High-resolution urban block data extraction and classification system

Abstract

The present invention relates to the field of remote sensing image processing technology, and specifically to a high-resolution urban block data extraction and classification system, including a data acquisition module, a data preprocessing module, a multi-scale feature extraction module, a feature fusion module, a classification recognition module, and a visualization module; wherein: the data acquisition module is used to obtain high-resolution image data; the data preprocessing module is used to preprocess the image data; the multi-scale feature extraction module is used to extract multi-scale feature information in the image based on different spatial scales; the feature fusion module is used to fuse the multi-scale feature information; the classification recognition module is used to classify the fused feature information. The present invention realizes the automation, precision extraction and classification of urban block information through multi-scale feature extraction and fusion technology combined with high-precision image processing methods, greatly improving data processing efficiency and classification accuracy.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing image processing, and in particular to a high-resolution urban block data extraction and classification system. Background Art

[0002] With the acceleration of urbanization, the application of high-resolution remote sensing image data has gradually become the mainstream in the extraction, planning and management of urban block information; traditional urban block data collection usually relies on manual annotation or low-resolution remote sensing images, however, this method has problems such as insufficient data accuracy, high annotation cost, and low processing efficiency; with the development of remote sensing technology, high-resolution images have gradually become an important data source for obtaining urban block information; these high-resolution images can provide richer detailed information, help analyze and identify elements such as buildings, roads, green belts in the city, and are of great value for applications such as urban planning, traffic management, and environmental monitoring; however, existing high-resolution image data extraction methods usually face the problem of how to accurately extract and classify urban block elements in complex scenes, especially in multi-scale and multi-angle environments. How to efficiently and accurately identify and annotate these elements is still a difficult point in technical research.

[0003] Existing high-resolution urban block data extraction methods usually rely on traditional image processing techniques, such as image segmentation and edge detection. However, these methods often have difficulty dealing with complex backgrounds, detail loss, and processing features of different scales in images, resulting in low classification accuracy. In addition, how to effectively stitch images from multiple perspectives and accurately match and fuse features extracted at different scales to further improve classification accuracy remains a technical challenge that needs to be solved urgently. Summary of the invention

[0004] Based on the above objectives, the present invention provides a high-resolution urban block data extraction and classification system.

[0005] The high-resolution urban block data extraction and classification system includes a data acquisition module, a data preprocessing module, a multi-scale feature extraction module, a feature fusion module, a classification and recognition module, and a visualization module; among which:

[0006] Data acquisition module: used to obtain high-resolution image data containing urban block information;

[0007] Data preprocessing module: used to preprocess the image data acquired by the data acquisition module, including denoising, image segmentation and enhancement processing;

[0008] Multi-scale feature extraction module: used to receive pre-processed image data and extract multi-scale feature information in the image based on different spatial scales, including geometric morphological information of buildings, roads and green belts;

[0009] Feature fusion module: connected to the multi-scale feature extraction module, used to receive the extracted multi-scale feature information, perform fusion processing, and generate a unified feature representation;

[0010] Classification and recognition module: connected to the feature fusion module, used to receive the fused feature information and classify it, identify and mark the buildings, roads and green belt elements in the urban blocks, and output the classification results;

[0011] Visualization module: connected with the classification recognition module, receives the classification results and displays the classification annotation data to the user in the form of a two-dimensional map through a graphical interface.

[0012] Optionally, the data acquisition module includes a satellite image acquisition unit, a sensor data acquisition unit and an image stitching unit; wherein:

[0013] Satellite image acquisition unit: used to acquire high-resolution image data covering the target city blocks from satellite or UAV equipment, where the resolution of the high-resolution image data is not less than 0.5 meters;

[0014] Sensor data acquisition unit: used to obtain real-time image data of urban blocks through ground sensors or mobile devices, and the resolution of the real-time image data is not less than 1 meter;

[0015] The image stitching unit is used to stitch image data acquired from multiple viewpoints or multiple devices, and align the images through an automated algorithm to generate complete high-resolution image data covering the target city blocks.

