Cloud recognition method, device, electronic device and readable storage medium for remote sensing images

Abstract

The present application discloses a cloud recognition method, device, electronic device and readable storage medium for remote sensing images. The method includes updating the Fmask cloud recognition algorithm applicable to multispectral sensors in advance according to the data preprocessing method and band information of the hyperspectral sensor to obtain a hyperspectral cloud recognition method; in response to the data input mode selection instruction, obtaining the target data to be identified from the hyperspectral remote sensing data. Based on the hyperspectral cloud recognition method, the target data to be identified is image preprocessed to obtain the apparent reflectance information; based on the apparent reflectance information, the hyperspectral cloud recognition method is called to perform cloud recognition and cloud shadow recognition respectively to obtain cloud recognition results and cloud shadow recognition results; based on the cloud recognition results and cloud shadow recognition results, the cloud information of the hyperspectral remote sensing data is determined, thereby effectively improving the cloud recognition accuracy of the hyperspectral images of domestic hyperspectral satellites.

Description

Technical Field

[0001] The present application relates to the field of remote sensing technology, and in particular to a cloud recognition method, device, electronic device and readable storage medium for remote sensing images. Background Art

[0002] Clouds, as an important part of the atmosphere, cover more than 50% of the global surface. They have a strong absorption and scattering effect on solar and infrared radiation, which causes the reflection of objects in remote sensing images to attenuate and cover the effect, making the ground observation based on visible light and infrared bands almost all interfered. In general, the brightening effect of cloud areas and the darkening effect of cloud shadows will greatly affect the quality of remote sensing information acquisition.

[0003] In remote sensing applications, clouds can cause large errors in the spectral information of objects in satellite images, especially in images with thick clouds. The underlying surface will be largely blocked by clouds, and important spectral information will be lost. In order to eliminate the impact of clouds on objects to the greatest extent, the first task is to perform cloud recognition in remote sensing images. The essence of cloud recognition is to find the difference between cloud pixels and other objects in spectral characteristics or texture characteristics, so that they can be separated. For clouds and cloud shadows, the high reflectivity and low temperature characteristics of clouds and the low reflectivity characteristics of cloud shadows are usually used for recognition. The current cloud recognition methods can be roughly divided into threshold methods, statistical methods and machine learning methods. Among them, the threshold method uses the spectral characteristics of clouds to construct threshold criteria for recognition. It is relatively simple and widely used, such as the ISCCP (International Satellite Cloud Climatology Project) algorithm, the CLAVR (The NOAA Cloud Advanced Very High Resolution Radiometer) algorithm, and the APOLLO (AVHRR Processing scheme Over Clouds, Land and Ocean) algorithm. Statistical analysis methods apply statistical principles to cloud recognition, such as the Bayesian method, the discriminant analysis method and the SVM method. Machine learning methods use artificial neural networks as classifiers to train sample data and perform cloud recognition. Although the neural network itself does not have physical meaning, it can extract spectral and spatial features through its powerful self-learning and data analysis capabilities, thereby identifying cloud areas on remote sensing images, such as the Otsu cloud detection method and the cloud detection method based on deep residual full convolutional networks. However, the above cloud recognition algorithms are all developed for multispectral sensors carried by satellites such as Landsat satellites and MODIS. Since hyperspectral sensors have not been around for a long time, there are not as many cloud recognition algorithms for hyperspectral sensors as for multispectral sensors. Especially for domestic hyperspectral satellites, the above cloud recognition algorithms may not be applicable and cannot achieve good cloud recognition results.

[0004] In view of this, how to improve the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites is a technical problem that technicians in the field need to solve. Summary of the invention

[0005] The present application provides a cloud recognition method, device, electronic device and readable storage medium for remote sensing images, which can effectively improve the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites.

[0006] To solve the above technical problems, the embodiments of the present invention provide the following technical solutions:

[0007] An embodiment of the present invention provides a cloud recognition method for a remote sensing image, comprising:

[0008] According to the data preprocessing method and band information of the hyperspectral sensor, the Fmask cloud recognition algorithm suitable for multispectral sensors is updated in advance to obtain a hyperspectral cloud recognition method;

[0009] In response to the data input mode selection instruction, the target data to be identified is obtained from the hyperspectral remote sensing data;

[0010] Based on the hyperspectral cloud recognition method, image preprocessing is performed on the target data to be identified to obtain apparent reflectance information;

[0011] Based on the apparent reflectance information, calling the hyperspectral cloud recognition method to perform cloud recognition and cloud shadow recognition respectively, to obtain a cloud recognition result and a cloud shadow recognition result;

[0012] Determining cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result;

[0013] Among them, the data input method is determined according to the band information of the hyperspectral sensor and the multispectral sensor, and the target data to be identified is the ground object image information corresponding to the matching bands of the hyperspectral sensor and the multispectral sensor.

[0014] Optionally, obtaining target data to be identified from hyperspectral remote sensing data includes:

[0015] When it is detected that the data input mode is the central wavelength input mode, the median of multiple single-band wavelength ranges of the multispectral sensor is read, corresponding to the image data at the corresponding band of the hyperspectral remote sensing data;

[0016] The solar azimuth angle, solar zenith angle information, image row and column numbers and annual accumulated days are read from the hyperspectral remote sensing data as target data to be identified.

[0017] Optionally, obtaining target data to be identified from hyperspectral remote sensing data includes:

[0018] When it is detected that the data input mode is a weighted fitting input mode, the radiation brightness value after radiation calibration of each band is determined according to the hyperspectral remote sensing data;

[0019] For each band, the approximate multispectral data of the current band is obtained by fitting the spectral response curve of the multispectral sensor in the current band and the radiant brightness value of the hyperspectral remote sensor in the current band;

[0020] The approximate multispectral data of each band is used as the target data to be identified.

[0021] Optionally, based on the hyperspectral cloud recognition method, image preprocessing is performed on the target data to be identified to obtain apparent reflectance information, including:

[0022] Acquire a gain parameter and a bias parameter from the target data to be identified, and calculate the radiation brightness value of each pixel according to the gain parameter and the bias parameter;

[0023] The apparent reflectance information is determined based on the radiance value of each pixel, the distance between the sun and the earth, the average solar irradiance at the top of the atmosphere, and the solar zenith angle.

