CN113822141A - Automatic glacier and snow extraction method and system based on remote sensing image - Google Patents
Automatic glacier and snow extraction method and system based on remote sensing image Download PDFInfo
- Publication number
- CN113822141A CN113822141A CN202110854709.8A CN202110854709A CN113822141A CN 113822141 A CN113822141 A CN 113822141A CN 202110854709 A CN202110854709 A CN 202110854709A CN 113822141 A CN113822141 A CN 113822141A
- Authority
- CN
- China
- Prior art keywords
- snow
- remote sensing
- glacier
- range
- image
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Abstract
The invention discloses a glacier and snow automatic extraction method based on remote sensing images, which comprises the following steps of: acquiring a satellite remote sensing image of a region to be extracted, and preprocessing the satellite remote sensing image; roughly extracting glacier and snow from the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and determining a suspected range; acquiring DEM auxiliary data of a region to be extracted, and calculating multi-dimensional features including spectral values, standard deviations and DEM features by combining remote sensing images; masking the multi-dimensional features in the area to be extracted by utilizing glacier and snow suspected ranges to obtain the multi-dimensional features in the suspected ranges; and collecting samples in a suspected range, selecting multi-dimensional characteristics, and realizing fine extraction of glaciers and accumulated snow by using a decision tree model.
Description
Technical Field
The invention relates to the technical field of remote sensing measurement, in particular to a method and a system for automatically extracting glaciers and accumulated snow based on remote sensing images.
Background
Glaciers and snow are important fresh water resources on the ground surface and generally exist in high latitude areas or high mountain areas with medium and low latitude. On a remote sensing image, glaciers and snow cover are represented as snow cover types, different from other ground object cover types, snow cover has unique spectral characteristics, high reflectivity in a visible light range is represented, the reflectivity is gradually reduced in a near infrared region, the reflectivity is reduced in a short wave infrared region, and the spectral characteristics of the snow cover are a physical basis for optical remote sensing monitoring of the snow cover.
At present, the common methods for extracting the accumulated snow comprise: 1) extracting the band operation; the band operation includes a band ratio method and a normalized snow index method, and the method has the defects of single characteristic, easy error extraction and unstable effect. 2) Extracting based on a classification algorithm; the extraction based on the classification algorithm comprises supervised classification and unsupervised classification, the unsupervised classification method does not need prior knowledge and performs classification according to the spectral statistical characteristics of the ground features, and when the spectral feature difference corresponding to the two ground feature types is small, the classification effect is not as good as that of the supervised classification, and in comparison, the supervised classification method is more effective. 3) Extracting information by an object-oriented technology; information extraction of object-oriented techniques is mainly used for high spatial resolution images, and is not suitable for medium and low resolution images. The spectral information of the high-resolution image is deficient, and the phenomena of 'same-object different spectrum and same-spectrum foreign matter' exist obviously, and the object-oriented technology is to break a research object into a homogeneous image spot object and perform information identification by combining multiple characteristics of textures, spectrums and shapes. In the prior art, the method for extracting the accumulated snow in the alpine mountain areas by using the high-resolution remote sensing images is to combine a depth confidence network with a fully-connected conditional random field model by using an object-oriented idea, fully utilize the spectrum, shape, texture and spatial relationship information of the high-resolution images and improve the accuracy of the accumulated snow extraction.
In order to improve the accuracy of snow extraction, more and more scholars comprehensively utilize multi-source data and multi-dimensional characteristics to perform snow extraction. For example, the NOAA AVHRR and TM information compounding technology is used for extracting the snow cover with high precision; roughly extracting an accumulated snow range by utilizing the normalized accumulated snow index and the reflectivity threshold of each wave band, and further identifying the accumulated snow through a DBN (database-based network); the method of the RS and the GIS is integrated, and the multi-source data such as the ASTER image, the DEM and the like are combined to identify glaciers and the like, and the result shows that the characteristics derived from the DEM data source have an auxiliary effect on the extraction precision.
