CN114494897A - High-resolution remote sensing image road extraction method fused with patch shape index - Google Patents
High-resolution remote sensing image road extraction method fused with patch shape index Download PDFInfo
- Publication number
- CN114494897A CN114494897A CN202111589580.9A CN202111589580A CN114494897A CN 114494897 A CN114494897 A CN 114494897A CN 202111589580 A CN202111589580 A CN 202111589580A CN 114494897 A CN114494897 A CN 114494897A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- classification
- road
- sensing image
- patch
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a high-resolution remote sensing image road extraction method fusing patch shape indexes, which comprises the following steps of (1) collecting and acquiring a high-resolution remote sensing image and topographic data of a research area and carrying out data preprocessing; (2) extracting the classification characteristic parameters of the remote sensing image; (3) constructing a high-resolution remote sensing image road automatic identification algorithm model fused with patch shape indexes, constructing a random forest algorithm model by using multiple characteristic variables extracted in the step (2), extracting importance information of the characteristic variables, selecting the characteristic variables according to importance degrees of the characteristic variables, reducing dimensions of high-dimensional data, and further obtaining a classification scheme with optimal precision and efficiency; (4) and remote sensing image classification and automatic extraction of road information. The invention can effectively distinguish buildings and roads and ensure that road information can be accurately and efficiently extracted.
Description
Technical Field
The invention relates to the field of high-resolution remote sensing image road identification and extraction, in particular to an automatic remote sensing image road information extraction method fusing patch shape indexes.
Background
Roads are used as important basic geographic information data and have important roles in map making and updating, road navigation, city planning, emergency disaster relief and the like. Therefore, the rapid identification and extraction of roads have important research value.
In recent years, with the increasing popularity of sub-meter high resolution satellite data such as SPOT, QuickBird, high-resolution two, high-view one, and beijing two, the extraction of roads by using remote sensing images has gradually become a main way to extract road information. At present, the extraction of road information based on remote sensing images mainly comprises the following steps: (1) the conventional classification method. The method mainly comprises the steps of identifying and extracting road information by creating corresponding features, wherein the common features comprise spectral features, textural features, geometric features and the like; (2) provided is a deep learning method. From the data perspective, the method automatically excavates the road image deep learning characteristics according to the road image samples provided in advance by utilizing the powerful generalization capability of deep learning, the fitting capability to any function and the extremely high stability, and realizes the prediction of the pixel-level road image probability value by utilizing the discriminant function so as to explore the internal relation of the road image. (3) An object-oriented segmentation method. The method comprises the steps of establishing a corresponding fuzzy discrimination rule by using related characteristic attributes such as spectrum, shape, texture, spatial position and the like of a target land class, segmenting a remote sensing image, forming pixels with the same characteristics into a homogeneous object, and further performing image classification and information extraction.
At present, the problems that the road is shielded by the shadows of buildings and vegetation, the difference of the color difference, the shape and the like of the road is large in different time periods and areas, the spectrum of the road and the buildings is easy to be confused, the road distribution sparsity and the like are solved, and the problems bring difficulty to the automatic extraction of the road on a remote sensing image. When the saliency of the geometric texture features of the road is reduced due to the facing of spatial heterogeneity (shadow, curvature and adjacent similar texture) in the traditional method, the degree of automation is low; although the deep learning method has the advantages of strong generalization, high automation degree and the like, no good sample and no good deep learning result exist; the object-oriented segmentation method only extracts roads with single width, and extracts road networks with different grades by adopting different segmentation scale grades, so that the universality is not strong.
Disclosure of Invention
In view of the above technical problems, an object of the present invention is to provide a method for extracting a road from a high-resolution remote sensing image, which performs random forest classification by fusing patch shape parameters with spectrum, texture and terrain feature parameters to complete automatic identification and extraction of road information.
The technical scheme adopted by the invention is a method for extracting a high-resolution remote sensing image road fused with a plaque shape index, which comprises the following steps:
step 1, collecting and obtaining high-resolution remote sensing images and topographic data of a research area and carrying out data preprocessing.