[0024] Optionally, before determining the apparent reflectance information according to the radiance value of each pixel, the distance between the sun and the earth, the average solar irradiance at the top of the atmosphere, and the solar zenith angle, the method further includes:

[0025] Acquire the central wavelength and half-wave width from the hyperspectral remote sensing data, and obtain the spectral curve of each band of the hyperspectral satellite according to the central wavelength and the half-wave width Gaussian fitting;

[0026] According to the spectral curve of each band and the solar spectrum curve, the average solar irradiance at the top of the atmosphere in each band is calculated.

[0027] Optionally, determining the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result includes:

[0028] Performing a water test on the target data to be identified to determine a non-water area in the target data to be identified;

[0029] For the non-water area, based on the apparent reflectivity information, an ATCOR algorithm is used to perform cloud shadow recognition to obtain a corrected cloud shadow recognition result;

[0030] Perform cloud matching on the cloud recognition result, the cloud shadow result and the corrected cloud shadow recognition result, respectively, to obtain a shadow area and a corrected shadow area;

[0031] If the difference between the shadow area and the corrected shadow area is greater than a preset area threshold, determining the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result;

[0032] If the difference between the shadow area and the corrected shadow area is less than or equal to a preset area threshold, cloud information of the hyperspectral remote sensing data is determined according to the cloud recognition result and the corrected cloud shadow recognition result.

[0033] Optionally, after performing a water test on the target data to be identified, the method further includes:

[0034] If the target data to be identified does not include a non-water area, the cloud information of the hyperspectral remote sensing data is determined according to the cloud identification result and the cloud shadow identification result.

[0035] Another aspect of the present invention provides a cloud recognition device for remote sensing images, including:

[0036] The method processing module is used to update the Fmask cloud recognition algorithm applicable to the multispectral sensor according to the data preprocessing method and band information of the hyperspectral sensor in advance, so as to obtain the hyperspectral cloud recognition method;

[0037] A data reading module, for responding to a data input mode selection instruction, and obtaining target data to be identified from the hyperspectral remote sensing data; the data input mode is determined according to the band information of the hyperspectral sensor and the multispectral sensor, and the target data to be identified is the ground object image information corresponding to the matching band of the hyperspectral sensor and the multispectral sensor;

[0038] A data preprocessing module, used to perform image preprocessing on the target data to be identified based on the hyperspectral cloud recognition method to obtain apparent reflectance information;

[0039] an identification module, configured to call the hyperspectral cloud identification method to perform cloud identification and cloud shadow identification respectively based on the apparent reflectance information, and obtain a cloud identification result and a cloud shadow identification result;

[0040] A result determination module determines the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result.

[0041] An embodiment of the present invention further provides an electronic device, comprising a processor, wherein the processor is configured to implement the steps of the cloud recognition method for remote sensing images as described in any of the preceding items when executing a computer program stored in a memory.

[0042] Finally, an embodiment of the present invention further provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the cloud recognition method for remote sensing images as described in any of the preceding items are implemented.

[0043] The advantage of the technical solution provided in the present application is that the data input method and data processing method are improved through the Fmask cloud recognition algorithm, thereby applying the Fmask algorithm based on cloud recognition of multispectral remote sensing satellite images to domestic hyperspectral satellites. Since the data input and preprocessing parts are set for the hyperspectral sensors of domestic hyperspectral satellites, accurate surface reflectance of ground objects can be obtained, making it suitable for the band setting of the hyperspectral sensor while also obtaining better recognition results, thereby effectively improving the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites.

[0044] In addition, the embodiments of the present invention also provide corresponding implementation devices, electronic devices and readable storage media for the cloud recognition method of remote sensing images, which further makes the method more practical, and the devices, electronic devices and readable storage media have corresponding advantages.

[0045] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic diagram of a flow chart of a cloud recognition method for remote sensing images provided by an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a flow chart of a cloud recognition method for remote sensing images provided by an embodiment of the present invention;

[0049] Figure 3 The unweighted recognition result of the Gaofen-5 image on July 7, 2019 in an exemplary application scenario provided by an embodiment of the present invention;

[0050] Figure 4 The unweighted recognition result of the Gaofen-5 image on November 1, 2018 in an exemplary application scenario provided by an embodiment of the present invention;

[0051] Figure 5 The unweighted recognition result of the Gaofen-5 image on March 11, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0052] Figure 6 The unweighted recognition result of Resource No. 1 02D image on June 1, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0053] Figure 7 The unweighted recognition result of Resource No. 1 02D image on September 7, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0054] Figure 8 The weighted recognition result of the Gaofen-5 image on July 7, 2019 in an exemplary application scenario provided by an embodiment of the present invention;

[0055] Fig. 9 The weighted recognition result of the Gaofen-5 image on November 1, 2018 in an exemplary application scenario provided by an embodiment of the present invention;

[0056] Fig.10 The weighted recognition result of the Gaofen-5 image on March 11, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0057] Fig.11 The weighted recognition result of Resource No. 1 02D image on June 1, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0058] Fig.12 The weighted recognition result of Resource No. 1 02D image on September 7, 2020 in an exemplary application scenario provided by an embodiment of the present invention;

[0059] Fig.13 A structural diagram of a specific implementation of a cloud recognition device for remote sensing images provided in an embodiment of the present invention;

[0060] Fig.14 A structural diagram of a specific implementation of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to enable those skilled in the art to better understand the scheme of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0062] The terms "first", "second", "third", "fourth", etc. in the specification and claims of this application and the above drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may include steps or units that are not listed.