[0016] Optionally, the image stitching unit includes:

[0017] Image preprocessing subunit: used to perform denoising, color correction and brightness adjustment on each acquired image to ensure consistent image quality during the stitching process;

[0018] Feature matching subunit: used to extract feature points of each image based on scale-invariant feature transformation algorithm, and calculate the matching degree of feature points between adjacent images through descriptors;

[0019] Image alignment subunit: used to use homography matrix to realize geometric transformation and alignment between images for matching feature point pairs;

[0020] Image stitching subunit: used to receive the aligned images, seamlessly stitch the overlapping areas of each image, and use the weighted average method to eliminate the seam phenomenon when stitching images;

[0021] Stitching result output subunit: Generates a complete view covering the target city block based on the stitched image data, and the resolution of the image data is not less than 0.5 meters.

[0022] Optionally, the data preprocessing module includes a noise removal unit, an image segmentation unit and an image enhancement unit; wherein:

[0023] Noise removal unit: uses Gaussian filtering algorithm to remove noise from the image data acquired by the data acquisition module;

[0024] Image segmentation unit: used to segment the denoised image data using a threshold-based segmentation algorithm;

[0025] Image enhancement unit: used for performing enhancement processing on the segmented image data, wherein the image enhancement processing includes histogram equalization.

[0026] Optionally, the multi-scale feature extraction module includes a scale decomposition unit, a feature extraction unit and a scale matching unit; wherein:

[0027] Scale decomposition unit: used to perform multi-level scale decomposition on the preprocessed image data, decomposing the image data into sub-images of different spatial scales, wherein the spatial scales include large scale, medium scale and small scale;

[0028] Feature extraction unit: used to extract geometric feature information of the corresponding scale at each scale, including geometric features of buildings, roads and green belts. The geometric morphological information includes the edge, shape, size and position parameters of the object;

[0029] Scale matching unit: used to match the geometric features extracted at different scales to ensure that the feature information extracted from the same object at different scales is consistent, and correct it through geometric constraints.

[0030] Optionally, the feature extraction unit includes:

[0031] Large-scale geometric feature extraction subunit: used to extract long-distance spatial features in images at a large scale. The spatial resolution of the large-scale image ranges from 0.5 meters to 10 meters. The long-distance features are extracted by calculating the edge information of adjacent pixels in the image.

[0032] Mesoscale geometric feature extraction subunit: used to extract medium-distance spatial features in an image at a mesoscale, where the spatial resolution of the mesoscale image ranges from 0.1m to 0.5m; specifically, a region growing algorithm is used to perform feature classification extraction through pixels;

[0033] Small-scale geometric feature extraction subunit: used to extract local or detail features in an image at a small scale. The spatial resolution of the small-scale image ranges from 0.01 meters to 0.1 meters. Feature extraction is performed by calculating the texture feature value of the local area of ​​the image.

[0034] Optionally, the scale matching unit includes:

[0035] Geometric feature extraction subunit: used to extract geometric feature information at each scale to obtain the geometric feature vector at each scale; for each scale, its feature vector is expressed by the following formula: ,in, Indicated in scale The geometric feature vector extracted under For the feature components, is the dimension of the feature, Representative scale;

[0036] Scale matching subunit: used to match geometric features of different scales. Specifically, the similarity metric based on Euclidean distance is used for matching. The formula is:

[0037] ,in, Indicated in scale and scale The Euclidean distance between geometric eigenvectors is and Respectively represent scale and scale Next The value of a feature;

[0038] Geometric constraint correction subunit: used to correct the matching results; specifically, the geometric features are corrected using the transformation matrix. Suppose the transformation matrix is ,expression: ,in, are translation, rotation and scale transformation parameters, is the affine transformation matrix, which is used to align geometric features at different scales.

[0039] Optionally, the feature fusion module includes a feature weighting unit and a feature fusion unit; wherein:

[0040] Feature weighting unit: used to assign weights to different scale features according to the importance of each scale feature. The assigned weights are calculated by the following formula: ,in, Indicates The weight of the scale feature, Indicates Importance measure of scale features, is the scale quantity;

[0041] Feature fusion unit: used to fuse the feature information of each scale by weighted average to generate a unified feature representation. The fusion formula is: ,in, represents the final fused feature representation, Indicates The feature representation of scale, For the The weight of the scale.

