A remote sensing classification method and device for periglacial landforms

Abstract

This invention discloses a remote sensing classification method and device for periglacial landforms. The method includes: establishing remote sensing classification rules for periglacial landforms, dividing periglacial landforms into primary and secondary periglacial landform types; collecting remote sensing images and digital elevation model data of the target area in permafrost regions at various resolutions, and performing radiometric calibration and atmospheric correction on the remote sensing images; performing first-scale segmentation and second-scale segmentation on the processed data to generate two-level image object layers, with the first-scale segmentation generating a primary image object layer corresponding to the primary periglacial landform type and the second-scale segmentation generating a secondary image object layer corresponding to the secondary periglacial landform type; extracting and analyzing features for each image object in the primary and secondary image object layers; and, based on the remote sensing classification rules for periglacial landforms, classifying the image objects in the primary and secondary image object layers into the corresponding primary and secondary periglacial landform types, respectively, and outputting a remote sensing classification map of periglacial landforms.

Description

Technical Field

[0001] This invention relates to the field of remote sensing classification technology, and in particular to a remote sensing classification method and device for periglacial landforms. Background Technology

[0002] Periglacial landforms are unique systems formed by the interaction between the active permafrost layer and the surface in cold regions. Their distribution and morphology are core data for global climate change monitoring, engineering safety in cold regions, and ecological protection. However, periglacial landforms are mostly located in high-altitude, high-latitude extreme cold regions, and field surveys face insurmountable limitations. First, they lack spatiotemporal continuity; snow cover and short-term melting periods in cold regions only allow for sporadic sampling, making dynamic monitoring impossible. Second, data accuracy is low; micro-scale units such as frost heave mounds and thermomelt collapses are difficult to accurately map, and human interpretation is highly subjective, resulting in poor consistency. Third, they are costly and risky; transportation in cold regions is inconvenient, and there are hazards such as avalanches and crevasses, making a single large-scale survey cost several times more than that in ordinary areas, which cannot meet the needs of routine surveys.

[0003] Currently, remote sensing technology has become the mainstream extraction method, but existing automatic classification methods are not well adapted to periglacial landform features:

[0004] First, traditional pixel-based methods have fundamental shortcomings. Periglacial landforms have continuous spatial structures and hierarchical distribution characteristics. Macroscopically, they are distributed in bands along elevation and slope, and microscopically, complete morphological recognition is required. However, pixel-level classification only relies on spectral information, which severs spatial correlation.

[0005] Secondly, while deep learning semantic segmentation has advantages, it faces significant bottlenecks in extracting periglacial landforms. Firstly, the black-box model is detached from geoscientific logic and cannot embed the coupling patterns of altitude, slope, and vegetation, potentially misclassifying low-altitude bare land as freeze-thaw bare land and high-altitude meadows as creeping soil slopes, resulting in a lack of geographical plausibility. Secondly, the contradiction between sample dependence and the scarcity of cold-region data is prominent. Deep learning requires thousands to tens of thousands of labeled samples, but cold-region images are affected by snow cover and cloud cover, making accurate sample acquisition costly and model generalization ability poor. Thirdly, the fusion of multi-source data is weak, making it difficult to integrate remote sensing spectral, textural, and DEM topographic features, resulting in insufficient accuracy in extracting topographically dependent landforms such as debris slopes and frost heave hills.

[0006] In summary, current remote sensing extraction of periglacial landforms faces challenges such as unsustainable field surveys, pixel classification disrupting spatial structure, and deep learning lacking geoscientific logic. There is an urgent need for new methods that integrate geographical laws and multi-source data to solve the problem of high-precision, high-reliability, and high-efficiency extraction of periglacial landforms in cold regions. Summary of the Invention

[0007] The purpose of this invention is to provide a remote sensing classification method and device for periglacial landforms, which solves the technical problems of limited field surveys and insufficient hierarchical reasoning in existing object-oriented methods.

[0008] To achieve the above objectives, the present invention provides a remote sensing classification method for periglacial landforms, the method comprising:

[0009] Establish remote sensing classification rules for periglacial landforms, dividing them into primary periglacial landform types and secondary periglacial landform types;

[0010] Remote sensing images and digital elevation model data of the target area in the permafrost region at various resolutions were collected, and radiometric calibration and atmospheric correction were performed on the remote sensing images.