[0063] The inventor of the present application has found through research that the Famsk (Function of mask) cloud recognition algorithm was disclosed in an article entitled Object-based cloud and cloud shadow detection in Landsat imagery published by Zhe Zhu et al. in the journal Remote Sensing of Environment in 2012. The algorithm was originally based on the TM and ETM+ sensors carried by the Landsat series of satellites, and then expanded to Landsat8 and Sentinel satellites, all of which achieved good cloud recognition results. The Fmask cloud recognition algorithm is a cloud recognition algorithm for multispectral satellite remote sensing images, which can identify clouds, cloud shadows, water bodies and snow pixels in the image with high accuracy. The algorithm uses the top of atmosphere (TOA, Top Of Atmosphere) reflectivity and brightness temperature (Brightness Temperature, BT) of the multispectral sensor as input, and constructs a criterion based on the spectral characteristics of clouds and other objects for recognition. In addition, the viewing angle and illumination angle of the satellite sensor are used to predict the possible cloud shadow position, making the recognition of clouds and cloud shadows more accurate. Compared with the classic cloud recognition algorithm ACCA (Automated Cloud-Cover Assessment), the cloud recognition accuracy of the Famsk algorithm has great advantages. For example, the overall accuracy of the Famsk algorithm is 96.41%, while the overall accuracy of the ACCA algorithm is 84.8%; the producer accuracy of the Famsk algorithm is 92.1%, while the producer accuracy of the ACCA algorithm is 72.1%; the user accuracy of the Famsk algorithm is 89.4%, while the user accuracy of the ACCA algorithm is 91.8%. The Fmask cloud recognition algorithm is divided into two parts. The first part is the determination of potential cloud areas, which determines possible cloud pixels through a series of threshold criteria; the second part calculates the probability of clouds on land and clouds on water according to the properties of the clouds. If the probability is greater than a certain value, it is determined to be a cloud.

[0064] Among them, (1) potential cloud area detection, namely PCPs test, includes basic test, whiteness test, HOT test and band 5 and 6 test. The basic test is shown in formula (1). BandSWIR2 is the apparent reflectance value of the seventh band of Landsat 8 satellite, and the wavelength range of the seventh band is 2.11μm-2.29μm; BT is the pixel temperature calculated from the thermal infrared band, in degrees Celsius; where NDSI (normalized difference snow cover index) and NDVI (normalized difference vegetation index) are defined by formulas (2) and (3). Among them, Band3, Band4, Band5 and Band6 are the apparent reflectance values ​​of the green, red, near infrared and SWIR1 (1.57-1.65μm) bands of Landsat 8 satellite, respectively, and the wavelength range of SWIR1 band is 1.57μm-1.65μm.

[0065] BandSWIR2>0.03 and BT<27 and NDSI<0.8 and NDVI<0.8 (1)

[0066]

[0067]

[0068] The whiteness test defines a new whiteness index, as shown in formula (4), and then the whiteness test is performed using the index, as shown in formula (5).

[0069]

[0070]

[0071] The HOT test definition is shown in formula (6).

[0072] HOT=(Band2-0.5)×(Band4-0.08)>0 (6)

[0073] The test definition for bands 5 and 6 is shown in formula (7).

[0074] Band5 / Band6>0.75 (7)

[0075] Pixels that pass all four tests are marked as potential cloud pixels, and pixels that fail at least one test are marked as absolutely clear sky pixels.

[0076] (2) Calculate the cloud probability over the water body. The temperature probability wTemperature_Prob is calculated as shown in formula (8), and the brightness probability Brightness_Prob is calculated as shown in formula (9).

[0077] wTemperature_Prob=(82.5%BT 晴空水体 -BT) / 4 (8)

[0078] Brightness_Prob=min(Band6,0.11) / 0.11 (9)

[0079] After obtaining the temperature probability and brightness probability of clouds above the water body, the cloud probability wCloud_Prob above the water body can be calculated according to formula (10). In the Fmask algorithm, if the value is greater than 0.5, it can be determined as a cloud pixel.

[0080] wCloud_Prob=wTemperture_Prob×Brightness_Prob (10)

[0081] (3) Calculate the cloud probability over land. The temperature probability is calculated as shown in formula (11-13), where BT 晴空陆地 is the temperature of land under clear sky conditions, BT is the temperature of potential cloud pixels over land, T low 、T high The clear sky land temperatures are 17.5% and 82.5% respectively, and lTemperature_Prob is the temperature probability over land. The variability probability Variability_Prob is calculated as shown in formula (14).

[0082] T high =82.5%×BT 晴空陆地 (11)

[0083] T low =17.5%×BT 晴空陆地 (12)

[0084] lTemperature_Prob=(T high +4-BT) / (T high +4-(T low -4)) (13)

[0085] Variability_Prob=1-max(abs(NDVI,NDSI,Whiteness)) (14)

[0086] After obtaining the temperature probability and variability probability, the probability of cloud over land lCloud_Prob can be calculated according to formula (15).

[0087] lCloud_Prob=lTemperature_Prob×Variability_Prob (15)

[0088] If the cloud probability is greater than Land_threshold (Formula (16)), it is determined to be a cloud pixel.

[0089] Land_threshold=82.5%lCloud_Prob 晴空 +0.2 (16)

[0090] Although the Famsk cloud recognition algorithm has a good recognition effect on clouds in multispectral satellite remote sensing images, the algorithm cannot preprocess domestic hyperspectral sensor data because the sensors carried by domestic hyperspectral satellites are completely different from the header file format and other related parameters of the multispectral sensors targeted by the Famsk cloud recognition algorithm. Furthermore, in the Famsk cloud recognition algorithm, the thermal infrared band is very important for the resolution of ground objects, but the hyperspectral sensors carried by domestic hyperspectral satellites such as Gaofen-5 and Ziyuan-1 02D do not have thermal infrared bands, resulting in the inability of domestic high-resolution satellites to use the Famsk cloud recognition algorithm. In addition, the Famsk cloud recognition algorithm cannot take advantage of the high spectral resolution and multiple bands of hyperspectral sensors.

[0091] Since the Famsk cloud recognition algorithm has a high-precision advantage in cloud recognition, this application improves the Famsk cloud recognition algorithm to make it suitable for hyperspectral sensors of domestic hyperspectral satellites, thereby improving the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites.

[0092] After introducing the technical solutions of the embodiments of the present invention, various non-limiting implementation methods of the present application are described in detail below.

[0093] See first Figure 1 , Figure 1 A schematic diagram of a flow chart of a cloud recognition method for a remote sensing image provided by an embodiment of the present invention, the embodiment of the present invention may include the following contents:

[0094] S101: According to the data preprocessing method and band information of the hyperspectral sensor, the Fmask cloud recognition algorithm applicable to the multispectral sensor is updated in advance to obtain a hyperspectral cloud recognition method.