[0042] Optionally, the classification recognition module includes a feature grouping unit, a classification model training unit and a classification recognition unit; wherein:

[0043] Feature grouping unit: used to divide the fused feature information into three categories: buildings, roads and green belts according to different spatial distribution and morphological characteristics;

[0044] Classification model training unit: used to build a classification model by training existing urban block data;

[0045] Classification and recognition unit: Use the classification model to classify and recognize the input feature information, and output the classification results. The classification results include the classification labels for each pixel in the urban block, including building labels, road labels and green belt labels.

[0046] Optionally, the visualization module includes a data input unit, a rendering unit, a map projection unit and a user interface unit; wherein:

[0047] Data input unit: used to receive the classification result data output by the classification recognition module, including the category label of each pixel and its corresponding image coordinates;

[0048] Rendering unit: used to map each pixel to a corresponding color or graphic logo according to the category label;

[0049] Map projection unit: used to convert the classification result data from the image coordinate system to the geographic coordinate system;

[0050] User interface unit: used to display the converted two-dimensional map on the graphical interface.

[0051] Beneficial effects of the present invention:

[0052] The present invention, through the collaborative work of multiple modules, can realize high-precision and high-efficiency automatic extraction and classification of urban block information; especially when processing high-resolution remote sensing images, multi-scale feature extraction and fusion technology is adopted, so that the system can simultaneously extract different types of spatial features such as buildings, roads and green belts, thereby significantly improving the accuracy of classification results and overcoming the misjudgment and omission problems caused by single-scale feature extraction in the prior art.

[0053] The present invention, by combining geometric feature information, can accurately identify and mark key elements in urban blocks, thereby improving the automation capability of data processing; and through a graphical interface visualization module, the classification and annotation results are displayed to users in the form of an intuitive two-dimensional map. This visualization method greatly improves the efficiency of urban planning and management, and provides reliable data support for relevant departments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the present invention or 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 in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 A schematic diagram of a system for extracting and classifying urban block data according to an embodiment of the present invention;

[0056] Figure 2 Schematic diagram of a multi-scale feature extraction module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0058] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0059] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0060] like Figure 1-Figure 2 As shown, the high-resolution urban block data extraction and classification system includes a data acquisition module, a data preprocessing module, a multi-scale feature extraction module, a feature fusion module, a classification and recognition module, and a visualization module; wherein:

[0061] Data acquisition module: used to obtain high-resolution image data containing urban block information, which includes spatial information of urban block elements such as buildings, roads, and green belts;

[0062] Data preprocessing module: used to preprocess the image data acquired by the data acquisition module, including denoising, image segmentation and enhancement processing, to generate image data suitable for subsequent analysis;

[0063] Multi-scale feature extraction module: used to receive pre-processed image data and extract multi-scale feature information in the image based on different spatial scales, including geometric morphological information of buildings, roads and green belts;

[0064] Feature fusion module: connected to the multi-scale feature extraction module, used to receive the extracted multi-scale feature information, perform fusion processing, and generate a unified feature representation;

[0065] Classification and recognition module: connected to the feature fusion module, used to receive the fused feature information and classify it, identify and mark the buildings, roads and green belt elements in the urban blocks, and output the classification results;

[0066] Visualization module: connected with the classification recognition module, receives the classification results and displays the classification annotation data to the user in the form of a two-dimensional map through a graphical interface.

[0067] The data acquisition module includes a satellite image acquisition unit, a sensor data acquisition unit and an image stitching unit; wherein:

[0068] Satellite image acquisition unit: used to obtain high-resolution image data covering the target city blocks from satellites or UAV equipment. The resolution of the high-resolution image data is not less than 0.5 meters. The image data includes clear images of urban block elements such as buildings, roads, and green belts;

[0069] Sensor data acquisition unit: used to obtain real-time image data of urban blocks through ground sensors or mobile devices, including information such as buildings, roads, traffic signs, street facilities, etc. The resolution of real-time image data is not less than 1 meter;

[0070] The image stitching unit is used to stitch image data obtained from multiple perspectives or multiple devices, and to align images through an automated algorithm to generate complete high-resolution image data covering the target city blocks, so as to ensure that the stitched images have no obvious seams and maintain the integrity of the high-resolution images. Through the collaborative work of the above units, the data acquisition module can establish accurate spatial relationships between image data from different sources, ensuring that the acquired city block image data has high resolution, integrity and timeliness.

