Method for extracting rural residential areas and determining building density with tile data

Abstract

The present application relates to the technical field of optical remote sensing image application, and proposes a method for extracting rural residential areas and determining building density by using tile data, which extracts rural residential areas and determines building density by combining map tile data and deep learning technology, selects 14-level tile images for extracting the distribution of rural residential areas through the tile resolution characteristics and the actual characteristics of rural residential areas, uses the model trained based on SegFormer to extract the distribution of rural residential areas after obtaining the 14-level tile images, selects 18-level tile images for extracting the distribution of rural buildings, uses the model trained based on DeepLabV3+ to extract rural buildings, and selects medium-resolution level images for interpretation, thereby improving the accuracy and efficiency of rural residential area extraction and building density determination.

Description

Technical Field

[0001] This invention relates to the field of optical remote sensing image application technology, specifically to a method for extracting rural residential areas and determining building density using tile data. Background Technology

[0002] Rural residential areas are an important part of rural geography. The ability to quickly and accurately extract the distribution of rural residential areas and determine their building density is of great significance in areas such as rural development planning, village consolidation, and population census. Traditional methods for extracting residential areas mainly rely on manual ground surveys, which suffer from problems such as large workload, high cost, long cycle, and untimely information acquisition, making it difficult to meet the requirements of high efficiency.

[0003] Remote sensing image interpretation can significantly reduce the number of staff and workload, but large-scale interpretation using raster imagery requires sophisticated electronic hardware and is inconvenient to operate. Currently, tile-based remote sensing image services are becoming increasingly common, offering convenient online viewing and access to data at different resolutions. For example, tile mapping services are an online service that follows a pyramid model, such as... Figure 1 As shown, the tile map pyramid model is a multi-resolution hierarchical model. From the bottom to the top of the tile pyramid, the resolution decreases, but the geographical area it represents remains unchanged. For the same geographical area, the higher the level of the tile pyramid, the lower the resolution and the smaller the data volume; the lower the level of the tile pyramid, the higher the resolution and the larger the data volume.

[0004] In recent years, the new generation of machine learning methods, represented by deep learning, has made breakthrough progress in computer vision-related fields and has gradually penetrated into the research of the intersection of computer vision and satellite remote sensing technology. Its application has made it possible to automate and efficiently process satellite image data, greatly improving the accuracy and efficiency of information extraction.

[0005] Therefore, how to combine map tile data and deep learning technology to extract rural residential areas and determine building density is a key technical problem that urgently needs to be solved. Summary of the Invention

[0006] To combine map tile data and deep learning technology to improve the accuracy and efficiency of rural residential area extraction and building density determination, this invention proposes the following technical solution:

[0007] Methods for extracting rural residential area data and determining building density using tile data, such as Figure 2 As shown, it includes the following steps:

[0008] Step 1: Preprocessing remote sensing images: Preprocess the sub-meter level satellite remote sensing images used, setting the corresponding resolutions to 14-level and 18-level tiles respectively, so as to obtain remote sensing images that can be used for the two types of land features: rural residential areas and rural buildings.

[0009] Step 2: Construct a rural residential area dataset and a rural building dataset for deep learning training;

[0010] Step 3: Train deep learning extraction models, including rural residential area extraction models and rural building extraction models;

[0011] Step 4: Crawl 14-level tile images within the prediction area. By combining the tiles, crop the rural residential area remote sensing image into image blocks of the same size as the rural residential area dataset, record their positions in the original image, and generate a remote sensing image that can be predicted.

[0012] In steps 5 and 4, the image patches are input into the rural residential area extraction model to obtain the rural residential area prediction results and obtain the contour vector patch that conforms to the actual boundary of the rural residential area.

[0013] Step 6: Based on the outline vector pattern of the actual boundary of the rural residential area in Step 5, crawl the 18-level tile image of the prediction area. By combining the tiles, crop the remote sensing image of the rural residential area into image blocks of the same size as the rural building dataset, record their positions in the original image, and generate a remote sensing image that can be predicted.

[0014] In steps 7 and 6, the image patches are input into the rural building extraction model to obtain the rural building prediction results and obtain the contour vector pattern that conforms to the actual boundary of the rural building.

[0015] Step 8: Determine the density level of rural buildings based on the rural building prediction results, assign the resulting rural building density level to each rural residential area, and obtain a vector file containing the distribution of rural residential areas and the building density attributes of each residential area.

[0016] Technical effects:

[0017] This invention combines map tile data and deep learning technology to extract rural residential areas and determine building density. Based on tile resolution characteristics and actual features of rural residential areas, it selects 14-level tile images to extract the distribution of rural residential areas. After obtaining the 14-level tile images, it uses a model trained on SegFormer to extract the distribution of rural residential areas and converts it into patch data.

[0018] Meanwhile, considering the small footprint of rural buildings, this invention selects high-resolution remote sensing imagery as the data source. Combining tile resolution characteristics and the actual features of rural buildings, 18-level tile imagery is chosen to extract the distribution of rural buildings. Based on the extent of rural residential area patches, 18-level tile imagery is selectively acquired for that area. This method excludes areas such as farmland and woodland where no residential areas exist, reducing the workload of prediction, minimizing hardware resource consumption, and greatly improving work efficiency. A model trained using DeepLabV3+ is used to extract rural buildings, and the density analysis results of rural residential area buildings are calculated.

[0019] In the process of large-scale rural residential area extraction, due to the large area involved, using high-resolution image data requires significant computational resources and results in a long prediction time. Given the distinct characteristics of rural residential areas, this invention innovatively selects medium-resolution imagery for interpretation, thereby improving the accuracy and efficiency of rural residential area extraction and building density determination. The method proposed in this invention can extract details of the distribution of some rural residential areas, such as... Figure 5 and Figure 6 As shown.

[0020] To verify the effectiveness of this invention in extracting residential areas using Jilin-1 sub-meter-level satellite imagery and its slice data, a specific region was selected for experimental verification. All residential areas within this region were manually delineated as accuracy verification samples. HRNet-OCR and DeepLabV3+ methods were compared, demonstrating that the proposed method has good performance. Producer accuracy (PA) and user accuracy (UA) were used for comparison, and the comparison accuracy is shown in the table below.

[0021] method PA UA HRNet-OCR 86.87％ 83.30％ DeepLabV3+ 84.21％ 82.60％ This invention, SegFormer 88.62％ 85.74％

[0022] While HRNet-OCR and DeepLabV3+ offer advantages in segmentation accuracy, their complex structures and high computational complexity may make them unsuitable for real-time applications or resource-constrained scenarios. SegFormer, based on a Transformer encoder structure, effectively models long-range dependencies, which is particularly important for residential area segmentation. This is because the boundaries and structures of residential areas can span a large spatial area, and SegFormer can handle residential areas of different scales while maintaining computational efficiency. Therefore, the method proposed in this invention achieves the highest numerical accuracy in the accuracy verification, demonstrating practical value. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of a pyramid model made of tile maps.

[0024] Figure 2 This is a flowchart of the overall process of this invention.

[0025] Figure 3 This is a block diagram of the network structure of the rural residential area extraction model.

[0026] Figure 4 This is a network structure diagram of the rural building extraction model.

[0027] Figure 5 This is the original map showing the distribution of rural residential areas within a certain predicted region.

[0028] Figure 6 This is a schematic diagram of the extraction results from rural residential areas after the present invention. Detailed Implementation

[0029] To better understand the technical solution of the present invention, the embodiments provided by the present invention are described in detail below with reference to the accompanying drawings, but the implementation of the present invention is not limited thereto.