With the increase of remote sensing satellite data and the annual accumulation of auxiliary data, how to utilize multi-source data and multi-dimensional characteristics to improve the accuracy and universality of snow detection is a key point of attention of people. In addition, in alpine mountain areas and arid and semiarid areas in China, the accumulated snow is an important fresh water resource for storage, and natural disasters such as snow slide, wind snow blowing and the like caused by the accumulated snow greatly threaten the safety of human life and property. Therefore, it is very significant to explore a new high-precision accumulated snow extraction method.
Disclosure of Invention
Aiming at the problems in the related art, the invention provides a method and a system for automatically extracting glaciers and snow based on remote sensing images, which are used for analyzing ground object characteristics from multi-dimensional characteristics, effectively fusing DEM and spectral information and avoiding the condition of low identification precision of the glaciers and the snow due to single characteristic dimension; meanwhile, a snow cover index method and a decision tree algorithm are comprehensively utilized, and on the basis of extraction of the snow cover index method, the advantages of data type and conventional type attributes can be simultaneously processed by the decision tree algorithm, so that glacier and snow cover extraction rules are excavated, and glacier and snow cover extraction accuracy is improved.
The technical scheme of the invention is realized as follows:
according to one aspect of the invention, a method for automatically extracting glaciers and snow based on remote sensing images is provided.
The glacier and snow automatic extraction method based on the remote sensing image comprises the following steps:
acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image, and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
roughly extracting glaciers and snow on the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and determining suspected ranges of the glaciers and the snow;
acquiring DEM auxiliary data in a region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
carrying out mask processing on DEM characteristic data and spectral characteristic data in the area to be extracted by utilizing glacier and snow suspected range to obtain multi-dimensional characteristics in the suspected range;
collecting samples of glaciers and snow in a suspected range and samples of non-glaciers and snow in a suspected range, selecting multi-dimensional characteristics, and finely extracting the glaciers and the snow by using a pre-configured decision tree model to obtain a glacier and snow range.
Wherein, the preprocessing the satellite remote sensing image comprises the following steps: and carrying out atmospheric correction processing, image registration processing and image fusion processing on the satellite remote sensing image.
In addition, when a suspected range is determined by roughly extracting glacier and snow from the preprocessed satellite remote sensing image by using a normalized snow index threshold method, the snow index NDSI is normalized, wherein Green represents a Green wave band ranging from 0.5 μm to 0.6 μm; SWIR1 represents the short-wave infrared band in the range of 1.55 μm to 1.75 μm.
The algorithm of the decision tree model is a CART algorithm, and a commonly used niki index is used as an index of attribute measurement.
In addition, the method further comprises: and taking the glacier and snow cover range data which is drawn manually as truth value data, and analyzing the accuracy of the glacier and snow cover range which is obtained by fine recognition, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow cover by the predicted total area of the glacier and snow cover.
According to another aspect of the invention, an automatic glacier and snow extraction system based on remote sensing images is provided.
This glacier and snow automatic extraction system based on remote sensing image includes:
the data acquisition module is used for acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
the rough extraction module is used for carrying out glacier and snow rough extraction on the preprocessed satellite remote sensing image by utilizing a normalized snow index threshold method to determine a glacier and snow suspected range;
the characteristic calculation module is used for acquiring DEM auxiliary data in the region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
the characteristic masking module is used for masking DEM characteristic data and spectral characteristic data in the area to be extracted by utilizing the suspected ranges of glaciers and snow to obtain multi-dimensional characteristics in the suspected ranges;
and the range identification module is used for acquiring samples of glaciers and snow in a suspected range and samples of non-glaciers and snow, selecting multi-dimensional characteristics, and performing fine extraction on the glaciers and the snow by using a pre-configured decision tree model to obtain a glacier and snow range.
The data acquisition module is used for preprocessing the satellite remote sensing image and comprises the following steps: and carrying out atmospheric correction processing, image registration processing and image fusion processing on the satellite remote sensing image.
