42 results about "Cloud top" patented technology

The invention discloses an RVM (relevant vector machine) method for RMW (radius of maximum wind) and typhoon eye dimension modeling, in order to establish a model of eye typhoon RMW and typhoon eye dimension based on the RVM. Hereto, according to the invention, the method comprises the following steps of: estimating the RMW through calculating the distance between the coldest temperature point at the top of the cloud and the warmest point within an eye wall on a typhoon cloud picture; dividing the typhoon eye based on patial differential equation, and measuring the dimension of the typhoon eye by the perimeter of the eye wall; and finally establishing the model of the RMW and the typhoon eye dimension based on the RVM, and comparing with the modeling methods of SVM (support vector machine), linear regression, radial basis function neural network and the like. Because the technical scheme provided by the invention is adopted, a good modeling result can be obtained; and compared with other modeling methods, the RVM-based modeling effect has much smaller errors, so that the inner core structure of a wind field at the low layer of typhoon can be depicted favorably.

The present invention is directed toward specially tinted lenses for use in flying airplanes. In the present invention, the far field of vision is highly tinted to reduce the light from a sunlit sky, over sunlit cloud tops, or in clouds. This enables an aircraft pilot to easily see out of the cockpit into highly illuminated environments. The intermediate field of vision is essentially clear to enable the pilot to easily see a dimly illuminated instrument panel as compared to outside the aircraft. The near field of vision may have a range of tinting from clear to up to a medium tinted value for reading maps and other material. In addition, the change in tinting is discontinuous, that is, sharp, from one field of vision to another. The present invention works because the aircraft structure, that is, the nose and instrument panel, blocks outside light from coming into the intermediate field of vision where the lens is clear.

The invention relates to a cloud radar and satellite detection data fusion method and a cloud radar and satellite detection data fusion system. The cloud radar and satellite detection data fusion method is characterized in that cloud top heights, cloud bottom heights, and average reflectivity factors detected by a millimeter-wave cloud radar can be achieved, and the satellite cloud picture brightness temperature data corresponding to the overhead part of the millimeter-wave cloud radar can be achieved; the cloud top heights, the cloud bottom heights, and the average reflectivity factors corresponding to the brightness temperature values of the area covered by the satellite cloud picture can be output in a real-time manner by the processing of the trained neural network of the satellite brightness temperatures, and the cloud top heights, the cloud bottom heights, and the average reflectivity factors detected by the millimeter-wave cloud radar. The horizontal distribution chart of the cloud top heights, the cloud bottom heights, and the average reflectivity factors of the area covered by the satellite cloud picture can be achieved, and then the new cloud information can be provided for the weather forecast analysis, the mathematical forecast, and the atmosphere scientific research.

The invention discloses a cloud top height retrieval method based on satellite tri-linear array CCD (charge coupled device) image. The method comprises the following steps: firstly selecting a cloud region in a foresight image, carrying out image matching to obtain a corresponding matching point in a back-sight image; secondly, obtaining the continuous static satellite image of adjacent time sections of the foresight image, and retrieving the wind speed and direction in the corresponding region; then, obtaining the time interval of cloud points in the foresight image and the back-sight image, and computing the displacement of the cloud points; and finally inferring the forward interaction formula of a moving target and combining the relative information of cloud point displacement and matching pair to obtain the cloud top height information. In the process of detecting the cloud top height by using the tri-linear array detector, the defects that height detection of the cloud tope height is greatly influenced caused by the moving of the cloud are compensated.

The invention relates to a method for assimilating cloud-to-ground lightning data through physical initialization. The method comprises the following steps that (1), three-dimensional agent radar echoes converted through the lightning data are screened in the vertical direction to obtain the maximum echo reflectivity factor Ze; (2), precipitation intensity RR'obs is obtained according to a formula, and the precipitation intensity is converted into the precipitation flux mode RRobs; (3), the cloud top height is set to be the top height of the three-dimensional agent radar echoes, cloud bottom height is approximately substituted by the isentropic condensation level of a background field; (4), vertical ascending velocity in a cloud is expressed; (5), when the precipitation intensity RR'obs is smaller than the area of 0.1, the vertical velocity is kept consistent with that of the background, specific humidity and the cloud water content are not changed, the cloud water content is set to be zero, maximum relative humidity cannot exceed 80 percent, and when the precipitation intensity RR'obs is larger the area of 0.1, the specific humidity and the cloud water content are adjusted. The method can effectively improve nowcasting precision of a mode on a severe convection system.