[0071] The image stitching unit includes:

[0072] Image preprocessing subunit: used to perform denoising, color correction and brightness adjustment on each acquired image to ensure consistent image quality during the stitching process;

[0073] Feature matching subunit: It is used to extract the feature points of each image based on the scale-invariant feature transform (SIFT) algorithm, and calculate the matching degree of feature points between adjacent images through descriptors; specifically, for two images, if the feature points in image A and the feature points in image B If the distance is less than the set threshold, it is considered a match. The calculation formula is:

[0074] ,in, Represents the feature points between image A and image B and If the Euclidean distance is less than the preset matching threshold, the two feature points are considered to be matched;

[0075] Image alignment subunit: used to use the homography matrix to achieve geometric transformation and alignment between images for the matching feature point pairs; let the corresponding matching points in image A and image B be and , then the homography matrix H satisfies the following relationship: , where H is The homography matrix represents the image To Image Perspective transformation of

[0076] Image stitching subunit: It is used to receive the aligned images, seamlessly stitch the overlapping areas of each image, use the weighted average method to eliminate the seam phenomenon when stitching images, and set the pixel value in the overlapping area to and , the expression of the spliced ​​image value is:

[0077] ,in, and Image A and image B are at positions The weight of the pixels is dynamically adjusted according to the degree of pixel overlap;

[0078] Stitching result output subunit: Generates a complete view covering the target city block based on the stitched image data, and the resolution of the image data is not less than 0.5 meters.

[0079] The data preprocessing module includes a noise removal unit, an image segmentation unit and an image enhancement unit; wherein:

[0080] Noise removal unit: Use Gaussian filtering algorithm to remove noise from the image data acquired by the data acquisition module; the formula of Gaussian filtering algorithm is: ,in, is the pixel value of the denoised image, For the original image at position The pixel value of is the standard deviation of the Gaussian filter, is the radius of the filter window;

[0081] Image segmentation unit: used to segment the denoised image data using a threshold-based segmentation algorithm; the specific steps of segmentation are as follows:

[0082] Threshold determination step: Calculate the global threshold of the image ,in Determined by Otsu's method to maximize the between-class variance, the expression is: ,in, is the between-class variance, defined as: ,in, and are the probabilities of foreground and background, respectively, and are the average gray values ​​of foreground and background, respectively;

[0083] Image segmentation step: based on a determined threshold , the image is divided into foreground and background, the segmentation formula is: ,in, is the binary image after segmentation, 1 represents foreground and 0 represents background;

[0084] Image enhancement unit: used to enhance the segmented image data. Image enhancement processing includes histogram equalization. The specific formula is: ,in, is the gray value after equalization, The gray value in the original image is The number of pixels, and are the number of rows and columns of the image respectively; through the collaborative work of the above units, the data preprocessing module can effectively remove image noise, accurately segment different areas in the image, and enhance image quality, providing high-quality image data for subsequent multi-scale feature extraction and classification recognition.

[0085] The multi-scale feature extraction module includes a scale decomposition unit, a feature extraction unit and a scale matching unit; wherein:

[0086] Scale decomposition unit: used to perform multi-level scale decomposition on the preprocessed image data, decomposing the image data into sub-images of different spatial scales, including large scale, medium scale and small scale;

[0087] Feature extraction unit: used to extract geometric feature information of the corresponding scale at each scale, including geometric features of buildings, roads and green belts. Geometric morphological information includes the edge, shape, size and position parameters of objects;

[0088] Scale matching unit: used to match the geometric features extracted at different scales to ensure that the feature information extracted at different scales of the same object is consistent, and to correct it through geometric constraints to ensure the consistency of feature position, size and shape between scales; through the above unit, the multi-scale feature extraction module can extract and fuse multi-scale geometric feature information from images of different scales, effectively improving the analysis accuracy of urban block data.