[0011] The processed data is segmented at a first scale and at a second scale to generate two levels of image object layers. The first scale segmentation generates a first-level image object layer corresponding to the first-level periglacial landform type, and the second scale segmentation generates a second-level image object layer corresponding to the second-level periglacial landform type.

[0012] Extract and analyze the features of each image object in the first-level image object layer and the second-level image object layer;

[0013] Based on the remote sensing classification rules for periglacial landforms and the extracted image object features, the image objects in the first-level image object layer and the second-level image object layer are respectively classified into the corresponding first-level periglacial landform type and second-level periglacial landform type, and a remote sensing classification map of periglacial landforms is output.

[0014] Preferably, the method for classifying the primary periglacial landform type includes:

[0015] The primary periglacial landform types are divided into: steep mountain slope bedrock erosion-deposition development zone, medium and gentle slope soil and rock transport-deposition development zone, and low and gentle flat land soil and rock deposition-erosion development zone.

[0016] Preferably, the method for classifying periglacial landforms into primary periglacial landform types and secondary periglacial landform types includes:

[0017] The bedrock erosion-deposition development zone on the steep mountain slope includes two secondary periglacial landform types: a cold peneplain and a debris slope.

[0018] The medium-slow slope soil and rock transport-deposition development zone includes two secondary periglacial landform types: stony glacial deposition surface and creeping soil and rock slope surface.

[0019] The low-lying, flat, sedimentary-erosional zone includes three secondary periglacial landform types: frost heave hills, freeze-thaw bare land, and thermocline collapse land.

[0020] Preferably, the method for classifying image objects in the primary image object layer and the secondary image object layer to the corresponding primary and secondary periglacial landform types based on the periglacial landform remote sensing classification rules includes:

[0021] If the elevation is greater than 4300 meters, the average slope is greater than 35°, the slope at the top of the slope is less than 8°, the normalized vegetation index is between 0 and 0.1, and the normalized water index is less than 0.3, it can be judged as the bedrock erosion-deposition development zone of the steep mountain slope.

[0022] If the elevation is between 4100 meters and 4300 meters, the average slope is between 20° and 35°, and the normalized vegetation index is between 0 and 0.1, it can be identified as the medium-slow slope soil transport-deposition development zone.

[0023] If the elevation is between 3900 meters and 4100 meters, the average slope is less than 20°, and the normalized vegetation index is between 0.1 and 0.4, it can be identified as the low-lying flat land sedimentary-erosion development zone.

[0024] The rest of the landforms are non-permafrost areas.

[0025] Preferably, the method for classifying the bedrock erosion-deposition development zone on steep slopes of permafrost mountains into the corresponding secondary periglacial landform type includes:

[0026] If the slope at the top of the slope is less than 8°, the texture entropy is greater than 5.0, the normalized water index is less than 0.3, and the normalized vegetation index is less than 0.1, it can be determined to be the so-called cold peneplain.

[0027] If the average slope is greater than 35°, the texture entropy is less than 2.0, and the aspect ratio is greater than 2, it can be determined to be the rock debris slope.

[0028] Preferably, the method for classifying the medium-slow slope soil transport-deposition development zone to the corresponding secondary periglacial landform type includes:

[0029] If the aspect ratio of the object is greater than 2, the compactness of the object is less than 0.4, the average slope of the downstream boundary of the object is greater than 30°, and the texture entropy is greater than 4.0, it can be determined to be the stone glacier deposition surface.

[0030] If the texture entropy is less than 3.0 and the normalized vegetation index is greater than 0.1, it can be determined to be the creeping rock and soil slope.

[0031] Preferably, the method for classifying the low-lying, flat, sedimentary-erosional development zone to the corresponding secondary periglacial landform type includes:

[0032] If the average curvature is greater than 0.3 and the normalized vegetation index is greater than 0.3, it can be determined to be the frost-heavy hill.

[0033] If the average curvature is less than -0.3, the normalized water index is greater than 0.3, or the variance of the internal normalized vegetation index is greater than 0.5, it can be determined to be the aforementioned thermocline collapse area;

[0034] If the average curvature is within ±0.1, the texture entropy is greater than 6.0, and the normalized vegetation index is less than 0.05, it can be determined as the frozen-thawed bare land.