And acquiring a high-resolution image and an elevation image in a research range, and preprocessing the images. According to the research condition, carrying out radiometric calibration and atmospheric correction on the acquired image by using a radiometric calibration function and a FLAASH atmospheric correction module in ENVI software; then defining projection parameters according to the geographical position of the research area, so that the high-resolution image and the elevation image keep consistent in space; thirdly, resampling is carried out according to the spatial resolution of the images, so that the spatial resolutions of the two images are kept consistent. And finally, cutting according to the research area range to obtain a remote sensing image of the research area range so as to facilitate subsequent data processing.
And 2, extracting the classification characteristic parameters of the remote sensing image.
201. The characteristic parameters to be extracted by the method need to normalize the vegetation index (NDVI), the normalized difference water body index (NDWI), the texture, the terrain and other information besides the spectral information of the remote sensing image. NDVI and NDWI can be obtained by calculation of an ENVI 5.5 middle waveband calculator according to a formula; the texture information uses a probability filtering tool in ENVI 5.5 to calculate eight texture parameter images of an acquired Mean (Mean), a Variance (Variance), Homogeneity (Homogeneity), Contrast (Contrast), Dissimilarity (similarity), Entropy (Entropy), Second Moment (Second _ Moment) and Correlation (Correlation); the Terrain information is extracted by using a Terrain module in ENVI 5.5 to obtain Elevation (Elevation), Slope (Slope) and Aspect (Aspect) information.
202. And extracting shape characteristic parameters. The plaque shape index is based on a super-pixel seed clustering method, plaques are extracted by using object-oriented eCooginion development Developer software multi-scale segmentation, and the plaques in a research region are obtained by cutting. And (4) counting the calculation geometry to obtain the area and the perimeter of each patch, calculating a result according to a formula of the patch shape index, and rasterizing the obtained value. The plaque shape index (S) is calculated as:
where P is the perimeter of the patch and A is the area of the patch. The patch shape index represents the degree of deviation of the shape from a square of the same area, the construction site is usually a small aggregated patch with a small degree of deviation, and the road is a scattered long patch with a large degree of deviation. Obviously, the addition of the plaque shape index is beneficial for the distinction of buildings and roads.
And 3, constructing a high-resolution remote sensing image road automatic identification algorithm model fused with the plaque shape index.
And (3) constructing a road identification and extraction model based on a random forest algorithm by using the spectrum, the texture, the terrain, the patch shape and other characteristic variables extracted in the step (2).
301. Randomly selecting N training sets from a total training sample by adopting a Bootstrap sampling method, and establishing a forest consisting of N CART decision trees; each training set is approximately 2/3 of the total training samples, and the remaining 1/3 samples are referred to as Out of Bag (OOB) data for internal error estimation, yielding OOB errors for model validation.
302. And calculating the importance of the characteristic variable. Calculating the error e of each decision tree according to the training sampletThen randomly changing the ith characteristic variable X of the test sampleiAnd calculating the error thereof
Obtaining a characteristic variable XiImportance of (A) V (X)i) The calculation formula is as follows:
in order to more accurately identify and extract road information, according to the characteristics of a random forest algorithm, extracting the importance information of characteristic variables, selecting the characteristic variables according to the importance degree of the characteristic variables, and reducing the dimension of high-dimensional data to obtain a classification scheme with the best precision and efficiency.
And 4, classifying the remote sensing image and automatically extracting road information.
Combining the data obtained in the steps 1, 2 and 3, removing vegetation and water bodies in a research area by adopting classification based on expert knowledge, obtaining buildings, roads and other land, considering the difference of the roads and the buildings in the aspects of texture and shape, and performing random forest classification by using remote sensing images fusing parameters such as spectrum, texture, terrain, patch shape and the like, thereby improving the accuracy of road identification. And then, processing the classification result by adopting a classification aggregation method, setting a minimum patch threshold value to remove classification noise, and vectorizing the classification result to obtain a road distribution range.
Further, the specific steps of performing dimension reduction on the high-dimensional feature variable in step 3 are as follows: firstly, calculating OOB errors by using a random forest algorithm, and analyzing the importance degree of different characteristic information in the classification process; then, selecting the feature information according to the importance of the feature variables, and removing the variables with small importance to reduce the dimension of the variables; and finally, the comprehensive information after dimension reduction is combined with a random forest algorithm to identify and extract road information, so that an optimal scheme between model operation efficiency and classification precision is realized.