The invention discloses a surface temperature inversion method based on intermediate infrared and thermal infrared data. The method comprises the steps that 1) the cloud top brightness temperatures ofintermediate infrared and thermal infrared channels under different atmosphere, surface and observation angle combination conditions are simulated, thereby generating a simulated data set; 2) according to the atmospheric column water vapor content, the cloud top brightness temperatures and the observation angles, simulated data in the simulated data set is divided into a plurality of sub-intervals; 3) surface temperature inversion algorithms of daytime data and night data are built in combination with a split window algorithm principle; and 4) a to-be-processed actual remote sensing image iscalculated by utilizing the surface temperature inversion algorithms of the daytime data and the night data, corresponding surface temperature inversion algorithm coefficients are selected according to the sub-intervals corresponding to the water vapor content, the cloud top brightness temperatures and the observation angles of all the pixels of the actual remote sensing image, then the surface emissivity of the actual remote sensing image is obtained in combination with a surface classification product, and a surface temperature inversion result is obtained. According to the method, the high-precision surface temperature can be obtained.

The present invention is directed toward specially tinted lenses for use in flying airplanes. In the present invention, the far field of vision is highly tinted to reduce the light from a sunlit sky, over sunlit cloud tops, or in clouds. This enables an aircraft pilot to easily see out of the cockpit into highly illuminated environments. The intermediate field of vision is essentially clear to enable the pilot to easily see a dimly illuminated instrument panel as compared to outside the aircraft. The near field of vision may have a range of tinting from clear to up to a medium tinted value for reading maps and other material. In addition, the change in tinting is discontinuous, that is, sharp, from one field of vision to another. The present invention works because the aircraft structure, that is, the nose and instrument panel, blocks outside light from coming into the intermediate field of vision where the lens is clear.

The invention relates to the technical field of satellite lightning imager data assimilation, in particular to a lightning imager data assimilation method which comprises the following steps: S1, gridding a data mode of a FengYun-4 satellite lightning imager; S2, gridding a cloud top height data mode of the FengYun-4 satellite; S3, creating a pseudo relative humidity field; S4, obtaining an analysis field; s5, forecasting a numerical mode; s6, forecasting and checking. According to the method, new cloud top height observation data is introduced in the data assimilation process of the FengYun-4satellite lightning imager, a pseudo relative humidity field close to observation is obtained by adjusting the relative humidity in a background field, and the pseudo relative humidity field is assimilated into a numerical mode by utilizing a 3DVAR assimilation system which is economical in calculation and convenient for business. The test result shows that the assimilation test obviously improves the forecasting of the numerical mode on the heavy rainfall, the ETS and FSS scores of the accumulated rainfall in one hour are improved, and the forecasting maintenance time can reach 6 hours.

The invention relates to a cloud base height inversion method based on a foundation cloud picture. The cloud base height inversion method comprises the following steps: (1) selecting places, deployingtwo whole-sky imaging instruments, setting a field angle of the two whole-sky imaging instruments to be not greater than 130 degrees, enabling the two whole-sky imaging instruments to perform synchronous observation, and enabling the observed foundation cloud picture to have an overlapped area; (2) acquiring the foundation cloud picture synchronously observed by the two all-sky imaging instruments, performing the image matching for the two foundation cloud pictures, and selecting the overlapped area with the smallest difference; and (3) for a cloud cluster selected in the overlapped area in amatching manner, and a cloud base height is inversed by utilizing a geometric relation to obtain an inversion result. Compared with the satellite cloud picture observed from the cloud top in the existing inversion method, the foundation cloud picture is observed on the ground by the whole-sky imaging instruments, i.e., the foundation cloud picture is observed from the cloud base, so that the accuracy of the cloud base height inversion result is improved, and the effect is more significant for the thick cloud and the low cloud. In addition, the application range of the whole-sky imaging instrument is also expanded.

