A desert region ground object discrimination method and system based on remote sensing image interpretation

Abstract

This invention discloses a method and system for identifying land cover in desert areas based on remote sensing image interpretation. It acquires multi-band reflectance and various remote sensing index data from satellite remote sensing images of the target area and performs band merging using a band combination tool to obtain merged image data. Based on statistically obtained remote sensing index threshold ranges corresponding to different land cover types, it constructs a hierarchical land cover separation and extraction rule. Based on the constructed hierarchical land cover separation and extraction rule, and combined with mask extraction, it separates and extracts different types of land covers from the band-merged image data. The interpretation method of this invention is relatively flexible, using different remote sensing index threshold ranges for different land cover types to effectively extract land cover feature information. Based on a rule-oriented object-oriented workflow, it can effectively extract various types of land covers, and the use of mask extraction effectively avoids misclassification of land covers.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image interpretation technology, specifically to a method and system for identifying land features in desert areas based on remote sensing image interpretation. Background Technology

[0002] With the development of 3S technology and the demand for social informatization, geographic information remote sensing technology has been increasingly widely applied in various fields of the national economy, and the demand for geographic information from all sectors of society is also increasing. Remote sensing imagery, due to its advantages such as large coverage area, intuitive information, and strong realism, has become an important information source for various geographic information products. How to quickly extract spatial data from these image data and achieve rapid updates of spatial data, enabling geographic information products to serve more users, is of great significance. Land desertification has always been one of the major global resource and environmental issues that has received much attention.

[0003] In remote sensing imagery, the content elements are primarily composed of images, supplemented by map symbols to represent or explain the mapped objects. Compared to ordinary maps, imagery maps possess rich ground information, clear content hierarchy, and are easy to read, fully demonstrating the dual advantages of imagery and maps. Remote sensing images are classified into aerial photographs (taken by aircraft) and satellite photographs (taken by satellite) according to their acquisition method. They can be categorized by spatial resolution into high-resolution, medium-resolution, and low-resolution imagery. Based on spectral resolution, they can be classified into spectral (hyperspectral), panchromatic, visible light, far-infrared, and near-infrared images. There are generally two methods for interpreting and classifying remote sensing images: pixel-based classification and object-based classification. Pixel-based classification mainly includes supervised and unsupervised classification, and classification based on expert knowledge decision trees; object-based classification typically refers to object-oriented classification methods.

[0004] Domestic and international scholars have conducted research on land use information extraction using 3S technology, employing methods such as supervised classification with maximum likelihood estimation, spectral enhancement classification, decision tree classification, and object-oriented classification. While these studies have achieved varying degrees of good extraction results, most employ a single method. For areas with spectral obfuscation in remote sensing images and complex land cover types, these extraction methods inevitably face limitations. Summary of the Invention

[0005] This application provides a method and system for identifying land features in desert areas based on remote sensing image interpretation, in order to solve the problems of low interpretation accuracy, easy mixing, misclassification, and omission in existing remote sensing image interpretation technologies.

[0006] According to the first aspect, one embodiment provides a method for identifying land features in desert areas based on remote sensing image interpretation, the method comprising: Obtain the multi-band reflectance and various remote sensing index data of the satellite remote sensing image in the target area, and use the band combination tool to merge the bands to obtain the merged image data; Based on the statistically obtained remote sensing index threshold ranges corresponding to different land cover types, construct the rules for hierarchical separation and extraction of land cover; Based on the constructed rules for hierarchical separation and extraction of land cover, and combined with the method of mask extraction, separate and extract different types of land cover from the image data obtained by band merging.

[0007] Further, obtaining the multi-band reflectance and various remote sensing index data of the satellite remote sensing image in the target area and using the band combination tool for band merging specifically includes: Use the ENVI band combination tool to merge the multi-band reflectance and spectral indices of the Landsat TM / OLI original image together as the raster data for multi-scale segmentation of the ENVI FX image.