Further, in step 4, in order to reduce workload, after setting a threshold value for the image and the parameters thereof, the water body and vegetation cover are removed, and the rest of construction land, roads and other lands are left. A random forest classification method is adopted, and based on a manual interpretation data set, buildings, roads and other land are classified to obtain road distribution. The method firstly adopts decision tree classification based on expert knowledge to preferentially remove irrelevant information of water body and vegetation coverage, thereby greatly improving the subsequent classification efficiency and shortening the working time; secondly, under the guidance of theory and experience common knowledge, a random forest classification means of texture and patch shape index is added to effectively distinguish buildings from roads, and the identification precision of the roads is effectively improved.
Compared with the prior art, the invention has the following advantages:
(1) the road information is extracted by utilizing the fusion of the plaque shape index and characteristic variables such as spectrum, geometric texture, terrain and the like, so that the road information is favorably distinguished from buildings and roads, the road acquisition precision is high, and the road information can be effectively extracted.
(2) The random forest algorithm can effectively utilize the difference between samples, calculate the importance of characteristic variables by using OOB errors, select the optimal variables for classification, and remarkably improve the identification efficiency of multi-dimensional information based on high-resolution images by the parallel processing capability of the random forest algorithm, thereby truly realizing the optimal scheme between the model operation efficiency and the classification precision.
(3) Aiming at the complex environment of the city, the road extraction method of the high-resolution remote sensing image fused with the plaque shape index is adopted, and the method has better pertinence and adaptability.
Drawings
The invention is further described below with reference to the accompanying drawings.
FIG. 1 is a flow chart of one example of the present invention.
Detailed Description
As shown in fig. 1, the method comprises the following specific steps:
the invention adopts the technical scheme that a method for extracting a high-resolution remote sensing image road fused with a plaque shape index comprises the following steps:
step 1, collecting and obtaining high-resolution remote sensing images and topographic data of a research area and carrying out data preprocessing.
And acquiring a high-resolution image and an elevation image in a research range, and preprocessing the images. According to the research condition, carrying out radiometric calibration and atmospheric correction on the acquired image by using a radiometric calibration function and a FLAASH atmospheric correction module in ENVI software; then defining projection parameters according to the geographical position of the research area, so that the high-resolution images and the elevation images keep consistent space; thirdly, resampling is carried out according to the spatial resolution of the images, so that the spatial resolutions of the two images are kept consistent. And finally, cutting according to the research area range to obtain a remote sensing image of the research area range so as to facilitate subsequent data processing.
Step 1, extracting the classification characteristic parameters of the remote sensing image.
201. The characteristic parameters to be extracted by the method need to normalize the vegetation index (NDVI), the normalized difference water body index (NDWI), the texture, the terrain and other information besides the spectral information of the remote sensing image. NDVI and NDWI can be obtained by calculation of an ENVI 5.5 middle waveband calculator according to a formula; the texture information uses a probability filtering tool in ENVI 5.5 to calculate eight texture parameter images of an acquired Mean (Mean), a Variance (Variance), Homogeneity (Homogeneity), Contrast (Contrast), Dissimilarity (similarity), Entropy (Entropy), Second Moment (Second _ Moment) and Correlation (Correlation); the Terrain information is extracted by using a Terrain module in ENVI 5.5 to obtain Elevation (Elevation), Slope (Slope) and Aspect (Aspect) information.
202. And extracting shape characteristic parameters. The plaque shape index is based on a super-pixel seed clustering method, plaques are extracted by using object-oriented eCooginion development Developer software multi-scale segmentation, and the plaques in a research region are obtained by cutting. And (4) counting the calculation geometry to obtain the area and the perimeter of each patch, calculating a result according to a formula of the patch shape index, and rasterizing the obtained value. The plaque shape index (S) is calculated as:
in the formula, P represents the circumferential length of the patch, and A represents the area of the patch. The patch shape index represents the degree of deviation of the shape from a square of the same area, the construction site is usually a small aggregated patch with a small degree of deviation, and the road is a scattered long patch with a large degree of deviation. Obviously, the addition of the plaque shape index is beneficial for the distinction of buildings and roads.