In addition, the rough extraction module performs glacier and snow rough extraction on the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and normalizes a snow index NDSI when determining a suspected range, wherein Green represents a Green wave band in a range of 0.5 μm to 0.6 μm; SWIR1 represents the short-wave infrared band in the range of 1.55 μm to 1.75 μm.
The algorithm of the decision tree model is a CART algorithm, and a commonly used niki index is used as an index of attribute measurement.
In addition, the above system further comprises: and the accuracy evaluation module is used for analyzing the accuracy of the glacier and snow range obtained by fine recognition by taking the manually drawn glacier and snow range data as truth value data, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow by the predicted total area of the glacier and snow.
Has the advantages that:
1) the method disclosed by the invention integrates DEM and spectral characteristics to generate multi-dimensional characteristics, and is beneficial to recognition of glaciers and snow. In a medium and low-dimensional area, non-seasonal snow has a certain rule on the DEM, and compared with the method of simply extracting snow by utilizing gray information of each wave band of an image, the method integrates the characteristics of the DEM, and can extract snow results with higher precision.
2) The characteristic of the accumulated snow on the remote sensing satellite image is distinct, and a corresponding logic expression can be easily deduced by utilizing a decision tree model. Under the condition of monitoring glaciers and snow in the same area for a long time, the glaciers and the snow can be extracted quickly and accurately according to the law expressed by the glaciers and the snow in multiple dimensions.
3) The snow cover index and the decision tree model are integrated to extract the accumulated snow, and therefore the false extraction caused by single wave band operation or decision tree model information extraction is avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a schematic flow chart of a method for automatically extracting glacier and snow based on remote sensing images according to an embodiment of the invention;
FIG. 2 is a block diagram schematically illustrating a structure of an automatic glacier and snow extraction system based on remote sensing images according to an embodiment of the present invention;
FIG. 3 is a remote sensing fused image map according to an embodiment of the present invention;
FIG. 4 is a diagram of NDSI feature results of a computed remote sensing fused image according to an embodiment of the invention;
FIG. 5 is a graph of results of NDSI threshold extraction of glaciers and snow according to an embodiment of the present invention;
FIG. 6 is a diagram of a result of detailed recognition of glaciers and snow according to an embodiment of the present invention;
FIG. 7 is a result image of information extraction by fusing DEM and spectral features according to an embodiment of the present invention;
FIG. 8 is a resulting image of information extraction with only input spectral features according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.
According to the embodiment of the invention, a method and a system for automatically extracting glaciers and snow based on remote sensing images are provided.
As shown in fig. 1, an automatic extraction method of glacier and snow based on remote sensing images according to an embodiment of the present invention includes the following steps:
step S101, acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image, and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
step S102, roughly extracting glaciers and snow from the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and determining suspected ranges of the glaciers and the snow;
s103, acquiring DEM auxiliary data in a region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
step S104, masking DEM characteristic data and spectral characteristic data in the area to be extracted by utilizing glacier and snow suspected range to obtain multi-dimensional characteristics in the suspected range;
step S105, collecting samples of glaciers and snow in a suspected range and samples of non-glaciers and snow, selecting multi-dimensional characteristics, and performing fine extraction on the glaciers and the snow by using a pre-configured decision tree model to obtain a glacier and snow range.
In specific application, the preprocessing of the satellite remote sensing image comprises the following steps: and carrying out atmospheric correction processing, image registration processing and image fusion processing on the satellite remote sensing image. The total radiance of the ground target finally measured by the sensor is not reflected by the real reflectivity of the ground, and the radiant quantity error caused by atmospheric absorption, especially scattering, is included. Therefore, in order to eliminate the radiation error caused by the atmospheric influence, the process of inverting the surface reflectivity of the ground object reality is firstly carried out with atmospheric correction processing on the remote sensing satellite before the extraction of the expansion information. And registration of the unfolded images is required to account for positional errors that exist between the panchromatic and multispectral images. And the image registration adopts relative registration, and the multispectral image is relatively registered by taking the panchromatic data as reference. The remote sensing image fusion is a process of complementing full-color high-spatial resolution and multispectral information, and processing is carried out by using an image fusion algorithm to obtain the remote sensing image which not only comprises spectral information but also has high-spatial resolution information. The method can improve the identification precision of the later information and enlarge the application range and the effect.