[0008] Further, obtaining the multi-band reflectance and various remote sensing index data of the satellite remote sensing image in the target area and using the band combination tool for band merging specifically includes: The original image bands for merging include blue, green, red, near-infrared, and short-wave infrared bands, and the spectral indices include EVI, MNDWI, NDBI, RVI, and DVI.

[0009] Further, obtaining the merged image data specifically further includes: By setting different segmentation scales, compare the differences between the regional classification effects and the actual situations under different segmentation scales, and obtain the optimal segmentation scale for subsequent setting of the optimal segmentation scale in the object-oriented spatial feature extraction process for land cover extraction.

[0010] Further, based on the statistically obtained remote sensing index threshold ranges corresponding to different land cover types, construct the rules for hierarchical separation and extraction of land cover, specifically including: The rules for hierarchical separation and extraction of land cover include: First, apply the threshold interval of 0.75 < EVI < 1. Those within the range of 0.75 < EVI < 1 are cultivated land and forest land pixels, otherwise they belong to water area, grassland, sandy land, and construction land pixels; For the separated cultivated land and forest land pixels, further apply the threshold interval of NIR > 0.3. If within the range of NIR > 0.3, they are cultivated land pixels, otherwise they are determined to be forest land pixels; For the isolated water, grassland, sandy land, and construction land pixels, further apply the threshold range of 0.35 < MNDWI < 1. If it is within the range of 0.35 < MNDWI < 1, it is a water pixel. Otherwise, further apply the threshold range of 0.075 < DVI < 0.17. If it is within the range of 0.075 < DVI < 0.17, it is a grassland pixel. Otherwise, further apply the threshold range of 1.05 < RVI < 1.3. If it is within the range of 1.05 < RVI < 1.3, it is a sandy land pixel. Otherwise, it is a construction land pixel.

[0011] Furthermore, based on the constructed hierarchical separation and extraction rules for ground objects, and combined with the mask extraction method, the image data obtained by band merging is separated and extracted for different types of ground objects, specifically including: Based on the constructed hierarchical separation and extraction rules for ground objects, different types of ground objects including water, cultivated land, forest land, grassland, sandy land, and construction land are extracted; Among them, to extract typical water ground objects, the threshold 0.35 < MDWI < 1 needs to be set in the rule-based object-oriented process; to extract typical cultivated land ground objects, the threshold 0.75 < EVI < 1 / NIR > 0.3 needs to be set in the rule-based object-oriented process; to extract typical forest land ground objects, the threshold 0.75 < EVI < 1 / NIR ≤ 0.3 needs to be set in the rule-based object-oriented process; to extract typical grassland ground objects, the thresholds DVI ≤ 0.075 and DVI ≥ 0.17 need to be set in the rule-based object-oriented process; to extract typical sandy land ground objects, the thresholds 0.075 < DVI < 0.17 / 1.05 < RVI < 1.3 need to be set in the rule-based object-oriented process; to extract typical construction land ground objects, the thresholds 0.075 < DVI < 0.17 / RVI ≥ 1.3 and RVI ≤ 1.05 need to be set in the rule-based object-oriented process.

[0012] Furthermore, based on the constructed hierarchical separation and extraction rules for ground objects, and combined with the mask extraction method, the image data obtained by band merging is separated and extracted for different types of ground objects, specifically including: Create masks for the classification results of each type of ground object; Load the masks of different types of ground objects into the object-oriented classification workflow, and extract the six types of ground objects in turn according to the reverse mask extraction method to obtain the mask extraction result maps of different types of ground objects.

[0013] According to the second aspect, in one embodiment, a desert area ground object discrimination system based on remote sensing image interpretation is provided. The system includes: An image segmentation module, configured to obtain the multi-band reflectance and various remote sensing index data of the satellite remote sensing image of the target area and use the band combination tool to perform band merging to obtain the merged image data; The feature separation rule construction module is used to construct step-by-step feature separation and extraction rules based on the remote sensing index threshold ranges corresponding to different feature types obtained statistically. The feature extraction module is used to separate and extract different types of features from the image data obtained by band merging based on the constructed feature separation and extraction rules and the mask extraction method.