The invention relates to a cloud layer height measurement system based on an aircraft. The cloud layer height measurement system comprises an air measuring device and a ground control device. The air measuring device comprises the aircraft, a first control module, a location module, an airborne communication module and an image shooting device, wherein the first control module, the location module, the airborne communication module and the image shooting device are arranged on the aircraft, the image shooting device is perpendicularly arranged on the surface of the aircraft upwards in the shooting direction, the ground control module comprises a second control module, a ground communication module and an information output device, and the ground communication module and the information output device are connected with the second control module. The invention meanwhile relates to a cloud layer height measurement method. By means of the cloud layer height measurement system and the cloud layer height measurement device based on the aircraft, the cloud top height or the cloud base height can be flexibly and accurately measured according to measurement requirements, and the measurement work efficiency is effectively improved.

The invention discloses a method for modeling the cumulus cloud scene based on a Landsat8 satellite image, belongs to the field of computer graphics, and can automatically construct the geometrical shape and the extinction coefficient of a cumulus cloud from the Landsat8 satellite image. The method comprises the steps of firstly, differentiating image pixels into two types which are cumulus cloud and earth surface through the difference of cloud and earth surface in an visible image and a long-wave infrared image, secondly, computing the height of cloud top through the linear relation between the temperature and the height of the cumulus cloud, further estimating the height of cloud base, then computing the optical thickness of the cloud through a bi-directional reflectance function, further computing the extinction coefficient of the cloud, and finally optimizing the shape of the cloud, performing sampling at the interior of the cloud, and forming a particle model of the cloud and performing drawing. The method has the advantages that a remote sensing theory is introduced to guide the construction of the cumulus cloud scene, the facticity and the fidelity of the cumulus cloud scene can be effectively improved, the modeling process is completely automatic, and the efficiency is high.

The invention provides a cloud sea natural landscape forecast method and system. The method provided by the invention comprises the steps that intraday first meteorological data acquired by a first automatic meteorological station are acquired, wherein the first automatic meteorological station is located at the height of cloud top above cloud sea; intraday second meteorological data acquired by a second automatic meteorological station are acquired, wherein the second automatic meteorological station is located in cloud sea; intraday weather forecast data released by the meteorology bureau are acquired; intraday cloud sea condition is acquired; according to the first meteorological data, the second meteorological data, the weather forecast data and the intraday cloud sea condition, the value of each independent variable in a forecast equation is acquired; and the forecast equation is input to acquire the cloud sea forecast result of the second day, wherein the forecast equation is acquired in advance through logistics regression analysis. The method and system can provide cloud sea landscape probability forecast of high accuracy rate, facilitate the travel plan of passengers and photographers, and promote the development of the local tourism industry.

The present invention discloses a cloud image retrieval method based on a shape feature. In effective combination with brightness temperature information of a cloud top of a cloud image, a cloud system is segmented by using an iteration threshold segmentation method; and a shape feature of the cloud system is extracted by using a geometric invariant moment that is relatively low in computational complexity and strong in robustness, thereby overcoming difficulty in analyzing the cloud image by using a conventional cloud image retrieval method based on the shape feature, and especially overcoming problems of method universality, computational complexity and robustness. In addition, according to the method, a bottom cloud image feature is transformed from euclidean space to sparse space, and a cloud image database is retrieved by using a distribution rule of the sparse space, thereby effectively alleviating a semantic gap problem of a conventional retrieval method.

A weather forecasting system may receive satellite image samples and identify an updraft and components of the updraft within a cloud. These satellite image samples are collected over time (e.g., at 30 second to 1 minute time intervals). The system may identify an area of rotation and / or divergence at cloud top in a cumulus cloud or mature convective storm over time by comparing the samples and determine a parameter indicative of the updraft based on the area of rotation and divergence. The system may estimate aspects of the environment related to storm development and predict the occurrence of a weather event in the future based on the parameter and generate an output indicative of the occurrence.

The invention discloses a multispectral rainfall detection method based on a random forest algorithm. The method comprises the steps: carrying out observation of the same cloud region at the same timethrough a visible and infrared spin scanning radiometer carried on a stationary satellite and a microwave temperature detector carried on a polar orbit satellite, and obtaining observation data; after visible and infrared spin scanning radiometer high-resolution cloud products are matched into microwave temperature detector phase elements, acquiring information on a surface of a cloud body in AMSU-A image elements of a microwave temperature detector and information inside the cloud body; and then simulating a nonlinear relationship between cloud top optical information and microwave rainfallinformation in the cloud body by using the random forest algorithm, and establishing a relationship between the optical information and the microwave rainfall information. The invention further provides a multispectral rainfall detection system based on the random forest algorithm. The system comprises a data acquisition module, a matching module, a modeling module, a model verification module anda rainfall detection module. Compared with a traditional detection method, the detection system and method have higher accuracy, higher detection rate and lower error rate.