In addition, considering the high reflectance of snow in the visible range, the reflectance gradually decreases in the near infrared region and decreases even lower in the short-wave infrared region. The coverage of Snow in the image is highlighted by a combination of visible (mainly green) and short-wave infrared bands using a Normalized Difference Snow Index (NDSI). And carrying out preliminary coarse extraction on glaciers and accumulated snow by using NDSI (New Desorption and design institute), and outputting the result of glaciers and accumulated snow extraction. Specifically, when a normalized snow index threshold value is used for roughly extracting a preprocessed satellite remote sensing image and determining glacier and snow suspected ranges, the normalized snow index NDSI is obtained, wherein Green represents a Green wave band in a range from 0.5 mu m to 0.6 mu m; SWIR1 represents the short-wave infrared band in the range of 1.55 μm to 1.75 μm.
And when calculating the spectral characteristics of the preprocessed image, the spectral characteristics include DN values and standard deviations of each wave band. The DN value can reflect the gray distribution condition of each pixel, and the standard deviation can reflect the dispersion degree between the pixels. For snow-covered surfaces, the standard deviation is generally low, and in addition, the DN value exhibited by snow is high in the visible range.
In addition, when the method is actually applied, the algorithm of the decision tree model is the CART algorithm, and the commonly used nikke index is used as the index of the attribute measurement. The classification tree (decision tree) used is a very common classification method, and is a tree structure, in which each internal node represents a test on an attribute, each branch represents a test output, and each leaf node represents a category. For decision trees, data preparation is often simple or unnecessary and can handle both data-type and conventional-type attributes simultaneously, enabling feasible and effective results for large data sources in a relatively short time.
In addition, the method further comprises: and taking the glacier and snow cover range data which is drawn manually as truth value data, and analyzing the accuracy of the glacier and snow cover range which is obtained by fine recognition, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow cover by the predicted total area of the glacier and snow cover.
As shown in fig. 2, an automatic extraction system of glacier and snow based on remote sensing images according to an embodiment of the present invention includes:
the data acquisition module 201 is used for acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
the rough extraction module 202 is configured to perform rough glacier and snow extraction on the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and determine a suspected glacier and snow range;
the characteristic calculation module 203 is used for acquiring DEM auxiliary data in the region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
the feature masking module 204 is configured to perform masking processing on DEM feature data and spectral feature data in the region to be extracted by using a suspected range of glaciers and snow to obtain a multi-dimensional feature in the suspected range;
the range identification module 205 is configured to collect samples of glaciers and snow in a suspected range and samples of non-glaciers and snow, select multidimensional features, and perform fine extraction on the glaciers and snow by using a pre-configured decision tree model to obtain a glacier and snow range.
In addition, the above system further comprises: and the precision evaluation module (not shown in the figure) is used for analyzing the accuracy of the glacier and snow range obtained by fine recognition by taking the manually drawn glacier and snow range data as truth value data, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow by the predicted total area of the glacier and snow.
For the convenience of understanding the above technical solutions of the present invention, the above technical solutions of the present invention will be described in detail below by taking glaciers and accumulated snow in Qinghai province as an example.
Taking glaciers and snow in Qinghai province as an example, the glaciers in Qinghai province are mainly distributed in Tanggula mountain, Qilian mountain and Kunlun mountain areas, Landsat8 remote sensing data of 2018 years in the Qilian mountain area is selected, 30 m national DEM data is selected as auxiliary data, glaciers and snow are extracted, and the method is specifically implemented by the following steps:
step 1: remote sensing data preprocessing
(1) Atmospheric corrections were made to the Landsat8 full and multi-spectral data. And the atmospheric correction of the remote sensing data is completed by utilizing two functional modules, namely ENVI5.3 radiometric calibration and FLAASH atmospheric correction.