[0014] According to a third aspect, one embodiment provides an electronic device, characterized in that the device includes: a processor and a memory; The memory is used to store one or more program instructions; The processor is configured to run one or more program instructions to perform the steps of a method for identifying land features in desert areas based on remote sensing image interpretation as described in any of the preceding claims.

[0015] According to a fourth aspect, one embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a method for identifying land features in desert areas based on remote sensing image interpretation as described in any of the preceding claims.

[0016] This application provides a method and system for identifying land features in desert areas based on remote sensing image interpretation. The interpretation method is relatively flexible, and different remote sensing index threshold ranges are matched for different land feature types to effectively extract land feature feature information. Based on a rule-oriented object-oriented workflow, it can effectively extract various types of land features, and the mask extraction method is used to effectively avoid misclassification of land features. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a method for identifying land features in desert areas based on remote sensing image interpretation, as provided in one embodiment of the present invention; Figure 2 An overall roadmap for a method for identifying land features in desert areas based on remote sensing image interpretation, provided as an embodiment of the present invention; Figure 3 A comparison of the multi-scale segmentation effect of ENVI FX images in a method for identifying land features in desert areas based on remote sensing image interpretation, provided in one embodiment of the present invention; Figure 4 This invention provides a step-by-step separation and extraction process for typical land cover types in a land cover identification method for desert areas based on remote sensing image interpretation, as an embodiment of the present invention. Figure 5 This invention provides a typical land cover type mask map in a method for identifying land cover in desert areas based on remote sensing image interpretation, as an embodiment of the present invention. Figure 6This is a schematic diagram of the logical structure of a desert area feature identification system based on remote sensing image interpretation, provided as an embodiment of the present invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0019] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0020] The first embodiment of this invention provides a method for identifying land cover in desert areas based on remote sensing image interpretation. This invention fully considers the brightness and color characteristics of typical land cover features characterized by visible-near-infrared remote sensing bands, and also makes full use of the abstract expression of land cover features by spectral indices. It applies an object-oriented remote sensing classification method to separate and extract target land cover types step by step. The following is a combination of... Figure 1 A detailed explanation is provided. The overall technical approach is as follows: Figure 2 As shown.

[0021] like Figure 1 As shown, in step S100, multi-band reflectance and various remote sensing index data of satellite remote sensing images of the target area are acquired, and band merging is performed using a band combination tool to obtain merged image data.

[0022] Specifically, image segmentation: multiple remote sensing indices such as MNDWI, EVI, NDBI, RVI, and DVI are used, along with Landsat multi-band fusion into a single band. Landsat-TM imagery mainly has 7 bands (blue band, green band, red band, near-infrared band, mid-infrared band, thermal infrared band, and far-infrared band), while Landsat-OLI imagery mainly has 11 bands (coastal band, blue band, green band, red band, near-infrared band, shortwave infrared 1, shortwave infrared 2, panchromatic band, cirrus band, thermal infrared 1, and thermal infrared 2). In this embodiment, the reflectance (blue, green, red, near-infrared, shortwave infrared) and spectral indices (EVI, MNDWI, NDBI, RVI, DVI) of the original Landsat TM / OLI image bands are combined together using the ENVI band combination tool to form raster data for multi-scale segmentation of the ENVI FX image. The difference in the upper boundary at different scales is controlled by adjusting the segmentation scale.

[0023] To achieve ideal segmentation results and ensure the accuracy of the final classification, this embodiment included a comparative experiment with nine segmentation scales: 5, 10, 15, 20, 25, 30, 35, 40, and 45. The results showed that at a segmentation scale of 20, using the original remote sensing image as the comparison object, the image's patch representation more closely matched the actual situation. Figure 3 Based on this, the patch approximate merging threshold and texture kernel were set to 80 and 3, respectively. Subsequently, the optimal segmentation scale was set in a rule-based object-oriented workflow for feature extraction.

