extraction, drone aerial photography and mapping, and total station field mapping; the geographic information image generation module includes establishing raster data for geographic information in the target area, and rendering the raster data to obtain a geographic information image in the target area; the geographic information image extraction module and the geographic information image generation module are connected by a computer to form signal transmission. By selecting raster data in the corrected two-dimensional geographic information space library, a local geographic information picture of the area can be obtained. By comparing the generated two-dimensional geographic information space library with the three-dimensional stereo model, the presentation of the geographic information image can be enhanced, which is convenient for staff to use.

[0004] However, the above technologies have at least the following technical problems: the existing technologies are not accurate enough in extracting complex terrain features, which limits the quality of analysis and decision-making; excessive computing resources are consumed when processing high-dimensional data, affecting the real-time performance of the system; and due to the lack of an effective dynamic weight allocation mechanism, important geographic information features are not fully utilized, reducing the accuracy of the final analysis results. Summary of the invention

[0005] The present invention solves the problems of insufficient accuracy in complex terrain feature extraction, limited analysis and decision-making quality in the prior art, excessive consumption of computing resources in high-dimensional data processing, affecting the real-time performance of the system, and failure to fully utilize important geographic information features due to the lack of an effective dynamic weight allocation mechanism, thereby reducing the accuracy of the final analysis results, by providing a feature extraction and generation system and method for geographic information images. The present invention significantly improves the accuracy and efficiency of geographic information image processing by dynamically fusing multi-scale feature extraction, adaptive scale adjustment, and feedback-based image generation strategies, while optimizing computing resource consumption, reducing information loss, and preventing data overfitting.

[0006] The present invention provides a system and method for extracting features and generating geographic information images, which specifically include the following technical solutions:

[0007] The feature extraction and generation system of geographic information images includes the following parts:

[0008] Data acquisition module, preprocessing module, preliminary feature extraction module, dynamic feature fusion module, adaptive feature scaling module, image generation module, feedback adjustment module;

[0009] The data acquisition module is used to collect geographic information image data; the data acquisition module is connected to the preprocessing module by means of data transmission;

[0010] The preprocessing module is used to perform standardization processing on the collected geographic information images; the preprocessing module is connected to the preliminary feature extraction module by means of data transmission;

[0011] The preliminary feature extraction module is used to analyze the standardized image by applying convolution kernels of different scales, and capture multi-level spatial features to form a multi-scale feature map; the preliminary feature extraction module is connected to the dynamic feature fusion module by means of data transmission;

[0012] The dynamic feature fusion module is used to dynamically analyze and fuse multi-scale feature maps; the dynamic feature fusion module is connected to the adaptive feature scaling module by means of data transmission;

[0013] The adaptive feature scaling module is used to adjust the scale of the fused feature map; the adaptive feature scaling module is connected to the image generation module by means of data transmission;

[0014] The image generation module is used to reconstruct the geographic information image according to the adjusted feature map in combination with the generation model; the image generation module is connected to the feedback adjustment module by means of data transmission;

[0015] The feedback adjustment module is used to evaluate the quality of the geographic information image reconstructed by the image generation module, provide feedback information, and guide the adjustment of the generation model; the feedback adjustment module is connected to the image generation module by means of data transmission.

[0016] The feature extraction and generation method of geographic information image includes the following steps:

[0017] S1. Standardize the collected geographic information images, use convolution kernels to extract multi-scale features from the standardized geographic information images to generate multi-scale feature maps, design a dynamic feature fusion mechanism, adjust the weight coefficients of the multi-scale feature maps, and obtain fused feature maps;

[0018] S2. Evaluate the information density of the fused feature map, dynamically adjust the scale of the fused feature map, design a feedback-based generation model, and generate geographic information images; evaluate the quality of the geographic information image through the feedback network and provide feedback, and then adjust the generation strategy and noise vector.

[0019] A feature extraction and generation system applied to the above-mentioned geographic information images.

[0020] Preferably, the S1 further specifically includes:

[0021] Multi-level spatial information is extracted through convolution kernels of different scales, and the maximum pooling operation is applied to obtain a multi-scale feature map.