[0035] Preferably, after performing radiometric calibration and atmospheric correction on the remote sensing image, the remote sensing image and DEM data are unified into the same coordinate system.

[0036] Preferably, the method for performing first-scale segmentation and second-scale segmentation to generate a two-level image object layer includes:

[0037] According to the multi-resolution segmentation algorithm, the first segmentation scale parameter is set to 200 to segment primary periglacial landforms; the second segmentation scale parameter is set to 60 to segment secondary periglacial landforms.

[0038] Secondly, the present invention provides a remote sensing classification device for periglacial landforms, the device comprising:

[0039] One or more processors;

[0040] A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the periglacial landform remote sensing classification method as described in the first aspect.

[0041] Compared with the prior art, the technical advantages of the present invention are as follows:

[0042] The remote sensing classification method for periglacial landforms provided by this invention incorporates hierarchical reasoning logic. By using multi-scale segmentation and multi-source feature fusion, pixel noise is eliminated. Moreover, it can accurately distinguish multi-level periglacial landforms without requiring a large number of labeled samples, thereby significantly improving the spatial consistency and reliability of the classification results. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of a remote sensing classification method for periglacial landforms provided in this embodiment.

[0044] Figure 2 This is the periglacial landform after first-scale segmentation provided in this embodiment;

[0045] Figure 3 This is the periglacial landform after second-scale segmentation provided in this embodiment;

[0046] Figure 4This is a schematic diagram of a remote sensing classification device for periglacial landforms provided in Embodiment 2. Detailed Implementation

[0047] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but it should be understood that the scope of protection of the present invention is not limited to the specific embodiments.

[0048] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0049] Example 1:

[0050] This embodiment provides a remote sensing classification method for periglacial landforms, such as... Figure 1 As shown, the method includes the following steps:

[0051] S101: Establish remote sensing classification rules for periglacial landforms, and divide periglacial landforms into primary periglacial landform types and secondary periglacial landform types.

[0052] According to different scales, periglacial landforms are divided into two levels, and the inclusion relationship between the two levels of periglacial landform types is shown in Table 1.

[0053] Table 1 Classification of Periglacial Landforms

[0054]

[0055] In this first embodiment, as Figure 2 As shown, the classification criteria for primary periglacial landforms are as follows:

[0056] Steep slope bedrock erosion-deposition zone: The elevation is above 4300 meters, the slope is greater than 35°, the slope at the top is less than 8°, the main body is bedrock and weathered debris, and there is no vegetation cover.

[0057] Medium-slow slope soil transport-deposition development zone: elevation between 4100-4300 meters, slope between 20-35°, vegetation cover is generally good, normalized vegetation index (NDVI) is between 0 and 0.1.

[0058] Low-lying flat soil and rock deposition-erosion development zone: elevation between 3900-4100 meters, slope less than 20°, with good vegetation cover and normalized difference vegetation index (NDVI) between 0.1 and 0.4.

[0059] Wherein, slope = arctan(elevation difference / horizontal distance).

[0060] In this first embodiment, as Figure 3As shown, the classification criteria for secondary periglacial landforms are as follows:

[0061] Cold-formed peneplains: flat or undulating plateaus or terraces, usually large in area; with gentle slopes, typically <10°; appearing as a flat "top surface," resembling a platform or steps in a landscape profile. One side or perimeter often connects to a steep cliff, representing a remnant surface left after the mountain recedes.

[0062] Rock debris slope: A continuous, straight or slightly convex slope that usually extends down from the edge of a ridge or peneplain to the valley floor; the slope is relatively steep, usually between 25° and 35° (close to the natural angle of repose of the clastic material); it appears as a linear slope connecting a cliff and the valley floor, with the edge of the cliff or peneplain above and the valley floor or alluvial fan below.

[0063] Lithoglacial deposits: typically tongue-shaped or leaf-shaped with clear boundaries, often featuring an arc-shaped advancing edge. The morphology is complete, resembling a "fluid." Steep and clearly defined, the slope often approaches the angle of repose of the clastic material. The surface material is rough, with numerous boulders, resulting in very coarse textures and uneven color tones in images. Due to the difficulty in vegetation growth, the Normalized Difference Vegetation Index (NDVI) value is low.