And 3, constructing a high-resolution remote sensing image road automatic identification algorithm model fused with the plaque shape index.
And (3) constructing a road identification and extraction model based on a random forest algorithm by using the spectrum, the texture, the terrain, the patch shape and other characteristic variables extracted in the step (2).
301. Randomly selecting N training sets from a total training sample by adopting a Bootstrap sampling method, and establishing a forest consisting of N CART decision trees; each training set is approximately 2/3 of the total training samples, and the remaining 1/3 samples are referred to as Out of Bag (OOB) data for internal error estimation, yielding OOB errors for model validation.
302. And calculating the importance of the characteristic variable. Calculating the error e of each decision tree according to the training sampletThen randomly changing the ith characteristic variable X of the test sampleiAnd calculating the error thereof
Obtaining a characteristic variable XiImportance of (A) V (X)i) The calculation formula is as follows:
in order to more accurately identify and extract road information, according to the characteristics of a random forest algorithm, extracting the importance information of characteristic variables, selecting the characteristic variables according to the importance degree of the characteristic variables, and reducing the dimension of high-dimensional data to obtain a classification scheme with the best precision and efficiency.
And 4, classifying the remote sensing image and automatically extracting road information.
Combining the data obtained in the steps 1, 2 and 3, removing vegetation and water bodies in a research area by adopting classification based on expert knowledge, obtaining buildings, roads and other land, considering the difference of the roads and the buildings in the aspects of texture and shape, and performing random forest classification by using remote sensing images fusing parameters such as spectrum, texture, terrain, patch shape and the like, thereby improving the accuracy of road identification. And then, processing the classification result by adopting a classification aggregation method, setting a minimum patch threshold value to remove classification noise, and vectorizing the classification result to obtain a road distribution range.
Further, the specific steps of performing dimension reduction on the high-dimensional feature variable in step 3 are as follows: firstly, calculating OOB errors by using a random forest algorithm, and analyzing the importance degree of different characteristic information in the classification process; then, selecting the feature information according to the importance of the feature variables, and removing the variables with small importance to reduce the dimension of the variables; and finally, the comprehensive information after dimension reduction is combined with a random forest algorithm to identify and extract road information, so that an optimal scheme between model operation efficiency and classification precision is realized.
Further, in step 4, in order to reduce workload, after setting a threshold value for the image and the parameters thereof, the water body and vegetation cover are removed, and the rest of construction land, roads and other lands are left. A random forest classification method is adopted, and based on a manual interpretation data set, buildings, roads and other land are classified to obtain road distribution. The method firstly adopts decision tree classification based on expert knowledge to preferentially remove irrelevant information of water body and vegetation coverage, thereby greatly improving the subsequent classification efficiency and shortening the working time; secondly, under the guidance of theory and experience common knowledge, a random forest classification means of texture and patch shape index is added to effectively distinguish buildings from roads, and the identification precision of the roads is effectively improved.
Claims (3)
1. A high-resolution remote sensing image road extraction method fused with patch shape indexes is characterized by comprising the following specific steps:
step 1, collecting and acquiring a high-resolution remote sensing image and topographic data of a research area and carrying out data preprocessing:
acquiring a high-resolution image and an elevation image in a research range, and preprocessing the images; according to the research condition, carrying out radiometric calibration and atmospheric correction on the acquired image by using a radiometric calibration function and a FLAASH atmospheric correction module in ENVI software; then defining projection parameters according to the geographical position of the research area, so that the high-resolution image and the elevation image keep consistent in space; thirdly, resampling is carried out according to the spatial resolution of the images, so that the spatial resolutions of the two images are kept consistent. Finally, cutting according to the research area range to obtain a remote sensing image of the research area range so as to facilitate subsequent data processing;