[0022] Preferably, the S1 further specifically includes:

[0023] A dynamic feature fusion mechanism is designed to calculate the information entropy of multi-scale feature maps. Combined with image complexity evaluation, an adjustment factor is introduced to adjust the weight coefficient of the feature map to obtain a fused feature map.

[0024] Preferably, the S2 further specifically includes:

[0025] In the process of evaluating the information density of the fused feature map, the information density of the fused feature map is calculated based on the local information volume and the global distribution, and then the scale of the fused feature map is dynamically adjusted.

[0026] Preferably, the S2 further specifically includes:

[0027] A feedback-based generative model is designed to generate geographic information images based on the adjusted feature maps and initial noise vectors.

[0028] Preferably, the S2 further specifically includes:

[0029] In the process of generating geographic information images, a dynamic adjustment mechanism is introduced to evaluate the geographic information images to obtain a comprehensive quality score.

[0030] Preferably, the S2 further specifically includes:

[0031] Based on the comprehensive quality score, the feedback adjustment module adjusts the noise vector through a feedback mechanism and repeats the process of evaluating and adjusting the noise vector for the geographic information image.

[0032] The beneficial effects of the technical solution of the present invention are:

[0033] 1. By using convolution kernels of different scales to capture multi-level spatial information from subtle to coarse in geographic information images, the features of complex terrain can be accurately extracted; the dynamic feature fusion mechanism further ensures the maximum utilization of information in the feature fusion process, thereby significantly improving the accuracy of feature extraction of geographic information images; the adaptive feature scaling module automatically adjusts the scale of the fused feature map by evaluating its information density, effectively reducing the dimension of feature representation, which not only ensures the retention of key information, but also significantly reduces the consumption of computing resources when processing high-dimensional data, and improves processing efficiency;

[0034] 2. By carefully extracting features at each scale and integrating these features using a dynamic weight allocation mechanism, the information loss that may occur during the feature fusion process is effectively reduced. Adaptive feature scaling further ensures that useful information is retained to the maximum extent while reducing the computational burden. The feedback-based generation model evaluates the quality of the generated geographic information images and adjusts the generation strategy based on the feedback. This not only improves the quality of the generated geographic information images, but also avoids overfitting of the model to the training data during the learning process through this dynamic adjustment mechanism, thereby improving the generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A diagram showing the composition of the feature extraction and generation system modules of a geographic information image provided by one embodiment of the present invention;

[0036] Figure 2 A flow chart of a method for extracting features and generating geographic information images provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiment 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.

[0038] 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.

[0039] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0040] See attached Figure 1 , a system for extracting features and generating geographic information images provided by an embodiment of the present invention includes the following parts:

[0041] Data acquisition module, preprocessing module, preliminary feature extraction module, dynamic feature fusion module, adaptive feature scaling module, image generation module, feedback adjustment module;

[0042] The data acquisition module is used to collect geographic information image data; the data acquisition module is connected to the preprocessing module by means of data transmission;

[0043] The preprocessing module is used to perform standardization processing on the collected geographic information images; the preprocessing module is connected to the preliminary feature extraction module by means of data transmission;

[0044] The preliminary feature extraction module is used to analyze the standardized image by applying convolution kernels of different scales, and capture multi-level spatial features to form a multi-scale feature map; the preliminary feature extraction module is connected to the dynamic feature fusion module by means of data transmission;

[0045] The dynamic feature fusion module is used to dynamically analyze and fuse multi-scale feature maps; the dynamic feature fusion module is connected to the adaptive feature scaling module by means of data transmission;

[0046] The adaptive feature scaling module is used to adjust the scale of the fused feature map to save computing resources; the adaptive feature scaling module is connected to the image generation module by means of data transmission;

[0047] The image generation module is used to reconstruct the geographic information image according to the adjusted and fused feature map in combination with the generation model; the image generation module is connected to the feedback adjustment module by means of data transmission;

[0048] The feedback adjustment module is used to evaluate the quality of the geographic information image reconstructed by the image generation module, provide feedback information, and guide the adjustment of the generation model; the feedback adjustment module is connected to the image generation module by means of data transmission.