[0095] In this embodiment, the hyperspectral cloud recognition method is a method for recognizing hyperspectral images obtained by improving the Fmask cloud recognition algorithm. The Fmask cloud recognition algorithm is a recognition algorithm suitable for multispectral images. Multispectral images are collected by multispectral sensors. Multispectral sensors are equipped with multispectral remote sensing satellites such as Landsat8 satellites. Hyperspectral images refer to images with a spectral resolution of 10 -2 Spectral images in the range of λ.

[0096] In this embodiment, since the read-in header file format and band of the multispectral sensor and the hyperspectral sensor are different, in order to make the cloud recognition algorithm applicable to the multispectral sensor applicable to the hyperspectral sensor, applicable to the domestic hyperspectral satellites such as Gaofen-5 and Ziyuan-1 02D satellite, this step improves the data input part, cloud and cloud shadow recognition part of the Fmask cloud recognition algorithm, and obtains a better cloud recognition result. For the data preprocessing part: This step modifies the code of the Fmask cloud recognition algorithm read-in header file part, reads in the basic information of the hyperspectral sensor corresponding to the target band of the multispectral sensor, and the basic information includes but is not limited to ground object information, solar azimuth and solar zenith angle. Taking the Landsat8 satellite as an example, this embodiment reads in the DN value (Digital Number, remote sensing image pixel brightness value) image file of the Gaofen-5 satellite or Ziyuan-1 02D satellite corresponding to the 2, 3, 4, 5, 6, 7, 9 bands of the OLI sensor in the Landsat8 satellite, as well as basic information such as the solar azimuth and solar zenith angle.

[0097] For the cloud recognition part: Since the band range of the visible short-wave infrared hyperspectral camera of hyperspectral sensors such as Gaofen-5 and Ziyuan-1 02D satellites is 400nm to 2500nm, and there is no thermal infrared band, the source code of the Fmask cloud recognition algorithm in the cloud recognition part is modified to remove the part using the thermal infrared band, including PCPs test, cloud probability calculation and all other temperature judgments and threshold calculations, that is, there is no thermal infrared band in the hyperspectral cloud recognition method, so that the Fmask cloud recognition algorithm can adapt to the band settings of hyperspectral satellites such as Gaofen-5 satellites to the greatest extent, and is suitable for the hyperspectral sensors of Gaofen-5 and Ziyuan-1 02D satellites. For the cloud shadow recognition part: Similarly, since there is no thermal infrared band, the cloud shadow detection code of the Fmask algorithm is modified to remove all parts related to temperature. This step modifies the data input and preprocessing parts according to the characteristics of the hyperspectral sensors of domestic hyperspectral satellites, and can obtain accurate surface reflectance of ground objects. The recognition of clouds and cloud shadows has been adjusted and improved according to the sensor to make it suitable for the band setting of the hyperspectral sensor and to obtain better recognition results.

[0098] S102: Responding to the data input mode selection instruction, obtaining target data to be identified from the hyperspectral remote sensing data.

[0099] The data input method of this embodiment is determined according to the band information of the hyperspectral sensor and the multispectral sensor, that is, since the data read is hyperspectral remote sensing data, and the Fmask cloud recognition algorithm is applicable to multispectral remote sensing data, in order to improve data processing efficiency and recognition effect, it is necessary to read some data matching the multispectral remote sensing data type from the hyperspectral remote sensing data, that is, to read data similar to the multispectral remote sensing data used by the original Fmask algorithm from the hyperspectral remote sensing data, and these data are determined based on the band information of the hyperspectral sensor and the multispectral sensor. The target data to be identified is the image information of the ground object corresponding to the matching band of the hyperspectral sensor and the multispectral sensor. The ground object image information is used to record the ground object feature data, such as the ground object gray value image file, and the remote sensing image pixel brightness value can be obtained from the remote sensing data as the data representing the ground object feature.

[0100] S103: Based on the hyperspectral cloud recognition method, image preprocessing is performed on the target data to be recognized to obtain apparent reflectance information.

[0101] In this step, the Fmask cloud recognition algorithm with modified read-in header file format is used to preprocess the target data to obtain the surface reflectivity data of the object.

[0102] S104: Based on the apparent reflectance information, a hyperspectral cloud recognition method is called to perform cloud recognition and cloud shadow recognition respectively to obtain a cloud recognition result and a cloud shadow recognition result.

[0103] In this step, the Fmask cloud recognition algorithm with the thermal infrared band removed is used to perform cloud recognition processing to obtain cloud recognition results, and the Fmask cloud recognition algorithm with the temperature part removed is used to perform cloud shadow processing to obtain cloud shadow recognition results.

[0104] S105: Determine cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result.

[0105] The cloud recognition results from the previous step are matched with the cloud shadow recognition results, and the cloud area in the hyperspectral remote sensing data is determined based on the matching results.

[0106] In the technical solution provided in the embodiment of the present invention, the data input method and the data processing method are improved through the Fmask cloud recognition algorithm, so that the Fmask algorithm based on cloud recognition of multispectral remote sensing satellite images is applied to the domestic hyperspectral satellite. Since the data input and preprocessing parts are set for the hyperspectral sensor of the domestic hyperspectral satellite, the accurate apparent reflectance of the ground object can be obtained, making it suitable for the band setting of the hyperspectral sensor while also obtaining better recognition results, thereby effectively improving the cloud recognition accuracy of the hyperspectral image of the domestic hyperspectral satellite.

[0107] It should be noted that there is no strict order of execution between the steps in this application. As long as they comply with the logical order, these steps can be executed simultaneously or in a certain preset order. Figure 1 This is just a schematic and does not mean that this is the only execution order.

[0108] In the above embodiment, there is no limitation on how to execute step S102. In this embodiment, two implementation methods of obtaining target data to be identified from hyperspectral remote sensing data are provided. Accordingly, the data input method includes two methods, which may include the following steps:

[0109] As an optional implementation, when it is detected that the data input method is the central wavelength input method, the median of multiple single-band wavelength ranges of the multispectral sensor is read, corresponding to the image data at the corresponding band of the hyperspectral remote sensing data; the solar azimuth angle, solar zenith angle information, number of image rows and columns, and annual accumulated days are read from the hyperspectral remote sensing data as the target data to be identified.

[0110] Taking the multispectral sensor OLI in Landsat 8 as an example, this embodiment reads in the hyperspectral single-band image at the median of the single-band wavelength range of OLI. For example, the wavelength range of the second band of Landsat 8 is 0.45-0.51 μm, then the band of GF-5 corresponding to this band read in is the 22nd band with a wavelength of 0.48 μm, and the band read in from the Resource-1 02D satellite is the 11th band with a wavelength of 0.48 μm, and so on, reading in data of seven bands, namely, 3, 4, 5, 6, 7, 9, etc.

[0111] The present application also provides another implementation method of obtaining target data to be identified from hyperspectral remote sensing data, which is parallel to the above-mentioned method and can be used as another optional implementation method, which may include:

[0112] When it is detected that the data input method is the weighted fitting input method, the radiation brightness value of each band after radiation calibration is determined according to the hyperspectral remote sensing data; for each band, the spectral response curve of the multispectral sensor in the current band and the radiation brightness value of the hyperspectral remote sensor in the current band are fitted to obtain the approximate multispectral data of the current band; the approximate multispectral data of each band is used as the target data to be identified.

[0113] This embodiment uses the fine bands of a hyperspectral satellite such as Gaofen-5 or Ziyuan-1 02D satellite to simulate the bands of a multispectral satellite such as Landsat 8 satellite to the greatest extent through weighted fitting. The fitting process can be performed according to the following fitting relationship (17):

[0114]

[0115] Among them, gi(λ) is the spectral response curve of each band of multispectral sensors such as Landsat8, X(λ) is the radiant brightness value of each hyperspectral band in the band range of λ1,λ2 after radiometric calibration of hyperspectral sensors such as Gaofen-5 or Ziyuan-1 02D satellite data, and λ1 and λ2 are the start and end wavelengths of the band respectively. Through weighted fitting, hyperspectral data can be transformed into multispectral data close to multispectral satellites such as Landsat 8 satellite, which is more in line with the original design of Fmask cloud recognition algorithm. Through fitting, the multi-band characteristics of hyperspectral data can be used to reduce the randomness of pixel values, so as to improve the shortcomings of single hyperspectral band input multispectral algorithm.

[0116] This embodiment provides two input methods, namely, central wavelength input and weighted fitting input. Users can flexibly select according to actual needs, focusing on different cloud types to obtain better results, and is also conducive to improving user experience.

[0117] The above embodiment does not impose any limitation on the calculation process of the apparent reflectance information. The present application also provides an optional implementation method for performing image preprocessing on the target data to be identified based on the hyperspectral cloud recognition method to obtain the apparent reflectance information, which may include:

[0118] The gain parameters and bias parameters are obtained from the target data to be identified, and the radiance value of each pixel is calculated according to the gain parameters and bias parameters; the apparent reflectance information is determined according to the radiance value of each pixel, the distance between the sun and the earth, the average solar irradiance at the top of the atmosphere, and the solar zenith angle.

[0119] In this embodiment, the data preprocessing method of the Fmask cloud recognition method is modified. The data preprocessing part of the hyperspectral cloud recognition method adopts a preprocessing method different from that of the multispectral remote sensing satellite Landsat 8. The gain and bias parameters in the file with the suffix RadCal provided by the hyperspectral remote sensing satellite such as Gaofen-5 or Ziyuan-1 02D satellite can be used to calculate the radiance of each pixel according to the following radiance relationship (18):

[0120] L=Gain×DN+Bias (18)

[0121] Among them, DN represents the pixel value in the original image of multispectral remote sensing satellites such as Landsat 8 satellite, L is the radiance corresponding to the pixel, Gain and Bias are the gain parameter and bias parameter provided by the hyperspectral remote sensing satellite. For the bias parameters of all bands of Gaofen-5 and Ziyuan-1 02D satellites, they are all 0, that is, you only need to multiply the DN value file by the gain to get the radiance value.

[0122]

[0123] After calculating the radiation brightness value of each pixel, the apparent reflectance can be calculated according to the ground object reflectance relationship 19, where L is the radiation brightness, D is the distance between the sun and the earth obtained by the imaging time, ESUN is the average solar irradiance at the top of the atmosphere, and θ is the solar zenith angle.

[0124] Among them, since the ESUN values ​​of each band of hyperspectral satellites such as Gaofen-5 and Ziyuan-1 02D satellite cannot be obtained, before the execution step determines the apparent reflectance information according to the radiance value of each pixel, the sun-earth distance, the average solar irradiance at the top of the atmosphere and the solar zenith angle, it also includes:

[0125] The central wavelength and half-wavewidth are obtained from the hyperspectral remote sensing data, and the spectral curve of each band of the hyperspectral satellite is obtained based on the central wavelength and half-wavewidth Gaussian fitting; based on the spectral curve of each band and the solar spectral curve, the average solar irradiance at the top of the atmosphere in each band is calculated.

[0126] In this embodiment, the central wavelength and half-wave width in the file with the suffix Spectralresponse provided by the Gaofen-5 or Ziyuan-1 02D satellite can be used to fit the spectral curve of each band of the hyperspectral satellite, and then combined with the solar spectrum curve, the ESUN value of each band of the Gaofen-5 and Ziyuan-1 02D satellite is finally calculated using the following relationship (20).

[0127]

[0128] Among them, E(λ) is the band spectrum curve, S(λ) is the solar spectrum curve, λ1 and λ2 are the start and end wavelengths of the band respectively. The start and end wavelengths used in this study are the center wavelength minus 3 times the standard deviation and the center wavelength plus 3 times the standard deviation. Substituting the obtained ESUN data into the above relationship (19) can obtain the apparent reflectance data required by the Fmask cloud recognition algorithm.

[0129] In order to partially compensate for the errors caused by removing temperature and matching clouds and cloud shadows, based on the above embodiments, the present application also adds the cloud shadow criterion in the ATCOR atmospheric correction model to the entire technical solution, as shown in formula (21). In the process of introduction, various situations are considered as comprehensively as possible. Correspondingly, Figure 2 As shown, the above steps may include the following steps for determining the cloud information of the hyperspectral remote sensing data based on the cloud recognition results and the cloud shadow recognition results:

[0130] Perform water test on the target data to be identified to determine the non-water area in the target data to be identified; for the non-water area, use the ATCOR algorithm to identify cloud shadows based on the apparent reflectance information to obtain the corrected cloud shadow identification results; perform cloud matching on the cloud identification results, the cloud shadow results and the corrected cloud shadow identification results, respectively, to obtain the shadow area and the corrected shadow area;

[0131] If the difference between the shadow area and the corrected shadow area is greater than the preset area threshold, the cloud information of the hyperspectral remote sensing data is determined based on the cloud recognition results and the cloud shadow recognition results; if the difference between the shadow area and the corrected shadow area is less than or equal to the preset area threshold, the cloud information of the hyperspectral remote sensing data is determined based on the cloud recognition results and the corrected cloud shadow recognition results.