(2) And (5) image registration. And taking the panchromatic waveband as a reference waveband, and relatively registering the multispectral waveband and the panchromatic waveband. And uniform homonymous points are collected in the range of the research area, and the registration correction of the image is realized through affine transformation.
(3) And (5) image fusion processing. And carrying out fusion processing on the panchromatic image and the multispectral image to obtain remote sensing data with the resolution of 15 meters. Common fusion methods include BROVEY transform fusion, IHS transform fusion, principal component transform fusion, wavelet transform fusion, and the like. The image fusion adopts a BROVEY transformation fusion method.
IhighRepresenting the panchromatic band, IlowRepresents a multi-spectral band, ∑ IlowRepresenting the sum of all bands of the multi-spectrum.
And calculating a first step output remote sensing fused image, which is shown in figure 3.
Step 2: glacier and snow are extracted by normalizing snow index threshold
Green represents a Green band ranging from 0.5 μm to 0.6 μm; SWIR1 represents the short-wave infrared band in the range of 1.55 μm to 1.75 μm. The NDSI characteristics of the output fusion image in the second step are calculated, and the calculation result is shown in figure 4, so that the snow is remarkably represented in the NDSI. The output NDSI threshold extraction result is calculated by means of threshold partitioning, as shown in fig. 5.
And step 3: snow extraction by using decision tree model based on fusion of multidimensional characteristics
(1) DEM characteristics and spectral characteristics are acquired.
DEM characteristic acquisition: acquiring DEM auxiliary data of a research area, considering the follow-up need of fusing DEM characteristics with other characteristics, resampling the DEM data of the research area to 15 meters, enabling the image resolution to be consistent with that of the fused Landsat8, and finally outputting the resampled DEM characteristics.
Obtaining spectral characteristics: landsat8 data blue, green, red, near red, short wave infrared 1, short wave infrared 2 were used as part of the spectral signature. In addition, standard deviations of blue, green, red, near red, short wave infrared 1, short wave infrared 2 bands are calculated as another input of spectral characteristics. Calculation method of standard deviation: the pixel to be calculated is used as the center, the window size is specified, the average value is subtracted from the gray value of all pixels in the window, the square sum of the average value is divided by the total number of the pixels, and the root number is formed, so that the standard deviation of the pixel can be obtained.
(2) Mask processing spectral features and DEM features
And (3) masking the spectral features and the DEM features by utilizing the crude extraction result range of glaciers and snow in the step (2), and outputting the spectral features and the DEM features in the range of the suspected glaciers and snow.
(3) Decision tree model information extraction
And integrating the elevation and spectral characteristics by using a wave band combination mode. On the basis of the fusion characteristics, glacier and snow samples and non-glacier and snow samples are respectively collected, a decision tree model is utilized to carry out fine recognition on the glacier and the snow, and the fine recognition result is shown in figure 6.
The decision tree adopts a greedy algorithm from top to bottom recursion, the attribute values of internal nodes of the decision tree are compared, and downward branches from the nodes are judged according to different attribute values, so that leaf nodes of the decision tree are obtained. In the process of establishing the decision tree, how to establish the optimal split node is very critical. At present, the attribute measurement methods include various methods such as a nity index, an information gain ratio, an information gain, G statistics, distance measurement, and the like. The CART algorithm utilized by the decision tree is used for taking the Nissy index as an index of attribute measurement.
Let the classification attribute C of the data set S have m different discrete attributes C1，c2，...，cmThe corresponding equation for the Nyquist index is:
wherein, PiIs a class ciThe frequency of occurrence. If the A attribute is utilized.
Partitioning S data set into S1And S2The divided formula of the Nyquist index is as follows:
Gini(S)＝(S1/S)*Gini(S1)+(S2/S)*Gini(S2)
and searching the divided optimal node, namely calculating the optimal dividing method of each attribute, and comparing the NiBu indexes of all the attributes, wherein the minimum NiBu index is regarded as the optimal divided node.