[0049] See attached Figure 2 , a method for extracting features and generating geographic information images provided by an embodiment of the present invention comprises the following steps:

[0050] S1. Standardize the collected geographic information images, use convolution kernels to extract multi-scale features from the standardized geographic information images to generate multi-scale feature maps, design a dynamic feature fusion mechanism, adjust the weight coefficients of the multi-scale feature maps, and obtain fused feature maps;

[0051] In the field of geographic information image processing, facing key technical problems such as insufficient extraction of complex terrain features, excessive consumption of computing resources, information loss, and data overfitting, a geographic information dynamic fusion network is constructed. By comprehensively using multi-scale feature extraction, dynamic weight allocation, adaptive feature scaling, and feedback-based image generation strategies, the aim is to improve the processing efficiency and accuracy of geographic information images.

[0052] The data acquisition module obtains geographic information images. In the design of the geographic information dynamic fusion network framework, the complexity of feature extraction of geographic information images is taken into account, including the multi-scale characteristics of the image and the inherent spatial relationship. The input geographic information image is first standardized to adjust the pixel value range of the image to meet the needs of subsequent deep network processing. The preprocessing module calculates the pixel value mean and standard deviation of the original geographic information image and applies the following standardization formula:

[0053]

[0054] Among them, I raw represents the original geographic information image, I is the processed image, and They refer to the mean and standard deviation of the pixel values ​​of the original geographic information image respectively.

[0055] In order to capture spatial features at different scales, the preliminary feature extraction module applies a series of convolution kernels of different scales to the processed image. Convolution kernels of different scales can extract multi-level spatial information from subtle to coarse, enriching the understanding of geographic information images. The convolution operation at the i-th scale is defined by the following formula:

[0056]

[0057] Among them, F′ si represents the unpooled i-th scale feature map, s is the scale, i is the index of the scale, K si represents the total number of convolution kernels used at the i-th scale, k∈[1, K si ], K si，k and b si，k are the kth convolution kernel and its bias term at the i-th scale, and σ represents a nonlinear activation function, such as ReLU. This captures the features of the image at different scales and provides a basis for subsequent feature fusion.

[0058] In order to reduce the spatial dimension of each scale feature map while retaining key feature information, a maximum pooling operation is applied to each scale feature map. The purpose of the maximum pooling operation is to reduce the size of the feature map by selecting the maximum value in each small area. This is done by the following formula:

[0059] F si =MaxPool(F′ si )

[0060] Among them, F si It is the i-th scale feature map after the maximum pooling operation, and MaxPool represents the maximum pooling operation.

[0061] After obtaining the multi-scale feature maps, the dynamic feature fusion module designs a dynamic feature fusion mechanism to automatically adjust the weight coefficients of the feature maps in order to more effectively fuse the feature maps. By calculating the information entropy of the feature maps at each scale and combining the evaluation of image complexity, the dynamic feature fusion mechanism can dynamically assign weights to maximize information retention and extraction. The dynamic calculation of weights not only depends on the amount of information in the feature map itself, but also takes into account the influence of image complexity, thereby ensuring the maximum utilization of information in the feature fusion process.

[0062] The complexity of the image is quantified by the entropy of its gradient. The gradient entropy of the feature map at the i-th scale is calculated to represent the image complexity. The formula is as follows:

[0063]

[0064] Among them, E g (F si ) is F si Gradient entropy, F si is the i-th scale feature map after the maximum pooling operation, V is F si The gradient value set of p(V) is the probability density function of the gradient value v. The calculation of the weight coefficient takes into account the combined influence of the information content of the feature map and the complexity of the image, so an adjustment factor δ is introduced, and its calculation formula is:

[0065]

[0066]

[0067] Among them, w si represents the weight coefficient of the feature map of the i-th scale, H(F si ) is the feature map F si The information entropy of F is β and δ, which are learnable parameters that adjust the influence of information entropy and gradient entropy on weights respectively. mergedIt is a fusion feature map that integrates multi-scale information, and n is the total number of scales.

[0068] S2. Evaluate the information density of the fused feature map, dynamically adjust the scale of the fused feature map, design a feedback-based generation model, and generate geographic information images; evaluate the quality of the geographic information image through the feedback network and provide feedback, and then adjust the generation strategy and noise vector.