[0024] like Figure 1 As shown, in step S200, based on the remote sensing index threshold ranges corresponding to different land cover types obtained statistically, a land cover separation and extraction rule is constructed.

[0025] Specifically, the creation of ground feature separation rules involves setting different threshold ranges for different ground features. Through extensive literature review and numerous experiments, the statistical threshold ranges of remote sensing band reflectance and spectral indices corresponding to ground survey samples of different ground feature types were used as a reference (Table 1). Through numerous repeated comparative experiments, it was found that by adjusting the threshold parameter range, different ground features could be extracted. The threshold value range determines the accuracy of ground feature extraction. Finally, a reasonable and effective step-by-step ground feature separation and extraction process was proposed. Figure 4). Applying the threshold range of 0.75 < EVI < 1, the six typical land cover types are divided into two groups. That is, the pixels within this range are cultivated land and forest pixels, otherwise they belong to water, grassland, sandy land, and construction land pixels. When the near-infrared reflectance NIR > 0.3 for the cultivated land and forest pixels, they are determined as cultivated land pixels, otherwise they are determined as forest pixels. Applying the threshold range of 0.35 < MNDWI < 1, water bodies can be separated from the images containing water, grassland, sandy land, and construction land. Then, applying the threshold range of 0.075 < DVI < 0.17 further separates the grassland. Finally, applying the threshold range of 1.05 < RVI < 1.3 separates the sandy land and construction land.

[0026] Table 1 Threshold Ranges of Remote Sensing Indices for Typical Land Cover Types

[0027]

[0028] As Figure 1 shown, in step S300, based on the constructed hierarchical separation and extraction rules for ground objects, and combined with the mask extraction method, different types of ground objects are separated and extracted from the image data obtained by band merging.

[0029] Specifically, for the extraction of typical ground objects: The six types of ground objects are extracted layer by layer using the mask extraction method. After the extraction is completed, a map of each typical ground object is obtained.

[0030] During the extraction, based on the hierarchical separation and extraction rules for ground objects, to extract the typical water body ground object, the threshold 0.35 < MDWI < 1 needs to be set in the object-oriented process based on the rules. To extract the typical cultivated land ground object, the threshold 0.75 < EVI < 1 / NIR > 0.3 needs to be set in the object-oriented process based on the rules. To extract the typical forest ground object, the threshold 0.75 < EVI < 1 / NIR ≤ 0.3 needs to be set in the object-oriented process based on the rules. To extract the typical grassland ground object, the thresholds DVI ≤ 0.075 and DVI ≥ 0.17 need to be set in the object-oriented process based on the rules. To extract the typical sandy land ground object, the thresholds 0.075 < DVI < 0.17 / 1.05 < RVI < 1.3 need to be set in the object-oriented process based on the rules. To extract the typical construction land ground object, the thresholds 0.075 < DVI < 0.17 / RVI ≥ 1.3 and RVI ≤ 1.05 need to be set in the object-oriented process based on the rules.

[0031] Based on the land cover classification rules, to avoid the impact of "different objects with the same spectrum" or "different spectra with the same object" phenomena on the classification results, this embodiment uses the spatial and spectral characteristics of land cover, geographical location, and other information to calculate the mask of each extracted land cover classification result. The calculation formula is (b1 ge 1)*1+(b1 lt 1)*0, where b1 represents the band, ge represents greater than or equal to, and lt represents less than. Then, the different land cover masks are loaded into the object-oriented classification workflow, and the six types of land cover, namely water area, cultivated land, forest land, construction land, sandy land, and grassland, are extracted in sequence according to the reverse mask extraction method. Figure 5 This is an image showing the effect of using land cover type mask extraction in this invention.