[0064] Creep-flowing soil slopes: These slopes are sheet-like, sheet-like, or strip-like, with relatively indistinct boundaries and a natural transition to the surrounding slopes. They are irregular in shape, with a gentle, transitional gradient and no obvious steep drops. The surface is relatively smooth, with more fine-grained material and a relatively uniform texture. The normalized difference vegetation index (NDVI) values ​​may be uneven due to sparse vegetation.

[0065] Frost-heavy mounds: isolated, round or elliptical protrusions, or distributed in a beaded pattern. They appear as local high points in digital elevation models (DEMs) or through topographic relief analysis. The surface is relatively smooth, especially at the top. Due to favorable moisture conditions, meadows may grow, forming "grass mounds." With vegetation cover, the Normalized Difference Vegetation Index (NDVI) value is higher; without vegetation cover, the spectrum is similar to that of the surrounding wetlands.

[0066] Freeze-thaw bare land: Large areas distributed in patches or strips, with an irregular but generally flat surface. There are no significant topographic reliefs; the surface may feel slightly rough due to disturbance, but there are no systematic protrusions or depressions. Extremely rough and fragmented, exhibiting a "honeycomb" or "spotted" pattern. The Normalized Difference Vegetation Index (NDVI) value is extremely low (almost no vegetation), and its spectral characteristics are similar to bare soil and rock, with a lighter and brighter hue.

[0067] Thermal melt subsidence areas: These are circular, elliptical, or irregularly shaped depressions, often occurring in clusters. On a digital elevation model (DEM), they appear as localized depressions with distinct rims (slopes). The texture is complex, with significant differences between the interior and surrounding areas. The bottom of the depression may be filled with water (spectral characteristics resembling water) or have recovered vegetation (spectral characteristics resembling vegetation). Waterlogged depressions resemble water in spectrum and have high Normalized Difference Water Index (NDWI) values, while vegetated depressions show locally high Normalized Difference Vegetation Index (NDVI) values.

[0068] S102: Collect multi-source remote sensing images and digital elevation model data of various resolutions for the target area in the permafrost region, and perform radiometric calibration and atmospheric correction on the remote sensing images.

[0069] In this first embodiment, the remote sensing data comes from Gaofen-1 and Gaofen-2 optical imagery, both with a resolution better than 2 meters. The terrain data comes from ASTER GDEM V3, and the digital elevation model data has a resolution of 30 meters.

[0070] In this first embodiment, the collected raw data includes two core data types with different spatial resolutions. The optical remote sensing image uses a high-resolution data source with a resolution better than 2 meters, which can accurately capture the microscopic morphological details of secondary periglacial landforms, such as the circular outline of frost heaves, the tongue-shaped boundary of stony glaciers, and the pit edge structure of thermal melt collapses, providing a clear foundation for the extraction of texture and shape features. The digital elevation model (DEM) uses a medium-resolution data source of 30 meters to calculate the macroscopic topographic features of the target area, such as the overall elevation range and slope gradient distribution, providing key topographic basis for the division of primary periglacial landform development zones. The two achieve an organic combination of microscopic detail recognition and macroscopic range coverage through resolution synergy, providing comprehensive data support for subsequent hierarchical classification.

[0071] In this first embodiment, the remote sensing image is radiometrically calibrated and subjected to FLAASH atmospheric correction, and then unified with the DEM data to the CGCS 2000 coordinate system.

[0072] S103: Perform first-scale segmentation and second-scale segmentation on the processed data to generate two-level image object layers. The first-scale segmentation generates a first-level image object layer corresponding to the first-level periglacial landform type, and the second-scale segmentation generates a second-level image object layer corresponding to the second-level periglacial landform type.

[0073] In this first embodiment, multiple resolution segmentation algorithms are used in the eCognition Developer software. Near-infrared, short-wave infrared, and DEM data are assigned weights of over 20%.