[0069] In order to solve the problem of excessive consumption of computing resources when processing high-dimensional data, an adaptive feature scaling module is further introduced. By evaluating the information density of the fused feature map, its scale is automatically adjusted, effectively reducing the dimension of the feature representation while ensuring the retention of key information. The calculation of information density takes into account the local information volume and global distribution, aiming to comprehensively evaluate the information richness of the feature map and adjust the scale accordingly. The calculation formula is as follows:

[0070]

[0071] Among them, D(F merged ) represents the fusion feature map F merged The information density of F is the combination of local information and global distribution. merged (x, y), Θ) is a local complexity function based on the pixel value and parameter set Θ, F merged (x, y) is the fusion feature map F merged The pixel value at position (x, y) in the middle, x represents the horizontal position index, y represents the vertical position index, W and H represent the width and height of the feature map respectively, Ω(F merged ) is a function that measures the complexity of the overall feature map, and λ is the balance coefficient.

[0072] The adaptive feature scaling module dynamically adjusts the scale of the feature map according to the information density of the fused feature map, retaining key information while reducing the size of the feature map to save computing resources. The scale adjustment can be achieved by the following formula:

[0073]

[0074] Among them, F adjusted Resize represents the scale-adjusted feature map. Resize represents the scale adjustment function, which adjusts the size of the feature map according to the given scale parameter. S(D(F merged )) is a function based on the information density D(F merged ) is used to determine the function of the new scale. The scale adjustment function S can be specifically defined as:

[0075] S(D)=max(min scale , min(1,κ-ηD))

[0076] Among them, min scale The lower limit of scale adjustment is defined to prevent information loss caused by the feature map being too small. κ and η are parameters determined based on experimental experience to adjust the influence of information density on scale adjustment.

[0077] Aiming at the information loss and data overfitting problems that may occur in the generation process, a feedback-based generation model is designed. After the initial generation of geographic information images, the generation model uses a feedback network to evaluate the quality of the generated geographic information images and provide specific feedback information. The generation strategy and noise vector are adjusted according to the feedback information to optimize the quality of the generated geographic information images. The calculation of the feedback information not only considers the image detail retention, color authenticity and content accuracy, but also introduces a dynamic adjustment mechanism to ensure the high quality of the generated geographic information images and the generalization ability of the model.

[0078] The image generation module uses the adjusted feature map F adiusted and the initial noise vector z init As input, a preliminary geographic information image G is generated by the generator G initial :

[0079] G initial =G(F adjusted , z init )

[0080] Among them, G is the generator network, which is responsible for generating geographic information images based on the given feature map and noise vector. The evaluation function is used to evaluate the generated geographic information image G initial The generation process not only relies on the initial generated images and feedback information, but also takes into account the dynamic adjustment mechanism in the generation process, including image detail retention, color authenticity and content accuracy, to improve the quality of the generated geographic information images. The evaluation function formula is set as follows:

[0081] Q score =ω1Q detail (G initial )+ω2Q color (G initial )+ω3Q content (G initial )

[0082] Among them, Q score Represents a comprehensive quality score obtained by evaluation, Q detail , Q color , Q content They represent the quality assessment functions of image details, color and content respectively, and ω1, ω2, ω3 are weight coefficients.

[0083] Based on the comprehensive quality score, the feedback adjustment module uses a feedback mechanism to adjust the noise vector so that a higher quality image is produced in the next generation process. The adjusted noise vector is calculated as follows:

[0084]

[0085] Among them, z adjusted represents the adjusted noise vector, It's Q score About the noise vector z init The gradient of , indicating how to adjust z init To improve image quality, α is the learning rate, which controls the magnitude of the adjustment.

[0086] The adjusted noise vector is used in the next generation process to generate a new geographic information image, and the above evaluation and adjustment process is repeated. Through multiple iterations, the noise vector is gradually optimized to generate a higher quality geographic information image.

[0087] In summary, the feature extraction and generation system and method of geographic information images described in the present invention are completed.