The invention discloses a cloud top height real-time calculation method based on satellite infrared detection element parameters, and the method comprises the steps: firstly collecting a cloud profile data set, satellite hyperspectral sensor parameters and a clear sky atmosphere profile data set, and carrying out cloud sky radiation transmission calculation and clear sky radiation transmission calculation on the data sets to obtain a clear sky environment brightness temperature difference value and a cloud sky environment brightness temperature difference value; calculating by combining the calculated bright temperature difference value of the clear sky and the cloud sky environment with the cloud profile data set to obtain a lookup table of the sub-pixel cloud cover, the cloud top height and the bright temperature difference; checking whether the cloud top height meets conditions or not by comparing the lookup table layer by layer; if yes, the cloud cover and the cloud top height of the hyperspectral pixel are obtained; and if not, continuing to execute the steps. The method is reliable in data calculation.

A specially tinted lens, a tri-colored lens for spectacles and sunglasses for use in flying airplanes. A far field of vision has a first tinting, such as green, to reduce the light from a sunlit sky, over sunlit cloud tops, or in clouds, which enables an aircraft pilot to easily see out of the cockpit into highly illuminated environments. An intermediate field of vision has a second tinting, such as yellow, to enable the pilot to easily see a dimly illuminated instrument panel as compared to outside the aircraft. A near field of vision has a third tinting, such as neutral in color, to enable said pilot to easily view maps and other information.

The invention discloses an automatic cloud signal recognition method based on a finite-state machine. According to the method, basic characteristics of the cloud signals are generalized into cloud base, cloud peak, cloud top and public boundary of two clouds according to the whole characteristics of the cloud signals; ceilometer signals can be separately chosen out by encoding ceilometer echo signals into cloud signal description sentences. The method comprises the steps that the cloud signals and related inversion parameters are smoothly filtered, and a first-order derivative of the echo signals is obtained; then the cloud base, the cloud peak, the cloud top and the public boundary of the two clouds are encoded into the cloud signal description sentences; the situation that the signals of rain, fog and peak interference are judged as the cloud signal in error is avoided by recognizing and logically analyzing the characteristics of the cloud signals; positions of the cloud base, the cloud peak and the cloud top in a cloud layer can be given out. The method can meet the situation of recognizing three layers of clouds to the maximum degree and improves the cloud signal recognition success rate.

The invention discloses a method for improving the spatial resolution of precipitation data. The method comprises the following steps: integrating precipitation data of a stationary satellite, performing precipitation calculation on the high spatial resolution of infrared cloud top brightness temperature of the stationary satellite, and downscaling coarse-resolution precipitation data. The spatialresolution of precipitation data on the inland and ocean in the global range can be improved from 0.25 DEG to 0.05 DEG, so a user can obtain finer precipitation spatial distribution data in the global range, and richer information is provided for flood disaster monitoring, drought monitoring and water resource distribution investigation in the global range. Compared with an existing downscaling method, the precipitation downscaling method provided by the invention can ensure that the downscaled precipitation data is not reduced in precision and time resolution compared with original data while the spatial resolution of the data is improved, and meanwhile, the spatial distribution trend of the original precipitation data is reserved.

The invention discloses a remote sensing detection system for the condensable water amount of an air cloud system. The remote sensing detection system includes a mie-scattering laser radar which is used for acquiring height values of a cloud bottom and a cloud top, a rotating Raman laser radar which is used for acquiring atmospheric temperature profile data, a wind measurement laser radar which is used for acquiring a vertical movement speed of an air flow, and a micro-rain radar which is used for judging whether rainfall exists. The Mie scattering laser radar, the rotating Raman laser radar, the wind measurement laser radar and the micro rain radar are all connected to the data processing system. The detection method comprises the following steps: 1, vertically observing the sky; 2, obtaining the height of the cloud bottom and the height of the cloud top; 3, calculating the saturated water vapor density of the cloud bottom and the cloud top; 4, obtaining the vertical airflow movement speed;5, calculating the condensable water quantity in the atmosphere within a period of time; and 6, calculating the condensable water quantity in the cloud. according to the invention, high-precision detection of the air condensable water quantity can be realized, and the problem that the air condensable water quantity cannot be observed in the existing atmospheric science field is solved.