[0074] In this first embodiment, the first-scale segmentation parameter is 200, used for segmenting primary periglacial landforms. The second-scale segmentation parameter is 60, used for segmenting secondary periglacial landforms. The segmentation scale parameter is the heterogeneity threshold used by the algorithm to determine whether adjacent pixels can be merged into a single image object. Heterogeneity represents the degree of difference in spectral, topographic, and other features between pixels: when the heterogeneity of adjacent pixels is lower than the parameter threshold, they will be aggregated into the same image object; when the heterogeneity is higher than the threshold, they will be divided into different objects. The value of the segmentation scale parameter is positively correlated with the size of the final generated image object: the larger the parameter value, the higher the allowed heterogeneity, the more pixels can be aggregated, and the larger the generated image object; the smaller the parameter value, the lower the allowed heterogeneity, the fewer pixels aggregated, and the smaller the generated image object. This can be implemented using conventional professional software, and will not be elaborated further in this first embodiment.

[0075] S104: Extract and analyze spectral and DEM-derived terrain features for each image object in the first-level image object layer and the second-level image object layer, and additionally extract and analyze texture and shape features for the image objects in the second-level image object layer.

[0076] In this first embodiment, the image object features include spectral features and DEM-derived terrain features, and the features of the secondary image object further include texture features and shape features.

[0077] Spectral characteristics: such as Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI).

[0078] Texture features: based on entropy and contrast of the gray-level co-occurrence matrix (GLCM).

[0079] Shape characteristics: area, length / width, compactness.

[0080] Topographic features: elevation, slope.

[0081] S105: Based on the remote sensing classification rules for periglacial landforms, the image objects in the first-level image object layer and the second-level image object layer are respectively classified into the corresponding first-level periglacial landform type and second-level periglacial landform type, completing the remote sensing classification of periglacial landforms in the target area and outputting the periglacial landform remote sensing classification map.

[0082] In this first embodiment, a hierarchical classification decision tree is constructed. The decision tree contains multiple hierarchical nodes, and each node corresponds to a classification rule based on object features. Through layer-by-layer discrimination, image objects are classified into different periglacial landform types. The decision tree logic is as follows:

[0083] If the elevation is greater than 4300 meters and the average slope is greater than 35°, it is a zone where bedrock erosion and deposition develop on steep slopes at the top of permafrost mountains.

[0084] If the elevation is between 4100 meters and 4300 meters and the average slope is between 20° and 35°, it is a gentle slope soil and rock transport and deposition development zone in the permafrost region;

[0085] If the elevation is between 3900 meters and 4100 meters and the average slope is less than 20°, it is considered a low-lying, flat, sedimentary-erosion zone in the permafrost region.

[0086] Otherwise, it is not a permafrost region.

[0087] For objects in the bedrock erosion-deposition development zone on steep mountain slopes:

[0088] If the slope at the top of the slope is less than 8°, the texture entropy is greater than 5.0, the NDWI is less than 0.3, and the NDVI is less than 0.1, then it is a cold-grown peneplain.

[0089] If the average slope is >35°, texture entropy is <2.0, aspect ratio is greater than 2, NDWI is <0.3, and NDVI is <0.1, it is a rock debris slope.

[0090] For objects in the soil and rock transport-deposition development zone on gentle slopes:

[0091] If the aspect ratio of an object in a medium-slow slope soil transport-deposition development zone is >2, the object compactness is <0.4, the average slope of the downstream boundary of the object is >30°, and the texture entropy is >4.0, it is considered a rock-glacial deposition surface;

[0092] If the texture entropy is <3.0 and the mean NDVI is >0.1, it is considered a creeping rock and soil slope.

[0093] For objects in low-lying, flat, sedimentary-erosion zones:

[0094] If the mean curvature is less than -0.3 and (mean NDWI > 0.3 OR internal NDVI variance > 0.5), it is a thermal fusion collapse land.

[0095] If the average curvature is close to 0, the texture entropy is >6.0, and the average NDVI is <0.05, it is considered a frozen-thawed bare land.

[0096] The variance of the normalized vegetation index within an object is calculated based on the image object generated by multi-scale segmentation, that is, the independent geographic unit corresponding to micro-landforms such as thermal collapse land. In this first embodiment, the normalized vegetation index value is first calculated pixel by pixel for the preprocessed remote sensing image, and then the normalized vegetation index data of all pixels within the boundary of the current image object to be analyzed are extracted, and the result is obtained through the statistical variance formula.