[0088] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0089] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0090] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0091] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for extracting features and generating geographic information images, characterized in that: The following steps are involved: S1. Standardize the collected geographic information images. The formula for standardization is as follows: Among them, I raw represents the original geographic information image; I is the standardized geographic information image; and They refer to the mean and standard deviation of the pixel values ​​of the original geographic information image respectively; The convolution kernel is used to extract multi-scale features from the standardized geographic information image to generate a multi-scale feature map. A dynamic feature fusion mechanism is designed to calculate the information entropy of the multi-scale feature map. Combined with the image complexity evaluation, an adjustment factor is introduced to adjust the weight coefficient of the feature map to obtain the fused feature map. The calculation formula is as follows: Among them, w si represents the weight coefficient of the feature map of the i-th scale, H(F si ) is the feature map F si Information entropy of E g (F si ) is the feature map F si Gradient entropy of; β and δ are learnable parameters; F merged is a fusion feature map that integrates multi-scale information; n is the total number of scales; S2. Evaluate the information density of the fused feature map. In the process of evaluating the information density of the fused feature map, the information density of the fused feature map is calculated based on the local information volume and the global distribution. The calculation formula is as follows: Among them, D(F merged ) represents the fusion feature map F merged The information density of merged (x, y), Θ) is a local complexity function based on the pixel value and the parameter set Θ; F merged (x, y) is the fusion feature map F merged The pixel value at position (x, y) in the middle; x represents the horizontal position index; y represents the vertical position index; W and H represent the width and height of the feature map respectively; Ω(F merged ) is a function that measures the complexity of the overall feature map; λ is the balance coefficient; Based on the information density of the fused feature map, the scale of the fused feature map is dynamically adjusted. The calculation formula is as follows: F adjusted =Resize(F merged ,S(D(F merged ))) Among them, F adjusted Represents the feature map after scale adjustment; Resize represents the scale adjustment function; S(D(F merged )) is a function based on the information density D(F merged ) to determine the function of the new scale; Based on the scale-adjusted feature map, a geographic information image is generated in combination with a feedback-based generation model. The quality of the geographic information image is evaluated and feedback is provided through a feedback network, thereby adjusting the generation strategy and noise vector.

2. The method for extracting features and generating geographic information images according to claim 1, characterized in that: The S1 specifically includes: Multi-level spatial information is extracted through convolution kernels of different scales, and the maximum pooling operation is applied to obtain a multi-scale feature map.

3. The method for extracting features and generating geographic information images according to claim 1, characterized in that: The S2 specifically includes: A feedback-based generative model is designed to generate geographic information images based on the adjusted feature maps and initial noise vectors.

4. The method for extracting features and generating geographic information images according to claim 3, characterized in that: The S2 specifically includes: In the process of generating geographic information images, a dynamic adjustment mechanism is introduced to evaluate the geographic information images to obtain a comprehensive quality score.

5. The method for extracting features and generating geographic information images according to claim 4, characterized in that: The S2 specifically includes: Based on the comprehensive quality score, the feedback adjustment module adjusts the noise vector through a feedback mechanism and repeats the process of evaluating and adjusting the noise vector for the geographic information image.

6. A system for extracting and generating features of geographic information images, applied to the method for extracting and generating features of geographic information images as claimed in claim 1, characterized in that: Includes the following parts: Data acquisition module, preprocessing module, preliminary feature extraction module, dynamic feature fusion module, adaptive feature scaling module, image generation module, feedback adjustment module; The data acquisition module is used to collect geographic information image data; the data acquisition module is connected to the preprocessing module by means of data transmission; The preprocessing module is used to perform standardization processing on the collected geographic information images; The preprocessing module is connected to the preliminary feature extraction module by means of data transmission; The preliminary feature extraction module is used to analyze the standardized image by applying convolution kernels of different scales, and capture multi-level spatial features to form a multi-scale feature map; the preliminary feature extraction module is connected to the dynamic feature fusion module by means of data transmission; The dynamic feature fusion module is used to dynamically analyze and fuse multi-scale feature maps; the dynamic feature fusion module is connected to the adaptive feature scaling module by means of data transmission; The adaptive feature scaling module is used to adjust the scale of the fused feature map; The adaptive feature scaling module is connected to the image generation module by means of data transmission; The image generation module is used to reconstruct the geographic information image according to the adjusted feature map in combination with the generation model; the image generation module is connected to the feedback adjustment module by means of data transmission; The feedback adjustment module is used to evaluate the quality of the geographic information image reconstructed by the image generation module, provide feedback information, and guide the adjustment of the generation model; the feedback adjustment module is connected to the image generation module by means of data transmission.

Images