[0030] Step 1: Preprocessing remote sensing images: The sub-meter resolution satellite remote sensing images used are preprocessed and set to the corresponding resolutions of 14-level and 18-level tiles respectively, so as to obtain remote sensing images that can be used for two types of land features: rural residential areas and rural buildings. The preprocessing process includes atmospheric correction, geometric correction, image fusion and image registration.

[0031] Step 2: Construct a rural residential area dataset and a rural building dataset for deep learning training;

[0032] On the low-resolution remote sensing image of rural residential areas obtained in step 1, the outlines of rural residential areas are manually delineated. The distribution of rural residential areas in the image should include the following characteristics: dense, normal distribution, sparse, and scattered. During the manual delineation process, the outlines of the residential areas should be closely matched as much as possible. The image data and labeled data are then divided into blocks to obtain the initial rural residential area dataset.

[0033] On the high-resolution rural building remote sensing image obtained in step 1, the outlines of rural buildings are manually delineated. During the selection of remote sensing images, data from areas containing a wide range of building types are chosen to ensure that the dataset includes basic rural building types such as single-story, multi-story, dense, and sparse buildings. During the manual delineation process, the outlines should be closely approximated to the outer edges of the buildings. The image data and labeled data are then divided into blocks to obtain the initial rural building dataset.

[0034] After data augmentation of the initial rural residential area dataset and the initial rural residential area dataset, a rural residential area dataset and a rural building dataset are obtained; further, the data augmentation includes horizontal flipping, vertical flipping, random rotation, and random combinations of affine transformation methods.

[0035] Step 3: Train deep learning extraction models, including rural residential area extraction models and rural building extraction models:

[0036] The network structure diagram of the rural residential area extraction model is as follows: Figure 3 As shown, in rural residential area extraction, the SegFormer model structure is selected as the basis for training the extraction model and prediction. SegFormer has a large effective receptive field, which enables the model to capture more contextual information, thereby improving the segmentation accuracy and making it more suitable for rural residential area extraction tasks. A hierarchical Transformer Block that does not require positional encoding is used in the process, enabling the capture of higher resolution features and thus obtaining richer detail information. Simultaneously, a lightweight, full MLP (Multi-Layer Perceptron) decoder is used, which does not require complex computation or high computational resources, and can obtain effective feature representation.

[0037] The specific process is as follows:

[0038] 3.1a. Input remote sensing images of rural residential areas Where H and W are the height and width of the image, respectively, and 3 represents the RGB three channels;

[0039] The input image is segmented into non-overlapping patches, each patch being P×P in size. The image is then segmented into... There are 10 patches, each patch is flattened into a vector, and the patch embedding is obtained through linear projection:

[0040]

[0041] Where C is the dimension of Patch Embedding, Flatten means flattening the Patch, and Linear means linear projection;

[0042] 3.1b. Based on the SegFormer model structure, the Patch Embedding is passed through multiple stages of Transformer Block in the encoder. The resolution of the output feature map gradually decreases and the number of channels gradually increases in each stage. The encoder can capture feature information from low level to high level through Transformer Block and generate multi-scale feature maps.

[0043] The encoder is the core of SegFormer. It adopts a hierarchical Transformer structure and can capture multi-scale feature information. It consists of multiple stages, each containing several Transformer Blocks.

[0044] This embodiment is divided into 4 stages (Block 1 to Block 4), as follows: Figure 3 As shown, the resolution of the output feature map gradually decreases and the number of channels gradually increases in each stage.

[0045] Phase 1 output

[0046] Phase 2 Output

[0047] Phase 3 Output

[0048] Phase 4 Output

[0049] Through multiple stages of Transformer Blocks, the encoder can capture feature information from low to high levels and generate multi-scale feature maps {F1, F2, F3, F4}, which correspond to the outputs of the above four stages.

[0050] Furthermore, the Transformer Block consists of an MHSA self-attention mechanism and an FFN feedforward neural network.