[0097] The core function of the normalized vegetation index variance within an object is to distinguish between waterlogged areas and vegetation restoration areas within a thermocline collapse site. The normalized vegetation index value is extremely low in the waterlogged areas within a thermocline collapse site, while it is relatively high in the vegetation restoration areas. When the difference in the normalized vegetation index value of a pixel is large, the variance will exceed 0.5, thereby enabling accurate identification of the landform of a thermocline collapse site.

[0098] Example 2:

[0099] This second embodiment provides a remote sensing classification device for periglacial landforms, such as... Figure 4 As shown, the device includes:

[0100] One or more processors;

[0101] A storage device for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the periglacial landform remote sensing classification method as described in Embodiment 1.

[0102] Figure 4 This is a schematic diagram of the structure of the periglacial landform remote sensing classification device provided in Embodiment 2 of the present invention. Figure 4 The remote sensing classification device for periglacial landforms shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0103] like Figure 4 As shown, the periglacial landform remote sensing classification device is presented in the form of a general-purpose device. The components of the periglacial landform remote sensing classification device may include, but are not limited to: one or more processors or processing units, memory, and buses connecting different system components (including memory and processing units).

[0104] A bus refers to one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0105] Remote sensing classification equipment for periglacial landforms typically includes a variety of computer-readable media. These media can be any available media that can be accessed by the remote sensing classification equipment for periglacial landforms, including volatile and non-volatile media, and portable and non-portable media.

[0106] The memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory. The periglacial landform remote sensing classification device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system may be used to read and write non-removable, non-volatile magnetic media (…). Figure 4 Not shown; usually referred to as a "hard drive"). Although Figure 4 As not shown, disk drives for reading and writing to removable non-volatile disks (e.g., "floppy disks") and optical disc drives for reading and writing to removable non-volatile optical discs (e.g., CD-ROMs, DVD-ROMs, or other optical media) may be provided. In these cases, each drive may be connected to a bus via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0107] A program / utility having a set (at least one) of program modules can be stored, for example, in memory. Such program modules include, but are not limited to, an operating system, one or more applications, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules typically perform the functions and / or methods described in the embodiments of this invention.

[0108] The periglacial landform remote sensing classification device can also communicate with one or more external devices (e.g., keyboard, pointing device, display, etc.), one or more devices that enable users to interact with the device, and / or any device that enables the device to communicate with one or more other devices (e.g., network card, modem, etc.). This communication can be performed through an input / output (I / O) interface. Furthermore, the device can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via a network adapter. Figure 4 As shown, the network adapter communicates with other modules of the periglacial landform remote sensing classification device via a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the periglacial landform remote sensing classification device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0109] The processing unit executes various functional applications and data processing by running programs stored in the memory, such as implementing the periglacial landform remote sensing classification method provided in any embodiment of the present invention.

[0110] It is worth noting that the information interaction and execution process between the modules and units in the above-mentioned device and system are based on the same concept as the processing method embodiment of the present invention. For details, please refer to the description in the method embodiment of the present invention, and will not be repeated here.

[0111] Those skilled in the art will understand that all or part of the steps in the various methods of the embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