The embodiment of the invention discloses a method and device for reconstructing a three-dimensional cloud body through satellite-ground cloud fusion, electronic equipment and a medium. The method for reconstructing the three-dimensional cloud body through satellite-ground cloud fusion comprises the steps of obtaining foundation cloud observation data of at least three cloud radars; acquiring multichannel scanning imaging radiometer data and cloud detection product data of the geostationary orbit meteorological satellite; performing space-time matching on the foundation cloud observation data, the multi-channel scanning imaging radiometer data and the cloud detection product data to obtain satellite-ground matching data; cloud top height inversion of a satellite in a preset area around each cloud radar installation point is realized based on the satellite-ground matching data, and the cloud top height of a cloud radar networking area is obtained; obtaining the cloud bottom height of the cloud radar networking area based on the ground-based cloud observation data; obtaining a three-dimensional cloud body boundary based on the cloud top height and the cloud bottom height; and obtaining a reconstructed three-dimensional cloud body based on the reflectivity factor profile and the radial speed profile in the foundation cloud observation data.

A weather forecasting system may receive satellite image samples and identify an updraft and components of the updraft within a cloud. These satellite image samples are collected over time (e.g., at 30 second to 1 minute time intervals). The system may identify an area of rotation and / or divergence at cloud top in a cumulus cloud or mature convective storm over time by comparing the samples and determine a parameter indicative of the updraft based on the area of rotation and divergence. The system may estimate aspects of the environment related to storm development and predict the occurrence of a weather event in the future based on the parameter and generate an output indicative of the occurrence.

The invention provides a spatial downscaling method and system for remote sensing precipitation data, and the method comprises the steps: introducing stationary satellite fusion data, and also comprises the following steps: carrying out the high-spatial-resolution precipitation calculation based on the infrared cloud top brightness temperature of a stationary satellite; and downscaling the coarse-resolution precipitation data. Based on the spatial downscaling method and system for remote sensing rainfall data, by introducing fusion data NCEP-IR of stationary satellite infrared band brightnesstemperature with high temporal-spatial resolution, establishing a downscaling method, spatial downscaling of coarse-resolution (0.25 degree*0.25 degree) rainfall data on land and sea is realized in aglobal medium and low latitude range (50N-50S, 180W-180E); and precipitation data with the spatial resolution of 0.05 degree*0.05 degree is obtained, and it is ensured that the precision and the timeresolution of the downscaled precipitation data are not reduced compared with those of the original precipitation data.

The invention discloses a ground surface temperature inversion method based on mid-infrared and thermal infrared data. The steps include: 1) simulating the cloud top brightness temperature of mid-infrared and thermal infrared channels under different combination conditions of atmosphere, surface and observation angles, and generating simulated Data set; 2) Divide the simulated data in the simulated data set into several sub-intervals according to the water vapor content of the atmospheric column, the cloud top brightness temperature and the observation angle; 3) Combining the principle of the split window algorithm to construct the surface temperature reflections of the daytime data and nighttime data respectively 4) Using the surface temperature inversion algorithm of daytime data and the surface temperature inversion algorithm of nighttime data to calculate the actual remote sensing image to be processed, according to the water vapor content of each pixel of the actual remote sensing image, cloud top brightness temperature and observation angle Select the corresponding surface temperature inversion algorithm coefficients for the corresponding sub-intervals, and then combine the surface classification products to obtain the surface emissivity of the actual remote sensing image, and obtain the surface temperature inversion results. The invention can acquire high-precision ground surface temperature.

The present invention provides a space downscaling method and system for remote sensing precipitation data, wherein the method includes introducing geostationary satellite fusion data, and further includes the following steps: high spatial resolution precipitation calculation based on geostationary satellite infrared cloud top brightness temperature; coarse resolution precipitation data downscaling. A method and system for spatial downscaling of remote sensing precipitation data proposed by the present invention, by introducing fusion data NCEP-IR of brightness temperature in the infrared band of stationary satellites with high temporal and spatial resolution, establishes a downscaling method, in the range of middle and low latitudes in the world ( 50N~50S, 180W~180E) to achieve spatial downscaling of precipitation data with a coarse resolution (0.25°×0.25°) on land and ocean, and obtain precipitation data with a spatial resolution of 0.05°×0.05° while ensuring downscaling Compared with the original precipitation data, the post-precipitation data does not decrease in accuracy and time resolution.