[0089] The feature extraction unit includes:

[0090] Large-scale geometric feature extraction subunit: used to extract long-distance spatial features in images at a large scale. The spatial resolution of large-scale images ranges from 0.5 meters to 10 meters. Long-distance features are extracted by calculating the edge information of adjacent pixels in the image. The edge detection formula is:

[0091] ,in, Indicates at location The edge strength at Represents the pixel grayscale value of the image, and The images are and Directional gradient;

[0092] Mesoscale geometric feature extraction subunit: used to extract medium-distance spatial features in images at a mesoscale. The spatial resolution of mesoscale images ranges from 0.1 meters to 0.5 meters. The regional growing algorithm is used to extract features through pixel classification. The formula is: ,in, For location The average gray value of the area at is the number of pixels in the area, For location The gray value of the pixel at ;

[0093] Small-scale geometric feature extraction subunit: used to extract local or detail features in images at a small scale. The spatial resolution of small-scale images ranges from 0.01 meters to 0.1 meters. Feature extraction is performed by calculating the texture feature values ​​of the local area of ​​the image. The texture feature calculation formula is:

[0094] ,in, Represents the texture feature value of the local area of ​​the image, For the local area The probability distribution of gray levels, is the total number of gray levels; through the above steps, the feature extraction unit can extract geometric information including buildings, roads and green belts at different scales, ensuring that features at different scales can be fully extracted and providing high-quality input data for subsequent feature fusion and classification.

[0095] The scale matching unit includes:

[0096] Geometric feature extraction subunit: It is used to extract geometric feature information at each scale, such as the boundary of the building, the curvature of the road, the shape of the green belt, etc., to obtain the geometric feature vector at each scale; for each scale, its feature vector is expressed by the following formula: ,in, Indicated in scale The geometric feature vector extracted under For the feature components, is the dimension of the feature, Representative scale (large scale, medium scale, small scale);

[0097] Scale matching subunit: used to match geometric features of different scales. Specifically, the similarity metric based on Euclidean distance is used for matching. The formula is:

[0098] ,in, Indicated in scale and scale The Euclidean distance between geometric eigenvectors is and Respectively represent scale and scale Next The value of a feature;

[0099] Geometric constraint correction subunit: used to correct the matching results; specifically, the geometric features are corrected using the transformation matrix. Suppose the transformation matrix is ,expression: ,in, are translation, rotation and scale transformation parameters, is an affine transformation matrix, which is used to align geometric features at different scales to achieve precise matching. Through the above steps, the scale matching unit can accurately match the geometric features extracted at different scales and correct them through geometric constraints to ensure the spatial consistency and accuracy of the final extracted features.

[0100] The feature fusion module includes a feature weighting unit and a feature fusion unit; wherein:

[0101] Feature weighting unit: used to assign weights to features of different scales according to the importance of each scale feature. The assigned weights are calculated using the following formula: ,in, Indicates The weight of the scale feature, Indicates Importance measure of scale features, is the scale quantity;

[0102] Feature fusion unit: used to fuse the feature information of each scale by weighted average to generate a unified feature representation. The fusion formula is: ,in, represents the final fused feature representation, Indicates The feature representation of scale, For the scale weight; through the above process, the feature fusion module can generate a unified set of feature representations, including geometric information from different scales, thereby providing a more accurate feature basis for subsequent classification and recognition.

[0103] The classification and recognition module includes a feature grouping unit, a classification model training unit and a classification and recognition unit; wherein:

[0104] Feature grouping unit: used to divide the fused feature information into three categories: buildings, roads and green belts according to different spatial distribution and morphological characteristics;

[0105] Classification model training unit: used to build a classification model by training existing urban block data. The classification model is trained based on known category labels and extracted feature information, and uses supervised learning algorithms to optimize model parameters and ensure that the model can accurately distinguish the features of buildings, roads, and green belts;

[0106] The specific construction steps are as follows:

[0107] Step 1: Label the image data obtained from different sensors or devices to obtain a training data set containing labels such as buildings, roads, and green belts, where each data point contains a feature vector and the corresponding labels ,in is the index of the data point;

[0108] Step 2: Initialize a classification model , the model can be a support vector machine (SVM), a decision tree, or a deep neural network, etc.;

[0109] Step 3: Use the training data set to adjust the parameters of the classification model through the optimization algorithm. Specifically, the objective function of maximizing the difference between categories is used for training. The objective function is as follows: ,in, are the parameters of the model, is the number of training samples, is the loss function, which represents the difference between the predicted result and the actual label. The cross entropy loss function or the square error loss function is often used. is the regularization parameter, is the complexity of the model to avoid overfitting;

[0110] Step 4: Use gradient descent or other optimization algorithms to minimize the objective function , thus obtaining the optimal model parameters , the formula of the optimization process is as follows: ,in, For the The parameters of the iterations, is the learning rate, is the gradient of the loss function with respect to the model parameters.