[0051] MHSA self-attention mechanism:

[0052]

[0053] Where Q, K, and V represent Query, Key, and Value, respectively, d k It is the dimension of the key;

[0054] FFN feedforward neural network:

[0055] FFN(x) = Linear(GELU(Linear(x))), where GELU is the activation function and Linear represents linear projection.

[0056] 3.1c. The decoder consists of an MLP layer, an upsampling layer, and a 1x1 convolutional layer. It can fuse the multi-scale feature maps extracted by the encoder and perform channel adjustment and feature enhancement on the multi-scale feature maps {F1, F2, F3, F4} through the MLP layer.

[0057] F i =MLP(F i ), i∈{1,2,3,4}

[0058] Then, the adjusted feature maps are upsampled to the same resolution (usually 1 / 4 the size of the input image) and stitched together:

[0059] Ffused =Concat(Upsample(F1'),Upsample(F2'),Upsample(F3'),Upsample(F4'))

[0060] Finally, the final segmentation result Y is generated through a lightweight MLP layer and a 1x1 convolutional layer:

[0061]

[0062] Among them, F fused This is a multi-scale fused feature map, where N is the number of categories;

[0063] Each pixel corresponds to a class probability distribution. The predicted class of each pixel is obtained through the argmax operation:

[0064] Prediction = argmax(Y),

[0065] The trained rural residential area extraction model is used for rural residential area range prediction.

[0066] Rural building extraction model network structure diagram as follows Figure 4 As shown, the specific process is as follows:

[0067] 3.2a. Input remote sensing images of rural residential areas H and W represent the height and width of the image, respectively, and 3 represents the RGB three channels. The input image is normalized to scale the pixel values ​​to the range of [0,1] to facilitate model training.

[0068] 3.2b. Based on the DeepLabV3+ model structure, the encoder passes through the SENet154 backbone network. Furthermore, its core feature is the introduction of the SE (Squeeze-and-Excitation) module, which adaptively adjusts the importance of channel features to extract multi-level feature information from the input image. This includes the Squeeze module, which compresses the spatial information of each channel into a scalar through Global Average Pooling (GAP).

[0069]

[0070] Where, x c (i,j) represents the feature value of the c-th channel at position (i,j), z c It is the compressed value of the c-th channel.

[0071] The Excitation module generates channel attention weights using a fully connected layer (ReLU) and an activation function (Sigmoid).

[0072] s=σ(W2·ReLU(W1·z))

[0073] Where W1 and W2 are the weights of the fully connected layer, σ is the Sigmoid function, and s is the channel attention weight.

[0074] Multiplying the channel attention weights by the original features yields the recalibrated features:

[0075] SENet154 outputs six feature maps at different scales. Among them, the feature map with a scale of 1 / 16 and 2048 channels is used as the main feature and is input into the ASPP module for information processing.

[0076] Furthermore, the ASPP module uses four parallel dilated convolution branches, each employing a different dilation rate and a global average pooling branch for multi-scale feature extraction. In this embodiment, the different dilation rates are 1, 6, 12, and 18. A dilation rate of 1 captures local detail information; dilation rates of 6, 12, and 18 capture a wider range of contextual information; and global average pooling captures global contextual information.

[0077] The formula for dilated convolution is:

[0078] Where r is the dilatation rate and w(m,n) is the convolution kernel weight;

[0079] The extracted multi-scale feature maps are concatenated and channel adjustment is performed using a 1x1 convolution:

[0080] F ASPP =Conv 1×1 (Concat(F1,F6,F 12 ,F 18 ,F GAP )),

[0081] Among them, F1, F6, F 12 F 18 These are characteristic maps of different void ratios, F GAP It is a global average pooling feature map.