Since the structure of the decision tree is easy to generate the over-fitting phenomenon, the decision tree needs to be pruned, and the common pruning method comprises pre-pruning and post-pruning. The CART decision tree adopts a post-pruning method, namely, the decision tree is generated firstly, then the tree is pruned, then cross validation is used for testing various pruning effects, and a pruning strategy with the best generalization capability is selected. The idea of pruning is as follows: first, pruning is performed to form a sub-tree sequence.
In the pruning process, the loss function of the subtree is calculated as follows:
Cα(T)＝C(T)+α|T|
where T denotes any one of the subtrees, and c (T) denotes the prediction error of the training data. | T | represents the number of leaf nodes, α ≧ 0, Cα(T) represents the loss of the subtree T when the parameter is α. Alpha is the degree of fit that balances the training data with the complexity of the model.
For a fixed α, C must be presentα(T) the smallest subtree, we denote it as Tα，TαAt Cα(T) is optimum in the sense of minimum. Then, the sub-tree sequences obtained by pruning are subjected to cross validation, and the optimal tree is selected from the sub-tree sequences, so that the decision tree can be pruned.
And 4, step 4: evaluation of accuracy
And (3) carrying out precision evaluation on the glacier and snow extraction result extracted by the decision tree model in the previous step by taking the glacier and snow sketched manually as truth data, wherein the precision evaluation adopts a correct rate index, and the formula is as follows:
accuracy ratio (correct prediction area/total prediction area)
Wherein the correctly predicted area represents the correctly extracted glacier and snow area and the predicted total area represents the extracted glacier and snow area.
The extraction precision of the glacier and the accumulated snow in 2018 years by the method provided by the invention is 98.7 percent respectively.
In addition, in order to verify the effect of the multi-dimensional features in this embodiment, the same samples are used, and only when spectral features are input, information extraction is performed by using a decision tree model, and a classification result is output. As shown in fig. 7 and 8, it is apparent from the visual effect analysis that the information extraction result by fusing the DEM and the spectral feature is more complete and the effect is better. From the analysis of the precision evaluation index, the accuracy of the classification chart output by combining the DEM and the spectral features is 98.7%, and the accuracy of the classification chart after inputting the spectral features is 92.6%.
In conclusion, by means of the technical scheme, the land feature is analyzed from the multi-dimensional features, the DEM and the spectrum information are effectively fused, and the condition that the recognition accuracy of glaciers and snow is low due to single feature dimension is avoided; meanwhile, by comprehensively utilizing the snow cover index method and the decision tree algorithm, on the basis of the extraction of the snow cover index method, the advantages of data type and conventional type attributes can be simultaneously processed by means of the decision tree algorithm, the glacier and snow cover extraction rule is excavated, and the glacier and snow cover extraction precision is improved.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A glacier and snow automatic extraction method based on remote sensing images is characterized by comprising the following steps:
acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image, and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
roughly extracting glaciers and snow on the preprocessed satellite remote sensing image by using a normalized snow index threshold method, and determining suspected ranges of the glaciers and the snow;
acquiring DEM auxiliary data in a region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
carrying out mask processing on DEM characteristic data and spectral characteristic data in the area to be extracted by utilizing glacier and snow suspected range to obtain multi-dimensional characteristics in the suspected range;
collecting samples of glaciers and snow in a suspected range and samples of non-glaciers and snow in a suspected range, selecting multi-dimensional characteristics, and finely extracting the glaciers and the snow by using a pre-configured decision tree model to obtain a glacier and snow range.
2. The method for automatically extracting glacier and snow based on remote sensing images according to claim 1, wherein preprocessing the satellite remote sensing images comprises:
and carrying out atmospheric correction processing, image registration processing and image fusion processing on the satellite remote sensing image.
3. The method for automatically extracting glacier and snow based on remote sensing images as claimed in claim 1, wherein the normalized snow index NDSI is used when rough glacier and snow extraction is performed on the preprocessed satellite remote sensing images by using a normalized snow index threshold method to determine a suspected range,
wherein, Green represents a Green wave band in the range of 0.5 to 0.6 mu m; the SWIR1 represents a short wave infrared band in the range of 1.55 to 1.75 μm.
4. The method for automatically extracting glacier and snow based on remote sensing images as claimed in claim 1, wherein the algorithm of the decision tree model is CART algorithm, and a commonly used Niey index is used as an index of attribute measurement.
5. The method for automatically extracting glaciers and snow based on remote sensing images as claimed in any one of claims 1-4, further comprising:
and taking the glacier and snow cover range data which is drawn manually as truth value data, and analyzing the accuracy of the glacier and snow cover range which is obtained by fine recognition, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow cover by the predicted total area of the glacier and snow cover.
6. The utility model provides a glacier and snow automatic extraction system based on remote sensing image which characterized in that includes:
the data acquisition module is used for acquiring a satellite remote sensing image of a region to be extracted, preprocessing the satellite remote sensing image and removing image interference factors in the satellite remote sensing image; the satellite remote sensing image comprises blue, green, red, near-red and short-wave infrared bands;
the rough extraction module is used for carrying out glacier and snow rough extraction on the preprocessed satellite remote sensing image by utilizing a normalized snow index threshold method to determine a glacier and snow suspected range;
the characteristic calculation module is used for acquiring DEM auxiliary data in the region to be extracted, and calculating multi-dimensional characteristics including spectral values, standard deviations and DEM characteristics by combining remote sensing images;
the characteristic masking module is used for masking DEM characteristic data and spectral characteristic data in the area to be extracted by utilizing the suspected ranges of glaciers and snow to obtain multi-dimensional characteristics in the suspected ranges;
and the range identification module is used for acquiring samples of glaciers and snow in a suspected range and samples of non-glaciers and snow, selecting multi-dimensional characteristics, and performing fine extraction on the glaciers and the snow by using a pre-configured decision tree model to obtain a glacier and snow range.
7. The automatic glacier and snow extraction system based on remote sensing images of claim 6, wherein the data acquisition module is used for preprocessing the satellite remote sensing images and comprises the following steps:
and carrying out atmospheric correction processing, image registration processing and image fusion processing on the satellite remote sensing image.
8. The automatic glacier and snow extraction system based on remote sensing images according to claim 6, wherein the rough extraction module is used for conducting rough glacier and snow extraction on the preprocessed satellite remote sensing images by utilizing a normalized snow index threshold method, and normalizing the snow index NDSI when a suspected range is determined, wherein Green represents a Green wave band in a range of 0.5 mu m to 0.6 mu m; the SWIR1 represents a short wave infrared band in the range of 1.55 to 1.75 μm.
9. The remote-sensing-image-based automatic glacier and snow extraction system according to claim 6, wherein the algorithm of the decision tree model is a CART algorithm, and a commonly used Niey index is used as an index of attribute measurement.