[0111] Classification and recognition unit: Use the classification model to classify and recognize the input feature information, and output the classification results. The classification results include the classification labels for each pixel in the urban block, including building labels, road labels and green belt labels. Through the collaborative work of the above units, the classification and recognition module can accurately classify the buildings, roads and green belt elements in the urban block after receiving the fused multi-scale feature information, and output the corresponding classification results for subsequent processing.

[0112] The visualization module includes a data input unit, a rendering unit, a map projection unit and a user interface unit; wherein:

[0113] Data input unit: used to receive the classification result data output by the classification recognition module, including the category label of each pixel (building label, road label and green belt label) and its corresponding image coordinates;

[0114] Rendering unit: used to map each pixel to a corresponding color or graphic identifier according to the category label. The mapping formula is:

[0115] ;

[0116] Among them, For location The pixel color at , and These are the preset colors for buildings, roads, and green belts. For location The category label of

[0117] Map projection unit: used to convert the classification result data from the image coordinate system to the geographic coordinate system. The conversion adopts the equidistant conic projection method. The formula is: and ,in, and is the projection coordinate in the geographic coordinate system, is the radius of the Earth, is the longitude, is the central meridian, is the latitude, is the standard parallel;

[0118] User interface unit: used to display the converted two-dimensional map on the graphical interface. The interface includes legend, zoom and navigation functions so that users can clearly view and operate the classification results; the above units are closely connected through data flows, wherein the data input unit receives the classification result data of the classification recognition module, the rendering unit maps the pixel points into colors or graphics according to the category labels, and the map projection unit converts the image coordinate system data into geographic coordinate system data, and finally displays the classification annotation results in the form of a two-dimensional map on the graphical interface through the user interface unit; the data dependency relationship between the units ensures that the system can accurately and efficiently visualize the classification results to the user.

[0119] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0120] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. High-resolution urban block data extraction and classification system, characterized by: It includes data acquisition module, data preprocessing module, multi-scale feature extraction module, feature fusion module, classification and recognition module and visualization module; among which: Data acquisition module: used to obtain high-resolution image data containing urban block information; Data preprocessing module: used to preprocess the image data acquired by the data acquisition module, including denoising, image segmentation and enhancement processing; Multi-scale feature extraction module: used to receive pre-processed image data and extract multi-scale feature information in the image based on different spatial scales, including geometric morphological information of buildings, roads and green belts; Feature fusion module: connected to the multi-scale feature extraction module, used to receive the extracted multi-scale feature information, perform fusion processing, and generate a unified feature representation; Classification and recognition module: connected to the feature fusion module, used to receive the fused feature information and classify it, identify and mark the buildings, roads and green belt elements in the urban blocks, and output the classification results; Visualization module: connected with the classification recognition module, receives the classification results and displays the classification annotation data to the user in the form of a two-dimensional map through a graphical interface; The multi-scale feature extraction module includes a scale decomposition unit, a feature extraction unit and a scale matching unit; wherein: Scale decomposition unit: used to perform multi-level scale decomposition on the preprocessed image data, decomposing the image data into sub-images of different spatial scales, wherein the spatial scales include large scale, medium scale and small scale; Feature extraction unit: used to extract geometric feature information of the corresponding scale at each scale, including geometric features of buildings, roads and green belts. The geometric morphological information includes the edge, shape, size and position parameters of the object; Scale matching unit: used to match geometric features extracted at different scales to ensure that the feature information extracted from the same object at different scales is consistent, and correct it through geometric constraints; The scale matching unit comprises: Geometric feature extraction subunit: used to extract geometric feature information at each scale to obtain the geometric feature vector at each scale; for each scale, its feature vector is expressed by the following formula: ,in, Indicated in scale The geometric feature vector extracted under For the feature components, is the dimension of the feature, Representative scale; Scale matching subunit: used to match geometric features of different scales. Specifically, the similarity metric based on Euclidean distance is used for matching. The formula is: ,in, Indicated in scale and scale The Euclidean distance between geometric eigenvectors is and Respectively represent scale and scale Next The value of a feature; Geometric constraint correction subunit: used to correct the matching results; specifically, the geometric features are corrected using the transformation matrix. Suppose the transformation matrix is ,expression: ,in, are translation, rotation and scale transformation parameters, is the affine transformation matrix, which is used to align geometric features at different scales.

2. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The data acquisition module includes a satellite image acquisition unit, a sensor data acquisition unit and an image stitching unit; wherein: Satellite image acquisition unit: used to acquire high-resolution image data covering the target city blocks from satellite or UAV equipment, and the resolution of the high-resolution image data is not less than 0.5 meters; Sensor data acquisition unit: used to obtain real-time image data of urban blocks through ground sensors or mobile devices, and the resolution of the real-time image data is not less than 1 meter; The image stitching unit is used to stitch image data acquired from multiple viewpoints or multiple devices, and align the images through an automated algorithm to generate complete high-resolution image data covering the target city blocks.

3. The high-resolution urban block data extraction and classification system according to claim 2 is characterized in that: The image stitching unit comprises: Image preprocessing subunit: used to perform denoising, color correction and brightness adjustment on each acquired image to ensure consistent image quality during the stitching process; Feature matching subunit: used to extract feature points of each image based on scale-invariant feature transformation algorithm, and calculate the matching degree of feature points between adjacent images through descriptors; Image alignment subunit: used to use homography matrix to realize geometric transformation and alignment between images for matching feature point pairs; Image stitching subunit: used to receive the aligned images, seamlessly stitch the overlapping areas of each image, and use the weighted average method to eliminate the seam phenomenon when stitching images; Stitching result output subunit: Generates a complete view covering the target city block based on the stitched image data, and the resolution of the image data is not less than 0.5 meters.

4. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The data preprocessing module includes a noise removal unit, an image segmentation unit and an image enhancement unit; wherein: Noise removal unit: uses Gaussian filtering algorithm to remove noise from the image data acquired by the data acquisition module; Image segmentation unit: used to segment the denoised image data using a threshold-based segmentation algorithm; Image enhancement unit: used for performing enhancement processing on the segmented image data, wherein the image enhancement processing includes histogram equalization.

5. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The feature extraction unit comprises: Large-scale geometric feature extraction subunit: used to extract long-distance spatial features in images at a large scale. The spatial resolution of large-scale images ranges from 0.5 meters to 10 meters. Long-distance features are extracted by calculating the edge information of adjacent pixels in the image. Mesoscale geometric feature extraction subunit: used to extract medium-distance spatial features in images at a mesoscale. The spatial resolution of mesoscale images ranges from 0.1 meters to 0.5 meters. The regional growing algorithm is used to extract features through pixel classification. Small-scale geometric feature extraction subunit: used to extract local or detail features in images at a small scale. The spatial resolution of small-scale images ranges from 0.01 meters to 0.1 meters. Feature extraction is performed by calculating the texture feature values ​​of the local area of ​​the image.

6. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The feature fusion module includes a feature weighting unit and a feature fusion unit; wherein: Feature weighting unit: used to assign weights to different scale features according to the importance of each scale feature. The assigned weights are calculated by the following formula: ,in, Indicates The weight of the scale feature, Indicates Importance measure of scale features, is the scale quantity; Feature fusion unit: used to fuse the feature information of each scale by weighted average to generate a unified feature representation. The fusion formula is: ,in, represents the final fused feature representation, Indicates The feature representation of scale, For the The weight of the scale.

7. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The classification and recognition module includes a feature grouping unit, a classification model training unit and a classification and recognition unit; wherein: Feature grouping unit: used to divide the fused feature information into three categories: buildings, roads and green belts according to different spatial distribution and morphological characteristics; Classification model training unit: used to build a classification model by training existing urban block data; Classification and recognition unit: Use the classification model to classify and recognize the input feature information, and output the classification results. The classification results include the classification labels for each pixel in the urban block, including building labels, road labels and green belt labels.

8. The high-resolution urban block data extraction and classification system according to claim 1, characterized in that: The visualization module includes a data input unit, a rendering unit, a map projection unit and a user interface unit; wherein: Data input unit: used to receive the classification result data output by the classification recognition module, including the category label of each pixel and its corresponding image coordinates; Rendering unit: used to map each pixel to a corresponding color or graphic logo according to the category label; Map projection unit: used to convert the classification result data from the image coordinate system to the geographic coordinate system; User interface unit: used to display the converted two-dimensional map on the graphical interface.

Images