[0082] 3.2c. In the decoder, high-level features are fused with low-level features. In this embodiment, the low-level feature is a low-level feature map F extracted from the SENet154 backbone network with a scale of 1 / 4 and 256 channels. low The high-level feature is the feature map F output by the ASPP module. ASPP Upsampled to a 1 / 4 scale; low-level and high-level features are concatenated and fused using a convolutional layer:

[0083] F fused′ =Conv 3×3 (Concat(F low Upsample(F ASPP )))

[0084] The fused features are mapped to the class space through an upsampling layer and a 1x1 convolutional layer, and the final segmentation result Z is output.

[0085]

[0086] Among them, F fused′ To fuse features, N is the number of classes, and the predicted class for each pixel is obtained through the argmax operation:

[0087] Prediction = argmax(Z),

[0088] The trained rural building extraction model is used for rural building prediction.

[0089] Step 4: Crawling 14-level tile images:

[0090] The vector file of the area to be predicted is parsed to determine its geographic boundaries. Based on the boundaries, tile crawling technology is used to crawl 14 levels of tile images within the area from left to right and from top to bottom, and the position of each tile is recorded. The raster image file of the predicted area is re-stitched based on the recorded tile positions, and geographic information is assigned according to the predicted range. The remote sensing image is cropped into image blocks of the same size as the dataset data, and their positions in the original image are recorded. In this embodiment, the overlap rate is set to 0.05.

[0091] Step 5: Input the image patch from Step 4 into the rural residential area extraction model to obtain the rural residential area prediction result and obtain the contour vector patch that conforms to the actual boundary of the rural residential area.

[0092] Furthermore, the raster-to-vector function in GDAL was used to convert the raster extraction results of the residential area into vector results, retaining the residential area portion and deleting the background value portion. The extracted residential area results were then morphologically optimized, and the Douglas-Peucker algorithm was used to reduce the amount of point data. Additionally, jagged edges in the contour were smoothed to obtain the final contour vector pattern that conforms to the actual boundary of the rural residential area.

[0093] Step 6: Extract 18 levels of tile imagery:

[0094] Based on the outline vector map of the actual boundary of the rural residential area in step 5, the geographical boundary it contains is analyzed. According to the range, the tile crawling technology is used to crawl 18 levels of tile images in the prediction area in the order from left to right and from top to bottom, and the position of each tile is recorded. The recorded tile positions are then used to re-stitch the raster image file of the prediction area. The remote sensing image of the rural residential area is cropped into image blocks of the same size as the rural building dataset by combining the tiles. The positions of these blocks in the original image are recorded to generate a remote sensing image that can be predicted. In this embodiment, the overlap rate is set to 0.05.

[0095] In steps 7 and 6, the image blocks are input into the rural building extraction model to obtain the rural building prediction results. All the prediction results are re-stitched into a result image of the same size as the original remote sensing image according to the recorded location information to obtain the final building raster prediction results. The building raster prediction results are converted into vector files to obtain the outline vector pattern that conforms to the actual boundary of the rural buildings.

[0096] Furthermore, each building patch can be optimized, including boundary optimization, removal of redundant points, filling of recessed lines, smoothing of burrs, and optimization of right angles, to make the building patch more closely resemble the actual building form.

[0097] Step 8: Determine the density level of rural buildings based on the rural building prediction results, assign the resulting rural building density level to each rural residential area, and obtain a vector file containing the distribution of rural residential areas and the building density attributes of each residential area.

[0098] In this embodiment, the density of rural buildings is defined as five sparseness types: scattered, sparse, normal, dense, and extremely dense. The area of ​​each rural residential area is calculated; the number of buildings in each rural residential area is recorded, and the area of ​​each building is calculated; the ratio of the building area to the area of ​​the residential area in each rural residential area is calculated.

[0099] Rural residential areas with an area of ​​less than 2,000 square meters, fewer than 10 buildings, or an area ratio of less than 0.1 are classified as scattered and sparse.

[0100] Rural residential areas with an area ratio greater than or equal to 0.1 and less than 0.3 are classified as sparse-density areas.