step 2, extracting classification characteristic parameters of the remote sensing image;
201. the characteristic parameters needing to be extracted in the method need information such as normalized vegetation index (NDVI), normalized difference water body index (NDWI), texture, terrain and the like besides spectral information of the remote sensing image; NDVI and NDWI can be obtained by calculation of an ENVI 5.5 middle waveband calculator according to a formula; the texture information uses a probability filtering tool in ENVI 5.5 to calculate eight texture parameter images of obtaining a Mean (Mean), a Variance (Variance), Homogeneity (Homogeneity), a Contrast (Contrast), a Dissimilarity (Dissimilarity), an Entropy (Entropy), a Second Moment (Second _ Moment) and a Correlation (Correlation); extracting Elevation (Elevation), Slope (Slope) and Aspect (Aspect) information by using a Terrain module in ENVI 5.5 for topographic information;
202. and extracting shape characteristic parameters. The plaque shape index is based on a super-pixel seed clustering method, and the plaque is extracted by using object-oriented eCooginion Developer software multi-scale segmentation and is cut to obtain the plaque in a research area; calculating the geometry by statistics to obtain the area and the perimeter of each patch, calculating the result according to a formula of the patch shape index, and rasterizing the obtained value; the plaque shape index (S) is calculated as:
wherein P is the circumference of the patch, and A is the area of the patch; the patch shape index represents the deviation degree of the shape from a square with the same area, the construction land is usually a gathered small patch, the deviation degree is small, the road is a dispersed long patch, and the deviation degree is large; obviously, the addition of the patch shape index is beneficial to the distinction of buildings and roads;
step 3, building a high-resolution remote sensing image road automatic identification algorithm model fused with the patch shapes:
constructing a road identification and extraction model based on a random forest algorithm by using the spectrum, texture, terrain, patch shape and other characteristic variables extracted in the step 2;
301. randomly selecting N training sets from a total training sample by adopting a Bootstrap sampling method, and establishing a forest consisting of N CART decision trees; each training set is approximately 2/3 of the total training samples, the remaining 1/3 samples are called Out of Bag (OOB) data for internal error estimation, and OOB errors are generated for model verification;
302. calculating the importance of the characteristic variable; calculating the error e of each decision tree according to the training sampletThen randomly changing the ith characteristic variable X of the test sampleiAnd calculating the error thereof
Obtaining a characteristic variable XiImportance of (A) V (X)i) The calculation formula is as follows:
in order to more accurately identify and extract road information, extracting importance information of characteristic variables according to the characteristics of a random forest algorithm, selecting the characteristic variables according to the importance degree of the characteristic variables, and reducing the dimensions of high-dimensional data to obtain a classification scheme with optimal precision and efficiency;
step 4, remote sensing image classification and road information automatic extraction:
combining the data obtained in the steps 1, 2 and 3, removing vegetation and water bodies in a research area by adopting classification based on expert knowledge, obtaining buildings, roads and other land, considering the difference of the roads and the buildings in the aspects of texture and shape, and performing random forest classification by using remote sensing images fusing parameters such as spectrum, texture, terrain, patch shape and the like so as to improve the accuracy of road identification; and then, processing the classification result by adopting a classification aggregation method, setting a minimum patch threshold value to remove classification noise, and vectorizing the classification result to obtain a road distribution range.
2. The method for extracting the high-resolution remote sensing image road fused with the plaque shape index according to claim 1, wherein in the step 3, a high-resolution remote sensing image road automatic identification algorithm model fused with the plaque shape index is constructed; the method comprises the following specific steps: in order to more accurately identify and extract road information, according to the characteristics of a random forest algorithm, a random forest algorithm-based model is constructed by using the spectrum, texture, terrain, patch shape and other characteristic variables extracted in the step 2, importance information of the characteristic variables is extracted, the characteristic variables are selected according to the importance degree of the characteristic variables, dimension reduction is performed on high-dimensional data, and then a classification scheme with the best precision and efficiency is obtained.