10. The remote-sensing-image-based automatic extraction system for glaciers and snow cover according to any one of claims 6-9, further comprising:
and the accuracy evaluation module is used for analyzing the accuracy of the glacier and snow range obtained by fine recognition by taking the manually drawn glacier and snow range data as truth value data, wherein the accuracy is obtained by dividing the correctly predicted area of the glacier and snow by the predicted total area of the glacier and snow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854709.8A CN113822141A (en) | 2021-07-28 | 2021-07-28 | Automatic glacier and snow extraction method and system based on remote sensing image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854709.8A CN113822141A (en) | 2021-07-28 | 2021-07-28 | Automatic glacier and snow extraction method and system based on remote sensing image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113822141A true CN113822141A (en) | 2021-12-21 |
Family
ID=78924189
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110854709.8A Pending CN113822141A (en) | 2021-07-28 | 2021-07-28 | Automatic glacier and snow extraction method and system based on remote sensing image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113822141A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114994675A (en) * | 2022-04-19 | 2022-09-02 | 西南交通大学 | Glacier classification method and system based on normalized intensity dispersion index |
| CN115469370A (en) * | 2022-11-14 | 2022-12-13 | 航天宏图信息技术股份有限公司 | Glacier boundary extraction method and device for eliminating ice lake interference |
| CN116129280A (en) * | 2023-04-17 | 2023-05-16 | 北京数慧时空信息技术有限公司 | Method for detecting snow in remote sensing image |
-
2021
- 2021-07-28 CN CN202110854709.8A patent/CN113822141A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114994675A (en) * | 2022-04-19 | 2022-09-02 | 西南交通大学 | Glacier classification method and system based on normalized intensity dispersion index |
| CN114994675B (en) * | 2022-04-19 | 2023-10-03 | 西南交通大学 | Glacier classification method and system based on normalized intensity dispersion index |
| CN115469370A (en) * | 2022-11-14 | 2022-12-13 | 航天宏图信息技术股份有限公司 | Glacier boundary extraction method and device for eliminating ice lake interference |
| CN116129280A (en) * | 2023-04-17 | 2023-05-16 | 北京数慧时空信息技术有限公司 | Method for detecting snow in remote sensing image |
| CN116129280B (en) * | 2023-04-17 | 2023-07-14 | 北京数慧时空信息技术有限公司 | Method for detecting snow in remote sensing image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109101894B (en) | A kind of remote sensing image clouds shadow detection method that ground surface type data are supported | |
| Wei et al. | Cloud detection for Landsat imagery by combining the random forest and superpixels extracted via energy-driven sampling segmentation approaches | |
| CN113822141A (en) | Automatic glacier and snow extraction method and system based on remote sensing image | |
| US8913826B2 (en) | Advanced cloud cover assessment for panchromatic images | |
| Fang | Canopy clumping index (CI): A review of methods, characteristics, and applications | |
| US7184890B2 (en) | Cloud shadow detection: VNIR-SWIR | |
| CN111160199B (en) | Highway disaster information detection method based on high-resolution remote sensing image | |
| CN112131946B (en) | Automatic extraction method for vegetation and water information of optical remote sensing image | |
| CN112183209A (en) | Regional crop classification method and system based on multi-dimensional feature fusion | |
| CN111582194A (en) | Multi-temporal high-resolution remote sensing image building extraction method based on multi-feature LSTM network | |
| CN110852381A (en) | Forest fire burned area extraction method and system | |
| CN111652092A (en) | Method for monitoring forest coverage change based on Sentinel-2A data | |
| CN112418133A (en) | Straw burning monitoring method based on multi-source remote sensing image | |
| Zhai | Inversion of organic matter content in wetland soil based on Landsat 8 remote sensing image | |
| CN114778483A (en) | Method for correcting terrain shadow of remote sensing image near-infrared wave band for monitoring mountainous region | |
| CN115170979A (en) | Mining area fine land classification method based on multi-source data fusion | |
| CN112669363A (en) | Urban green land three-dimensional green volume calculation method | |
| Li et al. | Multi-information collaborative cloud identification algorithm in Gaofen-5 Directional Polarimetric Camera imagery | |
| Xu et al. | Vegetation information extraction in karst area based on UAV remote sensing in visible light band | |
| CN117075138A (en) | Remote sensing measurement and calculation method, system and medium for canopy height of 30-meter forest in area | |
| CN116206211A (en) | Feature type identification method for integrating time sequence remote sensing information and KL-divergence | |
| Huang et al. | A 3D approach to reconstruct continuous optical images using lidar and MODIS | |
| CN116310864A (en) | Automatic identification method, system, electronic equipment and medium for crop lodging | |
| CN112577954B (en) | Urban green land biomass estimation method | |
| CN113902759B (en) | Space-spectrum information combined satellite-borne hyperspectral image segmentation and clustering method |
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 |