[0101] Rural residential areas with an area ratio greater than or equal to 0.3 and less than 0.55 are classified as normal density areas.

[0102] Rural residential areas with an area ratio greater than or equal to 0.55 and less than 0.75 are classified as dense-sparse areas.

[0103] Rural residential areas with an area ratio of 0.75 or higher are classified as extremely dense and sparse.

[0104] All content not described in detail in this specification belongs to the prior art known to those skilled in the art. Furthermore, for those skilled in the art, there will be changes in specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for extracting rural residential area data and determining building density using tile data, characterized in that, Includes the following steps: Step 1: Preprocessing remote sensing images: Preprocess the sub-meter level satellite remote sensing images used, setting the corresponding resolutions to 14-level and 18-level tiles respectively, so as to obtain remote sensing images that can be used for the two types of land features: rural residential areas and rural buildings. Step 2: Construct a rural residential area dataset and a rural building dataset for deep learning training; Step 3: Train deep learning extraction models, including rural residential area extraction models and rural building extraction models; Step 4: Crawl 14-level tile images within the prediction area. By combining the tiles, crop the rural residential area remote sensing image into image blocks of the same size as the rural residential area dataset, record their positions in the original image, and generate a remote sensing image that can be predicted. In steps 5 and 4, the image patches are input into the rural residential area extraction model to obtain the rural residential area prediction results and obtain the contour vector patch that conforms to the actual boundary of the rural residential area. Step 6: Based on the outline vector pattern of the actual boundary of the rural residential area in Step 5, crawl the 18-level tile image of the prediction area. By combining the tiles, crop the remote sensing image of the rural residential area into image blocks of the same size as the rural building dataset, record their positions in the original image, and generate a remote sensing image that can be predicted. In steps 7 and 6, the image patches are input into the rural building extraction model to obtain the rural building prediction results and obtain the contour vector pattern that conforms to the actual boundary of the rural building. Step 8: Determine the density level of rural buildings based on the rural building prediction results, assign the resulting rural building density level to each rural residential area, and obtain a vector file containing the distribution of rural residential areas and the building density attributes of each residential area.

2. The method for extracting rural residential areas and determining building density using tile data according to claim 1, characterized in that, The specific process for extracting the model for rural residential areas in step 3 is as follows: 3.1a. Input remote sensing images of rural residential areas Where H and W are the height and width of the image, respectively, and 3 represents the RGB three channels; The input image is segmented into non-overlapping patches, each patch being P×P in size. The image is then segmented into... There are 10 patches, each patch is flattened into a vector, and the patch embedding is obtained through linear projection: Where C is the dimension of Patch Embedding, Flatten means flattening the Patch, and Linear means linear projection; 3.1b. Based on the SegFormer model structure, the Patch Embedding is passed through multiple stages of Transformer Block in the encoder. The resolution of the output feature map gradually decreases and the number of channels gradually increases in each stage. The encoder can capture feature information from low level to high level through the Transformer Block and generate multi-scale feature maps. 3.1c. The decoder consists of an MLP layer, an upsampling layer, and a 1x1 convolutional layer. It can fuse the multi-scale feature maps extracted by the encoder and output the final segmentation result Y. Among them, F fused This is a multi-scale fused feature map, where N is the number of categories; Each pixel corresponds to a class probability distribution. The predicted class of each pixel is obtained through the argmax operation: Prediction = argmax(Y), The trained rural residential area extraction model is used for rural residential area range prediction.

3. The method for extracting rural residential areas and determining building density using tile data according to claim 2, characterized in that, The TransformerBlock consists of an MHSA self-attention mechanism and an FFN feedforward neural network. MHSA self-attention mechanism: Where Q, K, and V represent Query, Key, and Value, respectively, d k It is the dimension of the key; FFN feedforward neural network: FFN(x) = Linear(GELU(Linear(x))), where GELU is the activation function and Linear represents linear projection.