[0132] If the target data to be identified does not contain non-water areas, the cloud information of the hyperspectral remote sensing data is determined based on the cloud identification results and cloud shadow identification results.

[0133] In this embodiment, since the premise of the ATCOR algorithm is a non-water area, a water test is performed before executing the criterion. If it is a non-water area, the operation is performed. At the same time, since the formation principles of mountain shadows and cloud shadows are similar, in practice, the ATCOR cloud shadow criterion cannot distinguish between the two, so a change test is added after the water test, and the preset area threshold can be flexibly selected according to actual conditions. For example, the preset area threshold can be 5% of the entire image area. Correspondingly, it is determined whether the difference between the shadow area after matching the cloud shadow and the shadow area after adding the ATCOR algorithm exceeds 5% of the entire image area. If it exceeds, the ATCOR criterion is not executed. If it does not exceed, the ATCOR criterion is executed. Compared with the small error of a small part of the shadow being identified as a water body, it is more important to correct the huge error caused by the existence of mountain shadows.

[0134] 0.04<ρ(NIR)<0.12andρ(SWIR1)<0.20 (21)

[0135] Among them, ρ(NIR) is the reflectivity of the near-infrared band, and ρ(SWIR1) is the reflectivity of the first short-wave infrared band.

[0136] As can be seen from the above, this embodiment improves the recognition accuracy of the improved Fmask cloud recognition algorithm to the greatest extent by adding cloud and cloud shadow related recognition criteria, and further improves the accuracy of hyperspectral remote sensing images.

[0137] In order to verify that the technical solution provided by this application can effectively improve the cloud recognition accuracy of hyperspectral remote sensing images, this application also conducted verification tests, such as Figure 2 As shown, it may include the following:

[0138] The improved Fmask cloud recognition algorithm, also known as the hyperspectral cloud recognition method, was tested using three GF-5 images and two ZY-1 02D satellite images. The recognition results are shown in the figure. Figure 3-Figure 12 As shown, Figure 3-Figure 7 The central wavelength input method is adopted. Figure 8-Figure 12 A weighted fitting input method was used. Figure 3-Figure 5 is the cloud recognition result of the hyperspectral cloud recognition method without weighted fitting operation, Figure 6-Figure 7 This is the cloud identification result of the hyperspectral cloud identification method without weighted fitting operation. Figure 3 This is the unweighted recognition result of the Gaofen-5 image on July 7, 2019. Figure 4 This is the unweighted recognition result of the Gaofen-5 image on November 1, 2018. Figure 5 This is the unweighted recognition result of the Gaofen-5 image on March 11, 2020. Figure 6 This is the unweighted recognition result of the Resource-1 02D image on June 1, 2020. Figure 7 This is the unweighted recognition result of the Resource-1 02D image on September 7, 2020. Figure 8-Figure 10 The cloud recognition result of the hyperspectral cloud recognition method with weighted fitting operation is shown in Figure 2. Figure 11-Figure 12 Cloud identification results of the hyperspectral cloud identification method with weighted fitting operation. Figure 8 This is the weighted recognition result of the Gaofen-5 image on July 7, 2019. Fig. 9 This is the weighted recognition result of the Gaofen-5 image on November 1, 2018. Fig.10 This is the weighted recognition result of the Gaofen-5 image on March 11, 2020; Fig.11 This is the weighted recognition result of the Resource-1 02D image on June 1, 2020. Fig.12 This is the weighted recognition result of the Resource-1 02D image on September 7, 2020.

[0139] By comparing Figure 3-Figure 12 It can be seen that, compared with the cloud recognition results of other cloud recognition algorithms, it is confirmed that the hyperspectral cloud recognition method provided in this application can effectively improve the cloud recognition effect of hyperspectral images.

[0140] The embodiment of the present invention also provides a corresponding device for the cloud recognition method of remote sensing images, which further makes the method more practical. The device can be described from the perspective of functional modules and hardware. The cloud recognition device for remote sensing images provided by the embodiment of the present invention is introduced below. The cloud recognition device for remote sensing images described below and the cloud recognition method for remote sensing images described above can be referred to each other.

[0141] From the perspective of functional modules, see Fig.13 , Fig.13A structural diagram of a cloud recognition device for remote sensing images provided in an embodiment of the present invention in a specific implementation manner, the device may include:

[0142] The method processing module 131 is used to update the Fmask cloud recognition algorithm applicable to the multispectral sensor according to the data preprocessing method and band information of the hyperspectral sensor in advance to obtain a hyperspectral cloud recognition method;

[0143] The data reading module 132 is used to respond to the data input mode selection instruction and obtain the target data to be identified from the hyperspectral remote sensing data; the data input mode is determined according to the band information of the hyperspectral sensor and the multispectral sensor, and the target data to be identified is the ground object image information corresponding to the matching band of the hyperspectral sensor and the multispectral sensor;

[0144] The data preprocessing module 133 is used to perform image preprocessing on the target data to be identified based on the hyperspectral cloud recognition method to obtain the apparent reflectance information;

[0145] The recognition module 134 is used to call the hyperspectral cloud recognition method to perform cloud recognition and cloud shadow recognition respectively based on the apparent reflectance information, and obtain cloud recognition results and cloud shadow recognition results;

[0146] The result determination module 135 determines the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result.

[0147] Optionally, in some implementations of the present embodiment, the data reading module 132 may be used for: when it is detected that the data input mode is the central wavelength input mode, reading the median of multiple single-band wavelength ranges of the multispectral sensor, corresponding to the image data at the corresponding band of the hyperspectral remote sensing data; reading the solar azimuth angle, solar zenith angle information, number of image rows and columns, and annual accumulated days from the hyperspectral remote sensing data as target data to be identified.