Claims

1. A method of periglacial landform remote sensing classification, characterized in that, The method includes: Establishing remote sensing classification rules for periglacial landforms, and classifying periglacial landforms into primary periglacial landform types and secondary periglacial landform types; Among them, the primary periglacial landform types are divided into: the weathering-accumulation development zone of bedrock on the steep slopes of mountaintops, the transportation-accumulation development zone of rock and soil on the medium gentle slopes, and the sedimentation-erosion development zone of rock and soil on the low gentle flatlands; Among them, in the weathering-accumulation development zone of bedrock on the steep slopes of mountaintops, there are two secondary periglacial landform types, namely cryoplanation surfaces and debris slopes; In the transportation-accumulation development zone of rock and soil on the medium gentle slopes, there are two secondary periglacial landform types, namely accumulation surfaces of rock glaciers and creep slopes of rock and soil; In the sedimentation-erosion development zone of rock and soil on the low gentle flatlands, there are three secondary periglacial landform types, namely frost heave mounds, frost-thawed bare lands, and thermokarst collapse lands; Collecting remote sensing images and digital elevation model data with multiple resolutions in the target area of the permafrost region, and performing radiometric calibration and atmospheric correction processing on the remote sensing images; Performing first-scale segmentation and second-scale segmentation on the processed data respectively to generate two-level image object layers. The first-scale segmentation generates a primary image object layer corresponding to the primary periglacial landform types, and the second-scale segmentation generates a secondary image object layer corresponding to the secondary periglacial landform types; Extracting and analyzing the image object features for each image object in the primary image object layer and the secondary image object layer; Based on the remote sensing classification rules for periglacial landforms and the extracted image object features, classifying the image objects in the primary image object layer and the secondary image object layer into the corresponding primary periglacial landform types and secondary periglacial landform types respectively, and outputting a remote sensing classification map of periglacial landforms; Among them, the method for classifying the image objects in the primary image object layer into the corresponding primary periglacial landform types includes: If the elevation of the image object is greater than 4300 meters, the average slope is greater than 35°, the slope at the top of the slope is less than 8°, the normalized difference vegetation index is between 0 and 0.1, and the normalized difference water index is less than 0.3, then it is classified into the weathering-accumulation development zone of bedrock on the steep slopes of mountaintops; If the elevation of the image object is between 4100 meters and 4300 meters, the average slope is between 20° and 35°, and the normalized difference vegetation index is between 0 and 0.1, then it is classified into the transportation-accumulation development zone of rock and soil on the medium gentle slopes; If the elevation of the image object is between 3900 meters and 4100 meters, the average slope is less than 20°, and the normalized difference vegetation index is between 0.1 and 0.4, then it is classified into the sedimentation-erosion development zone of rock and soil on the low gentle flatlands.

2. The ice-marginal landform remote sensing classification method of claim 1, wherein, The method for classifying the weathering-accumulation development zone of bedrock on the steep slopes of mountaintops in the permafrost region into the corresponding secondary periglacial landform types includes: If the slope at the top of the slope is less than 8°, the texture entropy is greater than 5.0, the normalized difference water index is less than 0.3, and the normalized difference vegetation index is less than 0.1, it can be judged as the cryoplanation surface; If the average slope is greater than 35°, the texture entropy is less than 2.0, and the aspect ratio is greater than 2, it can be judged as the debris slope.

3. The ice-marginal landform remote sensing classification method of claim 1, wherein, The method for classifying the transportation-accumulation development zone of rock and soil on the medium gentle slopes into the corresponding secondary periglacial landform types includes: If the aspect ratio of the object is greater than 2, the compactness of the object is less than 0.4, the average slope of the downstream boundary of the object is greater than 30°, and the texture entropy is greater than 4.0, it can be determined as the rock glacier accumulation surface; If the texture entropy is less than 3.0 and the normalized vegetation index is greater than 0.1, it can be determined as the creep soil slope surface.

4. The periglacial landform remote sensing classification method according to claim 1, wherein The method for dividing the gentle flat ground geotechnical sedimentation-erosion development zone into the corresponding secondary periglacial landform types includes: If the average curvature is greater than 0.3 and the normalized vegetation index is greater than 0.3, it can be determined as the frost heave mound land; If the average curvature is less than -0.3, the normalized water body index is greater than 0.3 or the variance of the internal normalized vegetation index is greater than 0.5, it can be determined as the thermokarst depression land; If the average curvature is within the range of ±0.1, the texture entropy is greater than 6.0, and the normalized vegetation index is less than 0.05, it can be determined as the freeze-thaw bare land.

5. The periglacial landform remote sensing classification method according to claim 1, wherein After performing radiometric calibration and atmospheric correction processing on the remote sensing image, the remote sensing image and the DEM data are unified into the same coordinate system.

6. The periglacial landform remote sensing classification method according to claim 1, wherein The method for performing the first-scale segmentation and the second-scale segmentation to generate two-level image object layers includes: According to the multi-resolution segmentation algorithm, set the first segmentation scale parameter to 200 for segmenting the primary periglacial landform; set the second segmentation scale parameter to 60 for segmenting the secondary periglacial landform.

7. A periglacial landform remote sensing classification device, characterized in that the device Includes: One or more processors; A storage device for storing one or more programs, which when executed by the one or more processors, cause the one or more processors to implement the periglacial landform remote sensing classification method according to any one of claims 1 to 6.

Images