4. The method for extracting rural residential areas and determining building density using tile data according to claim 1, characterized in that, The specific process for extracting the rural building model in step 3 is as follows: 3.2a. Input remote sensing images of rural residential areas Where H and W are the height and width of the image, respectively, 3 represents the RGB three channels, and the input image is normalized. 3.2b. Based on the DeepLabV3+ model structure, the SENet154 backbone network is used in the encoder to extract multi-level feature information from the input image. SENet154 outputs 6 feature maps at different scales, among which the feature map with a scale of 1 / 16 and 2048 channels is used as the main feature and input into the ASPP module for information processing. 3.2c. In the decoder, high-level features are fused with low-level features. The low-level features are low-level feature maps F extracted from the SENet154 backbone network with a scale of 1 / 4 and 256 channels. low The high-level feature is the feature map F output by the ASPP module. ASPP Upsampled to a 1 / 4 scale; low-level and high-level features are concatenated and fused using a convolutional layer: F fused′ =Conv 3×3 (Concat(F low ,Upsample(F ASPP ))) The fused features are mapped to the class space through an upsampling layer and a 1×1 convolutional layer, and the final segmentation result Z is output. Among them, F fused′ To fuse features, N is the number of classes, and the predicted class for each pixel is obtained through the argmax operation: Prediction = argmax(Z), The trained rural building extraction model is used for rural building prediction.

5. The method for extracting rural residential areas and determining building density using tile data according to claim 4, characterized in that, The information processing in the ASPP module is specifically as follows: The ASPP module uses four parallel dilated convolution branches with different dilation rates of 1, 6, 12, and 18, as well as a global average pooling branch, for multi-scale feature extraction. The formula for dilated convolution is: Where r is the dilatation rate and w(m,n) is the convolution kernel weight; The extracted multi-scale feature maps are concatenated and channel adjustment is performed using a 1x1 convolution: F ASPP =Conv 1×1 (Concat(F1,F6,F 12 ,F 18 ,F GAP )), Among them, F1, F6, F 12 F 18 These are characteristic maps of different void ratios, F GAP It is a global average pooling feature map.

6. The method for extracting rural residential areas and determining building density using tile data according to claim 1, characterized in that, In step 5, the raster-to-vector function is used to convert the raster extraction results of the residential area into vector results, retaining the residential area part and deleting the background value part; the extracted residential area results are morphologically optimized, the Douglas-Puk algorithm is used to reduce the amount of point data, and the burrs in the contour are smoothed to obtain the final contour vector pattern that conforms to the actual boundary of the rural residential area.

7. The method for extracting rural residential areas and determining building density using tile data according to claim 1, characterized in that, In step 7, each building outline vector graphic patch is optimized, including boundary optimization, removal of redundant points, filling of concave line segments, smoothing of burrs, and optimization of right-angled edges, so that the building graphic patch is closer to the actual building shape.

8. The method for extracting rural residential areas and determining building density using tile data according to claim 1, characterized in that, In step 8, the density of rural buildings is defined as five types of sparseness: scattered, sparse, normal, dense, and extremely dense, and the area of ​​each rural residential area is calculated. Record the number of buildings in each rural residential area and calculate the area of ​​each building; Calculate the ratio of building area to residential area in each rural residential area; Rural residential areas with an area of ​​less than 2,000 square meters, fewer than 10 buildings, or an area ratio of less than 0.1 are classified as scattered and sparsely populated. Rural residential areas with an area ratio greater than or equal to 0.1 and less than 0.3 are classified as sparse-density areas; Rural residential areas with an area ratio greater than or equal to 0.3 and less than 0.55 are classified as normal density areas; Rural residential areas with an area ratio greater than or equal to 0.55 and less than 0.75 are classified as dense-sparse areas; Rural residential areas with an area ratio of 0.75 or higher are classified as extremely dense and sparse.

Images