[0148] As another optional implementation parallel to the above embodiment, the above data reading module 132 can be further used for: when it is detected that the data input mode is a weighted fitting input mode, determining the radiation brightness value of each band after radiation calibration according to the hyperspectral remote sensing data; for each band, obtaining the approximate multispectral data of the current band by fitting the spectral response curve of the multispectral sensor in the current band and the radiation brightness value of the hyperspectral remote sensor in the current band; and using the approximate multispectral data of each band as the target data to be identified.

[0149] Optionally, in some other implementations of the present embodiment, the above-mentioned data preprocessing module 133 can also be used to: obtain gain parameters and bias parameters from the target data to be identified, and calculate the radiation brightness value of each pixel according to the gain parameters and bias parameters; determine the apparent reflectance information according to the radiation brightness value of each pixel, the sun-earth distance, the average solar irradiance at the top of the atmosphere, and the solar zenith angle.

[0150] As an optional implementation of the above embodiment, the above data preprocessing module 133 can be further used to: obtain the central wavelength and half-wavewidth from the hyperspectral remote sensing data, and obtain the spectral curve of each band of the hyperspectral satellite according to the Gaussian fitting of the central wavelength and half-wavewidth; calculate the average solar irradiance at the top of the atmosphere of each band according to the spectral curve of each band and the solar spectral curve.

[0151] Optionally, in some other implementations of the present embodiment, the above-mentioned result determination module 135 can also be used to: perform a water test on the target data to be identified to determine the non-water area in the target data to be identified; for the non-water area, based on the apparent reflectance information, use the ATCOR algorithm to perform cloud shadow identification to obtain a corrected cloud shadow identification result; perform cloud matching on the cloud identification result with the cloud shadow result and the corrected cloud shadow identification result, respectively, to obtain the shadow area and the corrected shadow area; if the difference between the shadow area and the corrected shadow area is greater than a preset area threshold, then determine the cloud information of the hyperspectral remote sensing data based on the cloud identification result and the cloud shadow identification result; if the difference between the shadow area and the corrected shadow area is less than or equal to the preset area threshold, then determine the cloud information of the hyperspectral remote sensing data based on the cloud identification result and the corrected cloud shadow identification result.

[0152] As an optional implementation of the above embodiment, the above result determination module 135 can be further used to: if the target data to be identified does not contain a non-water area, determine the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result.

[0153] The functions of the functional modules of the cloud recognition device for remote sensing images described in the embodiment of the present invention can be specifically implemented according to the method in the above method embodiment. The specific implementation process can refer to the relevant description of the above method embodiment, which will not be repeated here.

[0154] It can be seen from the above that the embodiments of the present invention can effectively improve the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites.

[0155] The cloud recognition device for remote sensing images mentioned above is described from the perspective of functional modules. Furthermore, the present application also provides an electronic device, which is described from the perspective of hardware. Fig.14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application in one implementation manner. Fig.14As shown, the electronic device includes a memory 140 for storing a computer program; and a processor 141 for implementing the steps of the cloud recognition method for remote sensing images mentioned in any of the above embodiments when executing the computer program.

[0156] Among them, the processor 141 may include one or more processing cores, such as a 4-core processor, an 8-core processor, and the processor 141 may also be a controller, a microcontroller, a microprocessor or other data processing chip. The processor 141 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 141 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 141 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 141 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0157] The memory 140 may include one or more computer-readable storage media, which may be non-transitory. The memory 140 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the memory 140 may be an internal storage unit of an electronic device, such as a hard disk of a server. In other embodiments, the memory 140 may also be an external storage device of an electronic device, such as a plug-in hard disk equipped on a server, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Further, the memory 140 may also include both an internal storage unit of an electronic device and an external storage device. The memory 140 may not only be used to store application software and various types of data installed in the electronic device, such as: the code of the program for executing the vulnerability processing method, etc., but also be used to temporarily store data that has been output or is to be output. In this embodiment, the memory 140 is at least used to store the following computer program 1401, wherein, after the computer program is loaded and executed by the processor 141, the relevant steps of the cloud recognition method of remote sensing images disclosed in any of the aforementioned embodiments can be implemented. In addition, the resources stored in the memory 140 may also include an operating system 1402 and data 1403, and the storage method may be temporary storage or permanent storage. The operating system 1402 may include Windows, Unix, Linux, etc. The data 1403 may include, but is not limited to, data corresponding to the cloud recognition result of the remote sensing image.

[0158] In some embodiments, the electronic device may further include a display screen 142, an input / output interface 143, a communication interface 144 or a network interface, a power supply 145 and a communication bus 146. Among them, the display screen 142 and the input / output interface 143, such as a keyboard, belong to the user interface, and the optional user interface may also include a standard wired interface, a wireless interface, etc. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device, etc. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface. The communication interface 144 may optionally include a wired interface and / or a wireless interface, such as a WI-FI interface, a Bluetooth interface, etc., which are generally used to establish a communication connection between the electronic device and other electronic devices. The communication bus 146 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.14 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0159] Those skilled in the art will understand that Fig.14 The structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, for example, it may also include a sensor 147 to implement various functions.

[0160] The functions of the functional modules of the electronic device described in the embodiment of the present invention can be specifically implemented according to the method in the above method embodiment. The specific implementation process can refer to the relevant description of the above method embodiment, which will not be repeated here.

[0161] It can be seen from the above that the embodiments of the present invention can effectively improve the cloud recognition accuracy of hyperspectral images of domestic hyperspectral satellites.

[0162] It is understandable that if the cloud recognition method of remote sensing images in the above embodiment is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), electrically erasable programmable ROM, register, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, removable disk, CD-ROM, disk or optical disk, etc. Various media that can store program codes.

[0163] Based on this, an embodiment of the present invention further provides a readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the cloud recognition method for remote sensing images described in any of the above embodiments are performed.