The invention relates to a cloud base height inversion method based on a foundation cloud picture. The cloud base height inversion method comprises the following steps: (1) selecting places, deployingtwo whole-sky imaging instruments, setting a field angle of the two whole-sky imaging instruments to be not greater than 130 degrees, enabling the two whole-sky imaging instruments to perform synchronous observation, and enabling the observed foundation cloud picture to have an overlapped area; (2) acquiring the foundation cloud picture synchronously observed by the two all-sky imaging instruments, performing the image matching for the two foundation cloud pictures, and selecting the overlapped area with the smallest difference; and (3) for a cloud cluster selected in the overlapped area in amatching manner, and a cloud base height is inversed by utilizing a geometric relation to obtain an inversion result. Compared with the satellite cloud picture observed from the cloud top in the existing inversion method, the foundation cloud picture is observed on the ground by the whole-sky imaging instruments, i.e., the foundation cloud picture is observed from the cloud base, so that the accuracy of the cloud base height inversion result is improved, and the effect is more significant for the thick cloud and the low cloud. In addition, the application range of the whole-sky imaging instrument is also expanded.

The invention provides a cloud sea natural landscape forecast method and system. The method provided by the invention comprises the steps that intraday first meteorological data acquired by a first automatic meteorological station are acquired, wherein the first automatic meteorological station is located at the height of cloud top above cloud sea; intraday second meteorological data acquired by a second automatic meteorological station are acquired, wherein the second automatic meteorological station is located in cloud sea; intraday weather forecast data released by the meteorology bureau are acquired; intraday cloud sea condition is acquired; according to the first meteorological data, the second meteorological data, the weather forecast data and the intraday cloud sea condition, the value of each independent variable in a forecast equation is acquired; and the forecast equation is input to acquire the cloud sea forecast result of the second day, wherein the forecast equation is acquired in advance through logistics regression analysis. The method and system can provide cloud sea landscape probability forecast of high accuracy rate, facilitate the travel plan of passengers and photographers, and promote the development of the local tourism industry.

Provided is a millimeter wave cloud radar based on a frequency modulation interrupted continuous wave. The millimeter wave cloud radar comprises an antenna, a high-frequency box and an antenna base, the high-frequency box comprises a circulator, an emitter, a receiver, a signal processor, a data processing unit and a power supply module. The millimeter wave cloud radar outputs data including a cloud reflectivity intensity, a cloud vertical movement speed, a cloud top height, a cloud bottom height etc. given according to the height; and the millimeter wave cloud radar employs a frequency modulation interrupted continuous wave technology so that problems of large operating distance, high resolution and lack of blind area rapid detection can be well solved.

The invention relates to the field of meteorological analysis, discloses a daytime cloud phase inversion method based on a polar orbit meteorological satellite MERSI, and develops a daytime cloud phase identification product by utilizing radiation observation of a medium resolution spectral imager MERSI of a second generation polar orbit meteorological satellite FY-3D in China. Based on MERIS L1 data of radiometric calibration and geographic positioning and a cloud detection product, in combination with earth surface type data, by utilizing spectrum and texture characteristics of visible light, infrared and near-infrared channels and by utilizing different thermodynamic phase states of cloud top particles, cloud phase state identification is carried out on cloud pixels by adopting a threshold value method, and a cloud phase state product with the resolution of 1 km is obtained. By implementing the method, the identification precision of the cloud phase state can be effectively improved, and particularly, the mixed cloud phase state can be effectively identified.

The present invention discloses a cloud image retrieval method based on a shape feature. In effective combination with brightness temperature information of a cloud top of a cloud image, a cloud system is segmented by using an iteration threshold segmentation method; and a shape feature of the cloud system is extracted by using a geometric invariant moment that is relatively low in computational complexity and strong in robustness, thereby overcoming difficulty in analyzing the cloud image by using a conventional cloud image retrieval method based on the shape feature, and especially overcoming problems of method universality, computational complexity and robustness. In addition, according to the method, a bottom cloud image feature is transformed from euclidean space to sparse space, and a cloud image database is retrieved by using a distribution rule of the sparse space, thereby effectively alleviating a semantic gap problem of a conventional retrieval method.