[0032] The method in this embodiment is based on ENVI 5.3 and ArcGIS 10.8 software. It extracts remote sensing indices such as EVI, MNDWI, NDBI, DVI, and RVI from remote sensing images. Through grayscale segmentation and extensive experimental analysis and comparison, the optimal band threshold is selected to extract land use feature information. Then, the object-oriented spatial feature extraction module (feature extraction, FX) in ENVI is used to extract feature information. In the object-oriented spatial feature extraction process, the main software operation steps include: 1. Open the object-oriented land cover extraction tool in the software; 2. Parameter settings: In the opened tool box, select the image to be classified; 3. Select spectral index and color space information. Other settings are not required. Note that the spectral index is an index that is helpful for classification. For example, NDVI is very important when extracting vegetation, so select the near-infrared and red bands; 4. Select the segmentation and merging scale. You can click preview to check if the scale is appropriate; 5. Rule creation: Add different land cover categories, depending on how many categories to classify. You can modify the name and color. After creating a new category, you can create multiple rules. Rules can be band values, spectral index values, geometric area, length, etc.; 6. Result output: After completing the rule creation, select the output format to output the results.

[0033] The interpretation method of this invention is relatively flexible. It can effectively extract the feature information of land cover by matching different remote sensing index threshold ranges for different land cover types. Based on the rule-oriented object-oriented workflow, it can effectively extract various types of land cover. Furthermore, it adopts a mask extraction method to effectively avoid misclassification of land cover.

[0034] Corresponding to the aforementioned method for identifying land features in desert areas based on remote sensing image interpretation, this invention also discloses a system for identifying land features in desert areas based on remote sensing image interpretation, such as... Figure 6 As shown, it specifically includes: The image segmentation module is used to acquire multi-band reflectance and various remote sensing index data of satellite remote sensing images of the target area and to merge the bands using a band combination tool to obtain merged image data. The feature separation rule construction module is used to construct step-by-step feature separation and extraction rules based on the remote sensing index threshold ranges corresponding to different feature types obtained statistically. The feature extraction module is used to separate and extract different types of features from the image data obtained by band merging based on the constructed feature separation and extraction rules and the mask extraction method.

[0035] It should be noted that for a detailed description of the desert area feature identification system based on remote sensing image interpretation provided in the embodiments of the present invention, please refer to the relevant description of the desert area feature identification method based on remote sensing image interpretation provided in the embodiments of this application, which will not be repeated here.

[0036] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A method for identifying land features in desert areas based on remote sensing image interpretation, characterized in that, The method includes: Obtaining the multi-band reflectance and various remote sensing index data of the satellite remote sensing image of the target area, and using a band combination tool to merge the bands to obtain the merged image data; Based on the statistically obtained remote sensing index threshold ranges corresponding to different land cover types, constructing a rule for hierarchical separation and extraction of land cover; Based on the constructed rule for hierarchical separation and extraction of land cover, and combined with the method of mask extraction, separating and extracting different types of land cover from the image data obtained by band merging.

2. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 1, characterized in that, Obtaining the multi-band reflectance and various remote sensing index data of the satellite remote sensing image of the target area, and using a band combination tool to merge the bands, specifically including: Using the ENVI band combination tool to merge the multi-band reflectance and spectral indices of the Landsat TM / OLI original image together as the raster data for multi-scale segmentation of the ENVI FX image.

3. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 2, characterized in that, Obtaining the multi-band reflectance and various remote sensing index data of the satellite remote sensing image of the target area, and using a band combination tool to merge the bands, specifically including: The original image bands for merging include the blue, green, red, near-infrared, and short-wave infrared bands, and the spectral indices include EVI, MNDWI, NDBI, RVI, and DVI.

4. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 1, characterized in that, Obtaining the merged image data, specifically further including: By setting different segmentation scales, comparing the differences between the regional classification effects and the actual situation at different segmentation scales, obtaining the optimal segmentation scale for subsequent setting of the optimal segmentation scale in the object-oriented spatial feature extraction process for land cover extraction.

5. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 1, characterized in that, Based on the statistically obtained remote sensing index threshold ranges corresponding to different land cover types, constructing a rule for hierarchical separation and extraction of land cover, specifically including: The rule for hierarchical separation and extraction of land cover includes: First, apply the threshold range of 0.75 < EVI < 1. Those within the range of 0.75 < EVI < 1 are cultivated land and forest pixels, otherwise they belong to water, grassland, sandy land, and construction land pixels; For the separated cultivated land and forest pixels, further apply the threshold range of NIR > 0.

3. If within the range of NIR > 0.3, they are cultivated land pixels, otherwise they are determined to be forest pixels; For the separated water, grassland, sandy land, and construction land pixels, further apply the threshold range of 0.35 < MNDWI < 1. If within the range of 0.35 < MNDWI < 1, they are water pixels, otherwise further apply the threshold range of 0.075 < DVI < 0.

17. If within the range of 0.075 < DVI < 0.17, they are grassland pixels, otherwise further apply the threshold range of 1.05 < RVI < 1.

3. If within the range of 1.05 < RVI < 1.3, they are sandy land pixels, otherwise they are construction land pixels.

6. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 5, characterized in that, Based on the constructed rule for hierarchical separation and extraction of land cover, and combined with the method of mask extraction, separating and extracting different types of land cover from the image data obtained by band merging, specifically including: Based on the constructed rule for hierarchical separation and extraction of land cover, extracting different land cover types including water, cultivated land, forest, grassland, sandy land, and construction land. Among them, for extracting typical water features, it is necessary to set the threshold 0.35 < MDWI < 1 in the rule-based object-oriented process; for extracting typical cultivated land features, it is necessary to set the threshold 0.75 < EVI < 1 / NIR > 0.3 in the rule-based object-oriented process; for extracting typical forest features, it is necessary to set the threshold 0.75 < EVI < 1 / NIR ≤ 0.3 in the rule-based object-oriented process; for extracting typical grassland features, it is necessary to set the threshold DVI ≤ 0.075, DVI ≥ 0.17 in the rule-based object-oriented process; for extracting typical sandy land features, it is necessary to set the threshold 0.075 < DVI < 0.17 / 1.05 < RVI < 1.3 in the rule-based object-oriented process; for extracting typical construction land features, it is necessary to set the threshold 0.075 < DVI < 0.17 / RVI ≥ 1.3, RVI ≤ 1.05 in the rule-based object-oriented process.

7. The method for identifying land features in desert areas based on remote sensing image interpretation as described in claim 1, characterized in that, Based on the constructed hierarchical separation and extraction rules for features, and combined with the mask extraction method, the image data obtained by band merging is separated and extracted for different types of features, specifically including: Create a mask for the classification result of each type of feature; Load the masks of different types of features into the object-oriented classification workflow, and extract the six types of features in turn according to the reverse mask extraction method to obtain the mask extraction result maps of different types of features.

8. A feature identification system for desert areas based on remote sensing image interpretation, characterized in that, The system includes: An image segmentation module, which is used to obtain the multi-band reflectance and various remote sensing index data of the satellite remote sensing image of the target area and perform band merging using the band combination tool to obtain the merged image data; A feature separation rule construction module, which is used to construct hierarchical separation and extraction rules for features based on the remotely sensed index threshold ranges corresponding to different feature types obtained statistically; A feature extraction module, which is used to separate and extract different types of features from the image data obtained by band merging based on the constructed hierarchical separation and extraction rules for features and combined with the mask extraction method.

9. An electronic device, characterized in that, The device includes: a processor and a memory; The memory is used to store one or more program instructions; The processor is used to run one or more program instructions to execute the steps of a method for discriminating features in a desert area based on remote sensing image interpretation as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, the steps of a method for discriminating features in a desert area based on remote sensing image interpretation as described in any one of claims 1 to 7 are implemented.

Images