[0164] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the hardware disclosed in the embodiments, including devices and electronic devices, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0165] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0166] The above is a detailed introduction to a cloud recognition method, device, electronic device and readable storage medium for remote sensing images provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made to the present application without departing from the principles of the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A cloud recognition method for remote sensing images, characterized in that: include: According to the data preprocessing method and band information of the hyperspectral sensor, the Fmask cloud recognition algorithm suitable for multispectral sensors is updated in advance to obtain a hyperspectral cloud recognition method; In response to the data input mode selection instruction, the target data to be identified is obtained from the hyperspectral remote sensing data; Based on the hyperspectral cloud recognition method, image preprocessing is performed on the target data to be identified to obtain apparent reflectance information; Based on the apparent reflectance information, calling the hyperspectral cloud recognition method to perform cloud recognition and cloud shadow recognition respectively, to obtain a cloud recognition result and a cloud shadow recognition result; Determining cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result; Wherein, the data input method is determined according to the band information of the hyperspectral sensor and the multispectral sensor, and the target data to be identified is the ground object image information corresponding to the matching bands of the hyperspectral sensor and the multispectral sensor; The method of performing image preprocessing on the target data to be identified based on the hyperspectral cloud recognition method to obtain apparent reflectance information includes: Acquire a gain parameter and a bias parameter from the target data to be identified, and calculate the radiation brightness value of each pixel according to the gain parameter and the bias parameter; Acquire the central wavelength and half-wave width from the hyperspectral remote sensing data, and obtain the spectral curve of each band of the hyperspectral satellite according to the central wavelength and the half-wave width Gaussian fitting; According to the spectrum curve of each band and the solar spectrum curve, the average solar irradiance at the top of the atmosphere in each band is calculated; The apparent reflectance information is determined based on the radiance value of each pixel, the distance between the sun and the earth, the average solar irradiance at the top of the atmosphere, and the solar zenith angle; Wherein, determining the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result includes: Performing a water test on the target data to be identified to determine a non-water area in the target data to be identified; For the non-water area, based on the apparent reflectivity information, an ATCOR algorithm is used to perform cloud shadow recognition to obtain a corrected cloud shadow recognition result; Perform cloud matching on the cloud recognition result, the cloud shadow recognition result and the corrected cloud shadow recognition result, respectively, to obtain a shadow area and a corrected shadow area; If the difference between the shadow area and the corrected shadow area is greater than a preset area threshold, determining the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result; If the difference between the shadow area and the corrected shadow area is less than or equal to a preset area threshold, the cloud information of the hyperspectral remote sensing data is determined according to the cloud recognition result and the corrected cloud shadow recognition result.

2. The cloud recognition method for remote sensing images according to claim 1, characterized in that: The step of obtaining target data to be identified from hyperspectral remote sensing data includes: When it is detected that the data input mode is the central wavelength input mode, the median of multiple single-band wavelength ranges of the multispectral sensor is read, corresponding to the image data at the corresponding band of the hyperspectral remote sensing data; The solar azimuth angle, solar zenith angle information, image row and column numbers and annual accumulated days are read from the hyperspectral remote sensing data as target data to be identified.

3. The cloud recognition method for remote sensing images according to claim 1, characterized in that: The step of obtaining target data to be identified from hyperspectral remote sensing data includes: When it is detected that the data input mode is a weighted fitting input mode, the radiation brightness value after radiation calibration of each band is determined according to the hyperspectral remote sensing data; For each band, the approximate multispectral data of the current band is obtained by fitting the spectral response curve of the multispectral sensor in the current band and the radiant brightness value of the hyperspectral remote sensor in the current band; The approximate multispectral data of each band is used as the target data to be identified.

4. The cloud recognition method for remote sensing images according to claim 1, characterized in that: After the water test is performed on the target data to be identified, the method further includes: If the target data to be identified does not include a non-water area, the cloud information of the hyperspectral remote sensing data is determined according to the cloud identification result and the cloud shadow identification result.

5. A cloud recognition device for remote sensing images, characterized in that: include: The method processing module is used to update the Fmask cloud recognition algorithm applicable to the multispectral sensor according to the data preprocessing method and band information of the hyperspectral sensor in advance, so as to obtain the hyperspectral cloud recognition method; A data reading module, for responding to a data input mode selection instruction, and obtaining target data to be identified from the hyperspectral remote sensing data; the data input mode is determined according to the band information of the hyperspectral sensor and the multispectral sensor, and the target data to be identified is the ground object image information corresponding to the matching band of the hyperspectral sensor and the multispectral sensor; A data preprocessing module, used to perform image preprocessing on the target data to be identified based on the hyperspectral cloud recognition method to obtain apparent reflectance information; an identification module, configured to call the hyperspectral cloud identification method to perform cloud identification and cloud shadow identification respectively based on the apparent reflectance information, and obtain a cloud identification result and a cloud shadow identification result; A result determination module, which determines the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result; Wherein, the data preprocessing module is further used for: Acquire a gain parameter and a bias parameter from the target data to be identified, and calculate the radiation brightness value of each pixel according to the gain parameter and the bias parameter; Acquire the central wavelength and half-wave width from the hyperspectral remote sensing data, and obtain the spectral curve of each band of the hyperspectral satellite according to the central wavelength and the half-wave width Gaussian fitting; According to the spectrum curve of each band and the solar spectrum curve, the average solar irradiance at the top of the atmosphere in each band is calculated; The apparent reflectance information is determined based on the radiance value of each pixel, the distance between the sun and the earth, the average solar irradiance at the top of the atmosphere, and the solar zenith angle; Wherein, the result determination module is further used for: Performing a water test on the target data to be identified to determine a non-water area in the target data to be identified; For the non-water area, based on the apparent reflectivity information, an ATCOR algorithm is used to perform cloud shadow recognition to obtain a corrected cloud shadow recognition result; Perform cloud matching on the cloud recognition result, the cloud shadow recognition result and the corrected cloud shadow recognition result, respectively, to obtain a shadow area and a corrected shadow area; If the difference between the shadow area and the corrected shadow area is greater than a preset area threshold, determining the cloud information of the hyperspectral remote sensing data according to the cloud recognition result and the cloud shadow recognition result; If the difference between the shadow area and the corrected shadow area is less than or equal to a preset area threshold, the cloud information of the hyperspectral remote sensing data is determined according to the cloud recognition result and the corrected cloud shadow recognition result.

6. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the processor is used to implement the steps of the cloud recognition method for remote sensing images as claimed in any one of claims 1 to 4 when executing a computer program stored in the memory.

7. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the cloud recognition method for remote sensing images as claimed in any one of claims 1 to 4 are implemented.

Images