3. The method for extracting the high-resolution remote sensing image road fused with the plaque shape index according to claim 1, wherein the specific process of the step 4 is as follows: combining the data obtained in the steps 1, 2 and 3, removing vegetation and water bodies in a research area by adopting classification based on expert knowledge, obtaining buildings, roads and other lands, considering the difference of the roads and the buildings in the aspects of texture and shape, and using random forest classification fusing parameters such as spectrum, texture, terrain, patch shape and the like so as to improve the accuracy of road identification; then, processing the classification result by adopting a classification aggregation method, and setting a minimum plaque threshold value to remove classification noise; and vectorizing the classification result to obtain a road distribution range.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111589580.9A CN114494897A (en) | 2021-12-23 | 2021-12-23 | High-resolution remote sensing image road extraction method fused with patch shape index |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111589580.9A CN114494897A (en) | 2021-12-23 | 2021-12-23 | High-resolution remote sensing image road extraction method fused with patch shape index |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114494897A true CN114494897A (en) | 2022-05-13 |
Family
ID=81493329
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111589580.9A Pending CN114494897A (en) | 2021-12-23 | 2021-12-23 | High-resolution remote sensing image road extraction method fused with patch shape index |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114494897A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821333A (en) * | 2022-05-16 | 2022-07-29 | 中国人民解放军61540部队 | High-resolution remote sensing image road material identification method and device |
| CN115953674A (en) * | 2022-10-26 | 2023-04-11 | 中国科学院空天信息创新研究院 | Remote sensing detection and identification method and system for circular grave relics |
| CN117315501A (en) * | 2023-10-23 | 2023-12-29 | 中国水利水电科学研究院 | Remote sensing water body classification method based on water body plaque shape and adjacent relation |
-
2021
- 2021-12-23 CN CN202111589580.9A patent/CN114494897A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114821333A (en) * | 2022-05-16 | 2022-07-29 | 中国人民解放军61540部队 | High-resolution remote sensing image road material identification method and device |
| CN115953674A (en) * | 2022-10-26 | 2023-04-11 | 中国科学院空天信息创新研究院 | Remote sensing detection and identification method and system for circular grave relics |
| CN117315501A (en) * | 2023-10-23 | 2023-12-29 | 中国水利水电科学研究院 | Remote sensing water body classification method based on water body plaque shape and adjacent relation |
| CN117315501B (en) * | 2023-10-23 | 2024-04-12 | 中国水利水电科学研究院 | Remote sensing water body classification method based on water body plaque shape and adjacent relation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108846832B (en) | Multi-temporal remote sensing image and GIS data based change detection method and system | |
| CN114494897A (en) | High-resolution remote sensing image road extraction method fused with patch shape index | |
| CN108830870B (en) | Satellite image high-precision farmland boundary extraction method based on multi-scale structure learning | |
| WO2021232463A1 (en) | Multi-source mobile measurement point cloud data air-ground integrated fusion method and storage medium | |
| CN103400151B (en) | The optical remote sensing image of integration and GIS autoregistration and Clean water withdraw method | |
| CN107330875B (en) | Water body surrounding environment change detection method based on forward and reverse heterogeneity of remote sensing image | |
| CN107067405B (en) | Remote sensing image segmentation method based on scale optimization | |
| CN111626269B (en) | Practical large-space-range landslide extraction method | |
| CN112164062A (en) | Wasteland information extraction method and device based on remote sensing time sequence analysis | |
| CN112669363A (en) | Urban green land three-dimensional green volume calculation method | |
| Fan et al. | UAV image crop classification based on deep learning with spatial and spectral features | |
| Peng et al. | Building change detection by combining lidar data and ortho image | |
| Koc-San et al. | Plastic and glass greenhouses detection and delineation from worldview-2 satellite imagery | |
| Wang et al. | Adaptive scale selection in multiscale segmentation based on the segmented object complexity of GF-2 satellite image | |
| Liu et al. | Adaptive algorithm for automated polygonal approximation of high spatial resolution remote sensing imagery segmentation contours | |
| CN114066815B (en) | SAR image change detection method based on fuzzy c-means clustering | |
| CN115953685A (en) | Multi-layer multi-scale division agricultural greenhouse type information extraction method and system | |
| Magee | Segmentation, object-oriented applications for remote sensing land cover and land use classification | |
| CN115830464A (en) | Plateau mountain agricultural greenhouse automatic extraction method based on multi-source data | |
| CN113111794B (en) | High-resolution annual city green space remote sensing information extraction method for pattern spots | |
| Liu et al. | Semi-automatic extraction and mapping of farmlands based on high-resolution remote sensing images | |
| Yang et al. | Remote sensing image change saliency detection technology | |
| CN109886142B (en) | Crop interpretation method based on SAR technology | |
| CN114596489A (en) | High-precision multisource remote sensing city built-up area extraction method for human habitation index | |
| Zhan et al. | Pattern detection in airborne LiDAR data using Laplacian of Gaussian filter |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |