A method, medium and system for generating layers by cropping online satellite images

Abstract

The present invention provides a method, medium and system for generating layers by cropping online satellite images, belonging to the technical field of image data processing. Specifically, by establishing a satellite image spatial index database, multi-level image pyramids and multi-level image memory pools, the satellite image data to be processed is stored in partitions and hierarchically managed. Based on the parameters in the layer cropping request, the image slice data to be processed is obtained from the target pyramid level and allocated to the corresponding target memory pool level. The radiative transfer equation set is used for atmospheric correction processing and the weight calculation equation set is used for edge feathering processing, and finally the target layer data is generated. Among them, the multi-level image pyramids are stored in the system cache, and at the same time, multi-level image memory pools are established, and the number of levels of the multi-level image memory pools corresponds one-to-one with the number of levels of the multi-level image pyramids. The present invention solves the technical problem that the quality of the cropped layers formed after splicing in the prior art is not ideal.

Description

Technical Field

[0001] The present invention belongs to the technical field of image data processing. Specifically, it relates to a method, medium and system for generating layers by cropping online satellite images. Background Art

[0002] With the rapid development of remote sensing technology and the application of satellite images, satellite image data is increasingly widely used in fields such as geographic information systems, natural resource surveys, urban planning, and environmental monitoring. Satellite image data has the characteristics of a wide spatial range, high resolution, and high frequency. The data volume of a single scene of image data usually reaches several GB or even dozens of GB. In practical applications, users often need to perform online real-time processing on a large amount of satellite image data to obtain image layers of specific regions.

[0003] Currently, mainstream satellite image processing systems adopt the traditional batch processing mode, that is, first download the entire scene of image data to the local, and then perform processing such as cropping and correction. This processing mode has problems such as a large amount of data transmission, a long processing time, and high consumption of system resources. Especially in the application scenario facing network services, users need to quickly obtain image data of the area of interest, and the traditional batch processing mode is difficult to meet the real-time requirements.

[0004] Existing online satellite image processing technologies mainly adopt preprocessing and caching strategies, that is, preprocess satellite images by block processing and atmospheric correction, and store the processing results in the cache. Although this method can improve the data access speed, there are problems that the preset block size and correction parameters in the preprocessing method may not match the actual needs of users, resulting in unsatisfactory quality of the cropped layer formed after splicing. Summary of the Invention

[0005] In view of this, the present invention provides a method, medium and system for generating layers by cropping online satellite images, which can solve the technical problem that the quality of the cropped layer formed after splicing in the prior art is unsatisfactory.

[0006] The present invention is implemented as follows:

[0007] The first aspect of the present invention provides a method for generating layers by cropping online satellite images, including: partitioning and storing and hierarchically managing the satellite image data to be processed by establishing a satellite image spatial index database, a multi-level image pyramid, and a multi-level image memory pool, obtaining the image slice data to be processed from the target pyramid level based on the parameters in the layer cropping request and allocating it to the corresponding target memory pool level, performing atmospheric correction processing using the radiative transfer equation set and edge feathering processing using the weight calculation equation set, and finally generating the target layer data.

[0008] On the basis of the above technical solutions, the method for generating layers by online satellite image cropping according to the present invention can be further improved as follows:

[0009] Among them, the establishment of the satellite image spatial index database is specifically as follows: the satellite image data to be processed is stored in partitions, and the satellite image data to be processed includes original data of solar zenith angle, original data of extraterrestrial irradiance, original data of atmospheric optical thickness, original data of scattering phase function, and original data of surface reflectance.

[0010] Furthermore, the establishment of the multi-level image pyramid and the multi-level image memory pool is specifically as follows: a multi-level image pyramid is generated for the satellite image data to be processed according to a preset image scaling level, and the multi-level image pyramid is stored in the system cache. At the same time, a multi-level image memory pool is established, and the number of levels of the multi-level image memory pool corresponds one-to-one with the number of levels of the multi-level image pyramid.

[0011] Furthermore, the parameters based on the layer cropping request are specifically as follows: a layer cropping request is received, and according to the spatial range parameter and the target resolution parameter included in the layer cropping request, the target satellite image data is located from the satellite image spatial index database.

[0012] Furthermore, the obtaining of the image slice data to be processed from the target pyramid level is specifically as follows: according to the target resolution parameter, the corresponding target pyramid level in the multi-level image pyramid is selected, the image slice data to be processed in the target pyramid level is obtained, and the image slice data to be processed is allocated to the target memory pool level corresponding to the target pyramid level in the multi-level image memory pool.

[0013] Furthermore, the radiative transfer equation set includes a solar radiation calculation equation, an atmospheric scattering calculation equation, a surface reflection calculation equation, a sensor reception calculation equation, and a radiation component decomposition calculation matrix, which are used to perform atmospheric correction processing on the image slice data to be processed in the target memory pool level to obtain corrected image slice data.

[0014] Furthermore, the weight calculation equation set includes a distance weight calculation equation, a gradient weight calculation equation, a brightness weight calculation equation, a comprehensive weight calculation equation, and a weight component decomposition calculation matrix, which are used to perform dynamic stitching processing and edge feathering processing on the corrected image slice data in the target memory pool level.

[0015] Further, the generation of the target layer data is specifically as follows: performing a projection transformation process on the stitched image data according to the final pixel mixing weight value output by the weight calculation equation set and the apparent radiance value output by the radiative transfer equation set, performing a cropping process on the image data after the projection transformation according to the spatial range parameter, and outputting the image data after the cropping process as the target layer data.

[0016] For the radiative transfer equation set:

[0017] The solar radiation calculation equation is specifically expressed as follows:

[0018] ;

[0019] In the formula, is the incident radiation flux value, with the unit of ; is the extraterrestrial irradiance raw data, with the unit of ; is the solar zenith angle raw data, with the unit of radian; is the atmospheric optical thickness parameter; is the atmospheric scattering error term, with the range of 0.01 - 0.05.

[0020] The atmospheric scattering calculation equation is specifically expressed as follows:

[0021] ;

[0022] In the formula, is the atmospheric path radiation value, with the unit of ; is the scattering phase function raw data; is the single scattering albedo, with the range of 0 - 1; is the atmospheric optical thickness raw data; is the scattering calculation error term, with the range of 0.02 - 0.06.

[0023] The surface reflection calculation equation is specifically expressed as follows:

[0024] ;

[0025] In the formula, is the surface reflection radiation value, with the unit of ; is the surface reflectance raw data; is the surface reflection calculation error term, with the range of 0.01 - 0.04.

[0026] The sensor reception calculation equation is specifically expressed as follows:

[0027] ;

[0028] In the formula, is the apparent radiance value, with the unit of ; is the sensor response coefficient, with the range of 0.8 to 1.2; is the radiation correction value of the sensor channel; is the number of sensor channels; is the sensor reception error term, with the range of 0.03 to 0.07.

[0029] The radiation component decomposition calculation matrix is specifically expressed as follows:

[0030] ;

[0031] In the formula, is the radiation stable component matrix value; is the radiation variable component matrix value; is the element of the decomposition coefficient matrix, which is obtained by the least squares method.

[0032] The construction principle and meaning of each equation in the radiation transfer equation set are as follows:

[0033] 1. The solar radiation calculation equation is based on the Beer-Lambert law and considers the atmospheric attenuation effect. The exponential term represents the attenuation of solar radiation by the atmosphere, and the cosine term represents the influence of the solar incident angle;

[0034] 2. The atmospheric scattering calculation equation is based on the radiation transfer theory and considers the influence of single scattering and multiple scattering. The term in the denominator represents the solid angle integral of scattering, and the exponential term represents the optical path attenuation;

[0035] 3. The surface reflection calculation equation is based on the Lambert reflection model and considers the isotropic reflection characteristics of the surface. The

[0036] term is used for normalization, and the exponential term represents the atmospheric transmittance;

[0037] 5. The radiation component decomposition matrix uses the linear decomposition method to decompose the total radiation into stable components and variable components, which is helpful for subsequent radiation correction.

[0038] For the weight calculation equation set:

[0039] The distance weight calculation equation is specifically expressed as follows:

[0040] ;

[0041] In the formula, is the initial distance weight value, with a range of 0 to 1; is the pixel coordinate value of the overlapping area; is the coordinate value of the seam line in the overlapping area; is the distance influence factor, with a range of 1 to 5; is the distance weight calculation error term, with a range of 0.01 to 0.03.

[0042] The specific expression of the gradient weight calculation equation is as follows:

[0043] ;

[0044] In the formula, is the initial gradient weight value, with a range of 0 to 1; is the local pixel value of the overlapping area; is the local window radius, with a value of 2 to 5; is the pixel gradient modulus value; is the gradient influence factor, with a range of 10 to 50; is the gradient weight calculation error term, with a range of 0.02 to 0.04.

[0045] The specific expression of the brightness weight calculation equation is as follows:

[0046] ;

[0047] In the formula, is the initial brightness weight value, with a range of 0 to 1; is the current pixel value; is the average brightness value of the local area; is the brightness influence factor, with a range of 15 to 45; is the pixel value at the corresponding position of the adjacent image; is the brightness weight calculation error term, with a range of 0.01 to 0.05.

[0048] The specific expression of the comprehensive weight calculation equation is as follows:

[0049] ;

[0050] In the formula, is the initial pixel mixing weight value, with a range of 0 to 1; is the weight combination coefficient and satisfies ; is the weight exponent term, with a range of 0.5 to 2; is the comprehensive weight calculation error term, with a range of 0.02 to 0.06.

[0051] The weight component decomposition calculation matrix is specifically expressed as follows:

[0052] ;

[0053] In the formula, is the weight stable component matrix value; is the weight variable component matrix value; is the element of the weight decomposition coefficient matrix, which is obtained by singular value decomposition.

[0054] The calculation equation of the final pixel mixing weight value is specifically expressed as follows:

[0055] ;

[0056] In the formula, is the final pixel mixing weight value, and its range is 0 to 1; is the combination coefficient of the stable component and the variable component, and satisfies ; is the attenuation factor, and its range is 3 to 8; is the final weight calculation error term, and its range is 0.01 to 0.04.

[0057] The construction principle and meaning of each equation in the weight calculation equation set are as follows:

[0058] 1. The distance weight calculation equation uses a Gaussian function, which reflects the characteristic that the weight decays as the distance increases, and helps to maintain the smoothness of the edge transition;

[0059] 2. The gradient weight calculation equation combines local difference and global gradient information. The fractional part is used to suppress local noise, and the exponential part is used to maintain edge features;

[0060] 3. The brightness weight calculation equation takes into account both the difference between the pixel and the local mean and the brightness consistency of adjacent images. The fractional part is used to handle the brightness jump at the seam;

[0061] 4. The comprehensive weight calculation equation adopts a power function combination form, and adjusts the contribution degree of each component through the exponential term, which improves the flexibility and adaptability of the weight calculation;

[0062] 5. The weight component decomposition matrix decomposes the weight into two parts, stable and variable, through linear transformation, which is convenient for separate control and optimization;

[0063] 6. The final weight calculation equation adopts a weighted combination form with attenuation. The exponential attenuation term is used to suppress high-frequency changes and improves the stability of the stitching result.

[0064] For the establishment and management of the image memory pool, a multi-level image memory pool establishment and management equation set is adopted, which is specifically expressed as follows:

[0065] The hierarchical quantity calculation equation:

[0066] ;

[0067] In the formula, is the hierarchical quantity of the multi - level image memory pool; is the resolution of the original image; is the target minimum resolution; is the hierarchical redundancy term, with a value range of 1 to 2; represents the ceiling operation.

[0068] The calculation equation for the memory capacity of each layer:

[0069] ;

[0070] In the formula, is the capacity of the th layer memory pool, with the unit of MB; is the memory capacity of the base layer, with the unit of MB; is the capacity attenuation coefficient, with a range of 1.5 to 2.5; is the memory redundancy of the th layer, with a range of 10 to 50 MB.

[0071] The memory allocation optimization equation:

[0072] ;

[0073] In the formula, is the actually allocated memory amount of the th layer, with the unit of MB; is the access frequency factor of the th layer, with a range of 0 to 1; is the data density factor of the th layer, with a range of 0 to 1; is the total available memory amount of the system, with the unit of MB; is the memory allocation error term, with a range of 5 to 15 MB.

[0074] The memory load balancing equation:

[0075] ;

[0076] In the formula, is the load index of the th layer; is the currently used memory amount of the th layer, with the unit of MB; is the data access time of the th layer, with the unit of ms; is the maximum allowable access time, with the unit of ms; is the length of the cache queue at the maximum allowable queue length; is the balance coefficient and satisfies ; is the load calculation error term, with the range of 0.01 - 0.05.

[0077] The memory dynamic adjustment equation:

[0078] ;

[0079] In the formula, is the memory adjustment amount at the layer, with the unit of MB; is the system average load index; is the adjustment coefficient, with the range of 0.1 - 0.5; is the load change rate; is the adjustment amount error term, with the range of 1 - 5MB.

[0080] The construction principle and significance of the equations in the establishment and management equations of the multi - level image memory pool are as follows:

[0081] 1. The layer number calculation equation is based on the resolution ratio relationship, adopts the logarithmic form to ensure the rationality of layer division, and the redundant term ensures the fault - tolerance ability of the system;

[0082] 2. The memory capacity calculation equation adopts the exponential decay model, reflects the decreasing characteristic of the high - level data volume, and the memory redundancy is used to handle data fluctuations;

[0083] 3. The memory allocation optimization equation considers two key factors of access frequency and data density, and adopts normalization processing to ensure the rationality of allocation;

[0084] 4. The load balancing equation comprehensively considers three dimensions of memory utilization rate, access time and queue length, and realizes multi - objective optimization through weighting;

[0085] 5. The dynamic adjustment equation simultaneously considers the load difference and change trend, and adopts the proportional adjustment method to achieve smooth adjustment and avoid violent fluctuations.

[0086] The one - to - one correspondence relationship between the layer number of the multi - level image memory pool and the layer number of the multi - level image pyramid is specifically expressed as follows:

[0087] ;

[0088] In the formula, is the layer number of the multi - level image memory pool; is the number of levels of the multi - level image pyramid; is the total number of levels.

[0089] The mapping relationship equation between levels:

[0090] ;

[0091] In the formula, is the memory pool space of the th layer; is the pyramid image space of the th layer; represents a one - to - one correspondence.

[0092] The data consistency constraint equation for corresponding levels:

[0093] ;

[0094] In the formula, is the memory pool resolution of the th layer; is the pyramid resolution of the th layer; is the original resolution; is the resolution error term, and the range is 0.1 - 0.5 pixels.

[0095] The construction principles and meanings of these corresponding relationship equations are as follows:

[0096] 1. The one - to - one correspondence of the number of levels ensures the strict correspondence between the memory pool and the image pyramid in structure, facilitating data management and access;

[0097] 2. The mapping relationship equation between levels establishes a spatial mapping rule, ensuring the precise correspondence of data between the memory pool and the pyramid;

[0098] 3. The data consistency constraint equation for corresponding levels stipulates the descending relationship of resolutions, ensuring the continuity and consistency of data between different levels.

[0099] For step S70:

[0100] The projection transformation matrix:

[0101] ;

[0102] In the formula, is the scale transformation parameter, and the range is 0.5 - 2.0; is the shear transformation parameter, and the range is - 0.5 - 0.5; is the translation transformation parameter, and the unit is pixel; is the perspective transformation parameter, with a range of -0.01 to 0.01; is the normalization parameter, usually taken as 1; the projection transformation matrix parameters are calculated by the following equations:

[0103] ;

[0104] ;

[0105] ;

[0106] ;

[0107] In the formula, are the scaling factors of the x-axis and y-axis; is the rotation angle, in radians;

[0108] Perform projection transformation processing on the color-corrected image data:

[0109] ;

[0110] In the formula, are the coordinates after projection, in pixels; are the original coordinates, in pixels; is the scale factor for normalization, and the calculation method is: ; is the projection error term, with a range of 0.1 to 0.3 pixels; the final projection coordinates are calculated as: ;

[0111] Perform coordinate registration on the image data after projection transformation:

[0112] ;

[0113] In the formula, are the coordinates of the projection point, in pixels; are the coordinates of the corresponding point, in pixels; is the regularization coefficient, with a range of 0.01 to 0.1; is the optimization error term, with a range of 0.1 to 0.3 pixels; is the number of control points, not less than 4; the projection correction optimization process uses an iterative method:

[0114] ;

[0115] In the formula, is the number of iterations; is the learning rate, with a range of 0.001 to 0.01;

[0116] Evaluate the projection quality:

[0117] ;

[0118] Wherein, is the projection quality score, with the unit of pixel; are the coordinates of the control points after projection; are the coordinates of the target control points; is the evaluation error term, with the range of 0.05 - 0.15 pixels;

[0119] For step S80:

[0120] Determine the cropping boundary according to the spatial range parameter:

[0121] ;

[0122] Wherein, is the boundary indicating function; is the cropping area, determined by the four vertex coordinates ; is the boundary error term, with the range of 0.5 - 1.5 pixels; The determination of the cropping area adopts the following constraints:

[0123] ;

[0124] ;

[0125] Wherein, is the upper left corner coordinate of the cropping area; is the lower right corner coordinate of the cropping area;

[0126] Smooth the cropped edge area:

[0127] ;

[0128] Wherein, is the smoothed boundary; is the standard deviation of the Gaussian kernel, with the range of 1.0 - 3.0; is the center coordinate of the Gaussian kernel; represents the two-dimensional convolution operation; is the smoothing error term, with the range of 0.2 - 0.6 pixels;

[0129] Perform boundary optimization:

[0130] ;

[0131] Wherein, is the optimized boundary; is the edge intensity of the original image; is the boundary gradient; is the balance coefficient, with a range of 0.1 to 0.5; is the optimization error term, with a range of 0.3 to 0.8 pixels; The boundary gradient is calculated using:

[0132] ;

[0133] For step S90:

[0134] Perform compression encoding on the target layer data:

[0135] ;

[0136] In the formula, is the compressed image; is the original image; is the transform domain coefficient; is the compression coefficient, with a range of 0.1 to 1.0; is the compression error term, with a range of 0.01 to 0.05; The bit rate calculation in the compression process:

[0137] ;

[0138] In the formula, is the bit rate; is the probability of the level quantization level;

[0139] Evaluate the compression quality:

[0140] ;

[0141] In the formula, is the maximum pixel value; is the mean square error; is the mean of the image block; is the standard deviation; is the covariance; is the constant term, which are respectively ; is the pixel dynamic range; is the quality evaluation weight coefficient and satisfies ; is the evaluation error term, with a range of 0.01 to 0.05;

[0142] Optimize the compression encoding:

[0143] ;

[0144] In the formula, is the pixel probability distribution; is the coding coefficient; is a pixel of the neighborhood set; is the sparsity parameter, with a range of 0.01 to 0.1; is the spatial consistency parameter, with a range of 0.05 to 0.15; is the coding error term, with a range of 0.02 to 0.06; Coding optimization adopts an iterative method:

[0145] ; In the formula, is the number of iterations; is the update step size, with a range of 0.001 to 0.01.

[0146] The second aspect of the present invention provides a computer-readable storage medium, wherein program instructions are stored in the computer-readable storage medium, and when the program instructions run on a computer, they are used to execute the method for online satellite image cropping to generate layers described above.

[0147] The third aspect of the present invention provides a system for online satellite image cropping to generate layers, which includes the above-mentioned computer-readable storage medium.

[0148] Compared with the prior art, the method, medium and system for online satellite image cropping to generate layers provided by the present invention first establish an efficient satellite image spatial index database, and manage and quickly retrieve massive image data through multi-level spatial grids. At the same time, by adopting the method of multi-level image pyramids and dynamic memory pools, fast access and efficient caching of the original image data are realized.

[0149] In terms of image preprocessing, the present invention utilizes a complete set of radiative transfer models and weight calculation algorithms to comprehensively eliminate the radiative distortion caused by factors such as atmospheric effects, surface reflections, and sensor characteristics, and through an adaptive pixel stitching strategy, effectively solves the problems of inconsistent brightness and discontinuous edges during the image stitching process. In addition, a geometric correction method based on projective transformation and optimized registration is also adopted to ensure the geometric accuracy of the output image.

[0150] Finally, the present invention realizes the processing of image data such as cropping, resolution resampling, format conversion, and compression coding for GIS application requirements, and generates target layer data that meets user needs. Compared with the existing batch processing methods, this method has higher efficiency and flexibility, and solves the technical problem of the unsatisfactory quality of the cropped layers formed after stitching in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0151] Figure 1 is the flowchart of the method provided by the present invention;

[0152] Figure 2 It is a schematic diagram of the distribution of control points in Embodiment 2. Specific implementation manners

[0153] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0154] As Figure 1 shown, it is a flowchart of a method for generating a layer by cropping online satellite images provided in the first aspect of the present invention. The method includes the following steps:

[0155] S10. Establish a spatial index database for satellite images, and partition and store the satellite image data to be processed. The satellite image data to be processed includes original data of solar zenith angle, original data of extraterrestrial irradiance, original data of atmospheric optical thickness, original data of scattering phase function, and original data of surface reflectance.

[0156] S20. Generate a multi-level image pyramid for the satellite image data to be processed according to a preset image zoom level, and store the multi-level image pyramid in the system cache. Meanwhile, establish a multi-level image memory pool, and the number of levels of the multi-level image memory pool corresponds one-to-one with the number of levels of the multi-level image pyramid.

[0157] S30. Receive a layer cropping request, and locate the target satellite image data from the satellite image spatial index database according to the spatial range parameter and the target resolution parameter included in the layer cropping request.

[0158] S40. Select the corresponding target pyramid level in the multi-level image pyramid according to the target resolution parameter, obtain the to-be-processed image slice data in the target pyramid level, and allocate the to-be-processed image slice data to the target memory pool level corresponding to the target pyramid level in the multi-level image memory pool.

[0159] S50. Perform atmospheric correction processing on the to-be-processed image slice data in the target memory pool level by using a radiative transfer equation set to obtain corrected image slice data.

[0160] S60. Perform dynamic stitching processing on the corrected image slice data in the target memory pool level, and perform edge feathering processing by using a weight calculation equation set.

[0161] S70. Perform projection transformation processing on the stitched image data according to the final pixel mixing weight value output by the weight calculation equation set and the apparent radiance value output by the radiative transfer equation set.

[0162] S80. Crop the image data after projection transformation according to the spatial range parameter;

[0163] S90. Output the image data after cropping as the target layer data.

[0164] The specific implementation manners of the above steps are described in detail below:

[0165] The specific implementation manner of step S10 is as follows: First, establish a spatial partition index table in the satellite image spatial index database. Specifically, this step includes: Step 101, establish a multi-level spatial grid according to geographical location coordinates and establish a spatial partition index table in the satellite image spatial index database. This can better perform spatial management and retrieval on the satellite image data to be processed. Next, in step 102, collect the original data of solar zenith angle, extraterrestrial irradiance, atmospheric optical thickness, scattering phase function, and surface reflectance in the satellite image data to be processed. These original data are the basis for subsequent image processing. Then, in step 103, perform spatial partition storage on the satellite image data to be processed and establish a spatial partition table. This can improve the efficiency of data retrieval. Finally, in step 104, update the satellite image spatial index database according to the multi-level spatial grid and the spatial partition table. This step aims to establish an efficient satellite image spatial index system.

[0166] The specific implementation manner of step S20 is as follows: First, in step 201, determine the number of levels of the multi-level image pyramid and establish the multi-level image memory pool. This prepares for subsequent multi-scale processing. Next, in step 202, calculate the resolution of each level according to the level number calculation equation. This can ensure that the resolution relationship between different levels is reasonable. Then, in step 203, allocate memory space for each level of the multi-level image memory pool based on the memory capacity calculation equation. This can better manage memory resources. Next, in step 204, optimize the configuration of the multi-level image memory pool according to the memory allocation optimization equation. This step aims to improve the memory utilization efficiency. Then, in step 205, initialize the multi-level image memory pool using the load balancing equation. This can ensure that the memory pool is evenly loaded in the initial state. Finally, in step 206, establish a dynamic management mechanism for the multi-level image memory pool based on the memory dynamic adjustment equation. This can cope with changes in the data access pattern.

[0167] The specific implementation of step S30 is as follows: First, in step 301, a layer clipping request is received, and the spatial range parameter in the layer clipping request is parsed. This is to determine the target area. Next, in step 302, the target area is retrieved from the satellite image spatial index database according to the spatial range parameter. This helps to quickly locate the target image data. Then, in step 303, the target resolution parameter in the layer clipping request is parsed. This provides a basis for subsequent image scaling. Then, in step 304, the layer scaling ratio is calculated according to the target resolution parameter. This can determine the resolution of the final output layer. Finally, in step 305, the target satellite image data is located from the satellite image spatial index database according to the target area and the layer scaling ratio. This step realizes the quick location of the target image data.

[0168] The specific implementation of step S40 is as follows: First, in step 401, the target pyramid level is calculated according to the target resolution parameter. This provides a basis for subsequent data access. Next, in step 402, the target pyramid level is located in the multi-level image pyramid. This can quickly access the required image data. Then, in step 403, the image slice data to be processed is read from the target pyramid level. This prepares the data for subsequent image processing. Then, in step 404, the target memory pool level is located in the multi-level image memory pool. This can ensure that the data is allocated to the appropriate memory level. Next, in step 405, the image slice data to be processed is allocated to the target memory pool level. This can improve the data access efficiency. Finally, in step 406, the load status of the multi-level image memory pool is updated according to the load balancing equation. This helps to dynamically adjust the usage of the memory pool.

[0169] The specific implementation of step S50 is as follows: Calculate the radiation correction coefficient matrix value using the radiation transfer equation set, and perform matrix operations on the radiation correction coefficient matrix value and the data of the image slice to be processed to correct the solar radiation flux, atmospheric scattering, surface reflection, and sensor response in the data of the image slice to be processed. Modify the radiation value of the data of the image slice to be processed according to the correction result to obtain the corrected image slice data. This step aims to eliminate the radiation distortion in the image caused by factors such as atmospheric effects and surface reflection. At the same time, calculate the final pixel mixing weight value using the weight calculation equation set to prepare for subsequent stitching processing. Specifically, this step uses the solar radiation calculation equation, atmospheric scattering calculation equation, surface reflection calculation equation, sensor reception calculation equation, and radiation component decomposition calculation matrix, etc., comprehensively considering the influence of atmospheric and surface factors on the image radiation value, so as to achieve effective radiation correction. At the same time, the application of the weight calculation equation set provides a reliable weight basis for subsequent image stitching processing.

[0170] The specific implementation of step S60 is as follows: Calculate the final pixel mixing weight value using the weight calculation equation set. Specifically, this step applies the distance weight calculation equation, gradient weight calculation equation, brightness weight calculation equation, comprehensive weight calculation equation, and weight component decomposition calculation matrix, etc. Among them, the distance weight calculation equation is based on the Gaussian function, reflecting the characteristic that the weight decays with the increase of distance, which helps to maintain the smoothness of the edge transition. The gradient weight calculation equation combines local difference and global gradient information, which can suppress local noise and maintain edge features. The brightness weight calculation equation takes into account both the difference between the pixel and the local mean and the brightness consistency of adjacent images, which helps to handle the brightness jump at the seam. The comprehensive weight calculation equation adopts the form of power function combination, and the contribution degree of each component can be flexibly controlled by adjusting the exponential term. The weight component decomposition matrix decomposes the weight into stable and variable parts through linear transformation, which is convenient for separate optimization. The final weight calculation equation adopts the weighted combination form with attenuation, which can effectively suppress high-frequency changes and improve the stability of the stitching result. Generally speaking, this step makes full use of various weight calculation methods and provides a high-quality pixel mixing basis for subsequent stitching processing.

[0171] The specific implementation of step S70 is as follows: First, in step 701, the final pixel mixing weight value and the apparent radiance value are obtained. These data will provide a basis for subsequent processing. Next, in step 702, the stitched image data is subjected to brightness balance processing according to the final pixel mixing weight value. This helps to eliminate the brightness inconsistency in the image. Then, in step 703, the image data after brightness balance processing is subjected to color correction according to the apparent radiance value. This can ensure that the color performance of the image is more accurate. Then, in step 704, a projection transformation matrix is established. This matrix contains parameters such as scale transformation, shear transformation, translation transformation, and perspective transformation, and can achieve geometric correction of the image. Next, in step 705, the image data after color correction is subjected to projection transformation processing. This step can eliminate the geometric distortion in the image. Finally, in step 706, the coordinate registration of the image data after projection transformation is performed. This can ensure that the geographical location information in the image is consistent with the actual situation. Generally speaking, this step realizes the processing of brightness balance, color correction, geometric correction, and coordinate registration of the image, and prepares for subsequent cropping and output.

[0172] The specific implementation of step S80 is as follows: First, in step 801, the cropping boundary is determined according to the spatial range parameter. This provides a basis for subsequent cropping processing. Next, in step 802, the boundary check of the image data after projection transformation is performed. This helps to identify the abnormal areas at the image edges. Then, in step 803, the image data after projection transformation is cropped according to the cropping boundary. This can obtain the target area image required. Then, in step 804, the edge area after cropping is smoothed. This can eliminate the jagged effect caused by cropping. Next, in step 805, the cropped image data is resampled and the boundary is optimized according to the target resolution parameter, and the image data after cropping processing is obtained. This step can ensure that the output image meets the required resolution requirements. Generally speaking, this step realizes the processing of image cropping, edge smoothing, and resolution resampling, and prepares for the final output.

[0173] The specific implementation of step S90 is as follows: First, in step 901, the cropped image data is subjected to format conversion. This can ensure that the image data meets the requirements of a common output format. Next, in step 902, target layer metadata is generated. This metadata contains key attributes such as the geographical information, resolution, and bands of the image. Then, in step 903, the target layer metadata is associated with the cropped image data. This can ensure the consistency between the metadata and the image data. Subsequently, in step 904, the associated data is organized into target layer data. This step realizes the packaged output of the image data and the metadata. Next, in step 905, the target layer data is compressed and encoded. This can reduce the volume of the output data and improve the transmission and storage efficiency. Finally, in step 906, the compressed and encoded target layer data is output and saved. In this way, the entire image processing process is completed, and the required target layer data is obtained. Generally speaking, this step realizes processing such as format conversion, metadata generation, data packaging, and compression encoding of the image data, ensuring the quality and efficiency of the final output.

[0174] The second aspect of the present invention provides a computer-readable storage medium, wherein program instructions are stored in the computer-readable storage medium, and when the program instructions run on a computer, they are used to execute the method for generating layers by online satellite image cropping described above.

[0175] The third aspect of the present invention provides a system for generating layers by online satellite image cropping, which includes the above-mentioned computer-readable storage medium.

[0176] Specifically, the principle of the present invention is: making full use of the spatio-temporal characteristics of satellite image data, adopting a multi-level data organization and processing mechanism to achieve efficient management and rapid response to massive raw image data.

[0177] First, the established satellite image spatial index database stores the image data in partitions through multi-level spatial grids, greatly improving the efficiency of data retrieval. At the same time, dynamically maintaining the spatial partition table can quickly locate and obtain the image data of the target area. This lays a good foundation for subsequent image processing.

[0178] Secondly, by adopting the method of multi-level image pyramids and dynamic memory pools, rapid access and efficient caching of the raw image data are realized. By reasonably dividing the pyramid levels, both different resolution requirements are met and data redundancy is reduced. The dynamic memory management mechanism ensures that frequently accessed data can be kept in memory, improving the processing efficiency. This hierarchical data organization and adaptive caching mechanism are the key to the high-performance implementation of the present invention.

[0179] In terms of image preprocessing, the present invention adopts a comprehensive correction model based on the radiative transfer theory, including solar radiation calculation, atmospheric scattering calculation, surface reflectance calculation, sensor response calculation, etc., effectively eliminating various radiation distortions. At the same time, aiming at the problems of inconsistent brightness and discontinuous edges during the stitching process, an adaptive pixel blending weighted algorithm is proposed, which fully utilizes multi-faceted information such as distance, gradient, and brightness to generate a smooth and natural stitching result. In addition, geometric correction adopts an optimized projection transformation and control point registration method, which can accurately eliminate the geometric distortion of the image. The highly integrated preprocessing technologies ensure the radiative and geometric accuracy of the final output image.

[0180] Finally, in response to the requirements of GIS applications, the present invention implements processing such as cropping, resolution resampling, format conversion, and compression coding of image data. Among them, the cropping and edge smoothing algorithms ensure the clear boundary of the output image; the resolution resampling and optimized coding technologies ensure that the output image still maintains high visual quality with limited data volume. Generally speaking, this method fully meets the user's needs for customized and high-quality satellite image data, providing strong support for various spatial information services.

[0181] A specific Embodiment 1 of the present invention is provided below. The specific implementation manners of each step in this Embodiment 1 are described in detail as follows: The specific implementation manner of step S10 is: First, establish a spatial partition index table in the satellite image spatial index database. Specifically, this step includes: Step 101, establish a multi-level spatial grid according to geographical location coordinates and establish a spatial partition index table in the satellite image spatial index database. This can better perform spatial management and retrieval of the satellite image data to be processed. Next, in step 102, collect the original data of the solar zenith angle, the extraterrestrial irradiance, the atmospheric optical thickness, the scattering phase function, and the surface reflectance in the satellite image data to be processed. These original data are the basis for subsequent image processing. Then, in step 103, perform spatial partition storage on the satellite image data to be processed and establish a spatial partition table. This can improve the efficiency of data retrieval. Finally, in step 104, update the satellite image spatial index database according to the multi-level spatial grid and the spatial partition table. This step aims to establish an efficient satellite image spatial index system.

[0182] The specific implementation of step S20 is as follows: First, in step 201, determine the number of levels of the multi-level image pyramid and establish the multi-level image memory pool. This prepares for subsequent multi-scale processing. Next, in step 202, calculate the resolution of each level according to the level number calculation equation. This ensures that the resolution relationship between different levels is reasonable. Then, in step 203, allocate memory space for each level of the multi-level image memory pool based on the memory capacity calculation equation. This can better manage memory resources. Next, in step 204, optimize the configuration of the multi-level image memory pool according to the memory allocation optimization equation. This step aims to improve the memory utilization efficiency. Then, in step 205, initialize the multi-level image memory pool using the load balancing equation. This ensures that the memory pool is evenly loaded in the initial state. Finally, in step 206, establish a dynamic management mechanism for the multi-level image memory pool based on the memory dynamic adjustment equation. This can cope with changes in the data access pattern.

[0183] The specific implementation of step S30 is as follows: First, in step 301, receive a layer cropping request and parse the spatial range parameter in the layer cropping request. This is to determine the target area. Next, in step 302, retrieve the target area from the satellite image spatial index database according to the spatial range parameter. This helps to quickly locate the target image data. Then, in step 303, parse the target resolution parameter in the layer cropping request. This provides a basis for subsequent image scaling. Then, in step 304, calculate the layer scaling ratio according to the target resolution parameter. This can determine the resolution of the final output layer. Finally, in step 305, locate the target satellite image data from the satellite image spatial index database according to the target area and the layer scaling ratio. This step realizes the rapid location of the target image data.

[0184] The specific implementation of step S40 is as follows: First, in step 401, the target pyramid level is calculated according to the target resolution parameter. This provides a basis for subsequent data access. Next, in step 402, the target pyramid level is located in the multi-level image pyramid. This enables quick access to the required image data. Then, in step 403, the image slice data to be processed is read from the target pyramid level. This prepares the data for subsequent image processing. After that, in step 404, the target memory pool level is located in the multi-level image memory pool. This ensures that data is allocated to the appropriate memory level. Next, in step 405, the image slice data to be processed is allocated to the target memory pool level. This improves data access efficiency. Finally, in step 406, the load status of the multi-level image memory pool is updated according to the load balancing equation. This helps to dynamically adjust the usage of the memory pool.

[0185] The specific implementation of step S50 is as follows: The radiative correction coefficient matrix value is calculated using the radiative transfer equation set, and matrix operations are performed on the radiative correction coefficient matrix value and the image slice data to be processed to correct the solar radiation flux, atmospheric scattering, surface reflection, and sensor response in the image slice data to be processed. The radiation value of the image slice data to be processed is corrected according to the correction result to obtain the corrected image slice data. This step aims to eliminate the radiation distortion in the image caused by factors such as atmospheric effects and surface reflection. At the same time, the final pixel mixing weight value is calculated using the weight calculation equation set to prepare for subsequent stitching processing. Specifically, this step uses the solar radiation calculation equation, atmospheric scattering calculation equation, surface reflection calculation equation, sensor reception calculation equation, and radiation component decomposition calculation matrix, etc., comprehensively considering the influence of atmospheric and surface factors on the image radiation value, so as to achieve effective radiative correction. At the same time, the application of the weight calculation equation set provides a reliable weight basis for subsequent image stitching processing.

[0186] The specific implementation of step S60 is as follows: calculating the final pixel blending weight value using the weight calculation equations. Specifically, this step applies the distance weight calculation equation, gradient weight calculation equation, brightness weight calculation equation, comprehensive weight calculation equation, and weight component decomposition calculation matrix, etc. Among them, the distance weight calculation equation is based on the Gaussian function, reflecting the characteristic that the weight decays as the distance increases, which helps to maintain the smoothness of the edge transition. The gradient weight calculation equation combines local difference and global gradient information, which can suppress local noise and maintain edge features. The brightness weight calculation equation takes into account both the difference between the pixel and the local mean and the brightness consistency of adjacent images, which helps to handle the brightness jump at the seam. The comprehensive weight calculation equation adopts a power function combination form, and the contribution degree of each component can be flexibly controlled by adjusting the exponential term. The weight component decomposition matrix decomposes the weight into stable and variable parts through linear transformation, which is convenient for separate optimization. The final weight calculation equation adopts a weighted combination form with attenuation, which can effectively suppress high-frequency changes and improve the stability of the stitching result. Generally speaking, this step makes full use of various weight calculation methods, providing a high-quality pixel blending basis for subsequent stitching processing.

[0187] The specific implementation of step S70 is as follows: First, in step 701, obtain the final pixel blending weight value and the apparent radiance value. These data will provide a basis for subsequent processing. Next, in step 702, perform brightness balance processing on the stitched image data according to the final pixel blending weight value. This helps to eliminate the brightness inconsistency in the image. Then, in step 703, perform color correction on the image data that has undergone brightness balance processing according to the apparent radiance value. This can ensure that the color performance of the image is more accurate. Then, in step 704, establish a projection transformation matrix. This matrix contains parameters such as scale transformation, shear transformation, translation transformation, and perspective transformation, which can achieve geometric correction of the image. Next, in step 705, perform projection transformation processing on the image data that has undergone color correction. This step can eliminate the geometric distortion in the image. Finally, in step 706, perform coordinate registration on the image data after projection transformation. This can ensure that the geographical location information in the image is consistent with the actual situation. Generally speaking, this step realizes the processing of brightness balance, color correction, geometric correction, and coordinate registration of the image, preparing for subsequent cropping and output.

[0188] The specific implementation of step S80 is as follows: First, in step 801, the cropping boundary is determined according to the spatial range parameter, which provides a basis for subsequent cropping processing. Next, in step 802, boundary checking is performed on the image data after projective transformation, which helps to identify abnormal areas at the image edges. Then, in step 803, the image data after projective transformation is cropped according to the cropping boundary, and the desired target area image can be obtained. Then, in step 804, smoothing processing is performed on the cropped edge area to eliminate the sawtooth effect caused by cropping. Next, in step 805, the cropped image data is resampled and boundary optimized according to the target resolution parameter to obtain the image data after cropping processing. This step can ensure that the output image meets the required resolution requirements. Generally speaking, this step realizes processing such as image cropping, edge smoothing, and resolution resampling, preparing for the final output.

[0189] The specific implementation of step S90 is as follows: First, in step 901, format conversion is performed on the image data after cropping processing to ensure that the image data meets the requirements of a common output format. Next, in step 902, target layer metadata is generated. These metadata contain key attributes such as the geographic information, resolution, and bands of the image. Then, in step 903, the target layer metadata is associated with the image data after cropping processing to ensure the consistency between the metadata and the image data. Then, in step 904, the associated data is organized into target layer data. This step realizes the packaged output of the image data and the metadata. Next, in step 905, compression encoding is performed on the target layer data to reduce the volume of the output data and improve the transmission and storage efficiency. Finally, in step 906, the target layer data after compression encoding is output and saved. In this way, the entire image processing process is completed, and the required target layer data is obtained. Generally speaking, this step realizes processing such as format conversion of image data, metadata generation, data packaging, and compression encoding, ensuring the quality and efficiency of the final output.

[0190] To better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: A certain GIS center is responsible for spatial data management and geographic information services within the city. The center owns multiple latest-generation high-resolution remote sensing satellites and obtains a large amount of satellite image data every day, covering urban areas, suburban areas, and some mountainous and rural areas. These image data not only provide basic data support for urban and rural planning, environmental monitoring, etc., but also provide online map services and spatial query functions for citizens.

[0191] To effectively manage and efficiently utilize these massive satellite image data, the municipal government GIS center has introduced the method of online satellite image cropping to generate layers proposed in the present invention. This method mainly includes the following steps:

[0192] The first step is to establish a satellite image spatial index database. The specific approach is as follows:

[0193] Firstly, according to the geographical location coordinates of the image data, a 5-level spatial grid index is established using the quadtree algorithm. As shown in Table 1, the spatial range and numbering information of each grid cell at each level are recorded in the index table, laying a foundation for subsequent data retrieval.

[0194] Table 1 Satellite Image Spatial Grid Index Table

[0195]

[0196] Next, the technical staff of the GIS center collected high-resolution satellite image data of this area in the past year through the ground station receiving equipment and preprocessed it. Specifically, it includes:

[0197] 1) Extract the original data of solar zenith angle, extraterrestrial irradiance, atmospheric optical thickness, scattering phase function, and surface reflectance in the image data to prepare for subsequent radiation correction.

[0198] 2) Partition and store the preprocessed image data according to the spatial grid, and establish a corresponding spatial partition table to record information such as the image data range, data volume, and data format within each grid cell, providing a basis for quickly retrieving image data in the target area.

[0199] 3) Finally, update all the above spatial grid information and spatial partition information to the satellite image spatial index database to facilitate subsequent rapid positioning and access to the required image data.

[0200] The second step is to establish a multi-level image pyramid and a dynamic memory pool. Specifically as follows:

[0201] 1) According to experience, determine that the number of levels of the multi-level image pyramid and the memory pool is 8 levels, that is . According to the calculation formula of , take meters (original resolution), meters (minimum resolution), and we can get . This can meet various application requirements from high resolution to low resolution.

[0202] 2) For each layer of pyramid image, calculate its resolution , as shown in Table 2. This can ensure that the resolution relationship between different levels is reasonable.

[0203] Table 2 Resolution of Multi-level Image Pyramid

[0204]

[0205] 3) Based on the multi-level image pyramid, a corresponding dynamic memory pool was established. First, the initial capacity was allocated to each layer of the memory pool according to the formula , where MB, . This can ensure that the memory capacity of the lower layer is larger and the memory capacity of the upper layer gradually decreases.

[0206] 4) Considering the differences in access frequency and data density of data at different levels, the optimized formula was used to dynamically adjust the memory capacity, where and were respectively set to . This can ensure that the frequently accessed high-resolution data can be kept in memory.

[0207] 5) During initialization, the load index of each layer of the memory pool was calculated according to the load balancing equation , where were respectively set to . This can ensure that the load of the memory pool is balanced in the initial state.

[0208] 6) Finally, a dynamic memory management mechanism was established. When the load of a certain layer of the memory pool is too high, dynamic adjustment is carried out according to to supplement or release memory resources to maintain the balance of the overall load.

[0209] Through the above steps, the GIS center established an efficient multi-level image data organization and dynamic memory management mechanism, laying a foundation for subsequent fast image processing.

[0210] The third step is to receive the user's layer clipping request and locate the target image data. The specific process is as follows:

[0211] 1) A certain urban planning department submitted a layer clipping request to the GIS center, requesting a 1:5000 scale vector base map within a certain area. The request included the spatial range parameters of the target area, such as the upper left coordinate (108.25°, 22.25°) and the lower right coordinate (108.75°, 22.00°).

[0212] 2) The processing system in the GIS center first parsed the spatial range parameters in the request and quickly located 3 grid cells covering the area: 1-1-1-1-2, 1-1-1-1-3, and 1-1-1-1-4 based on the satellite image spatial grid index established previously.

[0213] 3) At the same time, the system also parsed the target resolution parameter in the user request, that is, a scale of 1:5000, corresponding to a field resolution of approximately 5 meters. Based on this parameter, the required layer scaling ratio was calculated as .

[0214] 4) Finally, based on the above target area grid cell information and scaling ratio, the system located the original image data covering the area from the satellite image spatial index database, preparing for subsequent image processing.

[0215] Step 4: Read and process the target image data from the multi-level image pyramid and dynamic memory pool. Specifically as follows:

[0216] 1) First, based on the target resolution of 5 meters requested by the user, the system calculated the corresponding pyramid level as the 5th level.

[0217] 2) In the multi-level image pyramid, 3 target image slices at the 5th level were located, corresponding to the aforementioned 3 grid cells respectively.

[0218] 3) Then, the system located 3 memory pool layers corresponding to the 5th level pyramid image in the dynamic memory pool. If it was found that some slice data had not been cached in memory, it would be automatically loaded from the hard disk into the corresponding memory layer.

[0219] 4) Next, the system performed radiometric correction on the image slice data in these 3 memory pool layers using the previously established radiative transfer equation set. Specifically including:

[0220] Using the solar radiation calculation equation To calculate the incident radiation flux value; using the atmospheric scattering calculation equation To calculate the atmospheric path radiation value; using the surface reflection calculation equation To calculate the surface reflected radiation value; using the sensor reception calculation equation To calculate the apparent radiance value; using the radiation component decomposition calculation matrix to decompose the apparent radiance value into a stable component and a variable component, obtaining the radiometric correction coefficient matrix.

[0221] 5) Then, the system performed adaptive stitching processing on the image slice data in these 3 memory pool layers using the previously established weight calculation equation set. Specifically including:

[0222] Using the distance weight calculation equation Calculate the initial distance weight; use the gradient weight calculation equation Calculate the initial gradient weight; use the brightness weight calculation equation Calculate the initial brightness weight; use the comprehensive weight calculation equation Calculate the initial pixel mixing weight; use the weight component decomposition calculation matrix to decompose the initial weight into a stable component and a variable component, and finally according to Calculate the final pixel mixing weight.

[0223] Through the above radiation correction and pixel stitching processing, the system generates a high-quality image data covering the target area.

[0224] Step 5: Perform geometric correction and cropping processing on the stitched image data. The specific steps are as follows:

[0225] 1) First, the system determines the 4 vertex coordinates of the cropping area according to the spatial range parameters requested by the user: (108.25°, 22.25°), (108.25°, 22.00°), (108.75°, 22.25°), (108.75°, 22.00°). And establish a boundary indication function to identify the cropping area.

[0226] 2) Then, the system establishes a projection transformation matrix , where , to achieve the scale transformation and shear transformation of the image. Next, perform projection transformation on the stitched image data, , where , and the final coordinates are .

[0227] 3) To further optimize the projection result, the system uses 4 ground control points, (108.25°, 22.20°), (108.25°, 22.05°), (108.70°, 22.25°), (108.70°, 22.00°), and fine-tunes the projection transformation matrix parameters through an iterative method, making the reprojection error of the control point coordinates less than 0.5 meters.

[0228] 4) Finally, the system crops the image data after projection transformation according to the previously determined cropping area boundary . At the same time, apply Gaussian smoothing filtering to the cropped edge area, , to eliminate the jagged effect caused by cropping.

[0229] Through the above geometric correction and cropping processing, the system generates a 1:5000 scale vector base map layer that covers the user-specified area and has good geometric accuracy and edge smoothness. As Figure 2As shown, a schematic diagram of the control point distribution and the reprojection error range is presented.

[0230] Step 6: Perform format conversion, metadata generation, and compression encoding on the final layer data. Specifically as follows:

[0231] 1) First, the system converts the cropped image data into a standard GIS raster data format, such as GeoTIFF. This ensures that the data can be recognized and used by common GIS software.

[0232] 2) Then, the system generates corresponding metadata based on information such as the geographic extent, resolution, and bands of the image, and associates it with the image data to form a complete layer data packet.

[0233] 3) Finally, the system performs lossy compression encoding on the layer data. The specific algorithm used is: , where is the compressed image, is the original image, is the transform domain coefficient. Through this sparse optimization encoding method, the volume of the layer data can be compressed to about 30% of the original, while the image quality index can still reach a relatively high level.

[0234] At this point, the GIS center has successfully generated a vector base map layer at a scale of 1:5000 for the urban planning department. The layer data has undergone effective management, rapid preprocessing, and customized output. It not only has a small volume and high transmission efficiency, but also meets high standards in terms of geometric accuracy, radiometric correction, visual effects, etc., fully meeting the user's requirements.

[0235] In addition, the GIS center can also flexibly generate customized layer data of various scales and coverage ranges according to user needs for different application scenarios, such as environmental monitoring, urban planning, emergency response, etc., giving full play to the advantages of the method of the present invention.

[0236] It should be noted that the variable explanations involved in the present invention are shown in Tables 3 and 4.

[0237] Table 3 Variable Explanation Table

[0238]

[0239] Table 4 Variable Explanation Table - 2

[0240]

[0241] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for generating layers by cropping online satellite images, characterized in that, Including: By establishing a satellite image spatial index database, multi-level image pyramids, and multi-level image memory pools, the satellite image data to be processed is partitioned and hierarchically managed. Image slice data to be processed is obtained from the target pyramid level based on the parameters in the layer clipping request and assigned to the corresponding target memory pool level. The radiative transfer equation set is used for atmospheric correction processing, and the weight calculation equation set is used for edge feathering processing to finally generate the target layer data; The equation sets used for establishing and managing the multi-level image memory pools include a level number calculation equation, a memory capacity calculation equation for each level, a memory allocation optimization equation, a memory load balancing equation, and a memory dynamic adjustment equation; among them, the specific steps for establishing the multi-level image pyramids and multi-level image memory pools are: generating multi-level image pyramids for the satellite image data to be processed according to the preset image scaling levels, storing the multi-level image pyramids in the system cache, and simultaneously establishing multi-level image memory pools. The number of levels of the multi-level image memory pools corresponds one-to-one with the number of levels of the multi-level image pyramids; Among them, the level number calculation equation: ; In the formula, is the number of levels of the multi-level image memory pool; is the original image resolution; is the target minimum resolution; is the level redundancy term; represents the ceiling operation; The memory capacity calculation equation for each level: ; Wherein, is the capacity of the -th layer memory pool; is the memory capacity of the reference layer; is the capacity attenuation coefficient; is the memory redundancy of the -th layer; The memory allocation optimization equation: ; Wherein, is the actually allocated memory amount of the th layer; is the access frequency factor of the th layer; is the data density factor of the th layer; is the total available memory amount of the system; is the memory allocation error term; The memory load balancing equation: ; In the formula, is the load index of the th layer; is the currently used memory amount of the th layer; is the data access time of the th layer; is the maximum allowed access time; is the cache queue length of the th layer; is the maximum allowed queue length; is the balance coefficient, and satisfies ; is the load calculation error term; The memory dynamic adjustment equation: ; Wherein, is the memory adjustment amount of the th layer; is the system average load index; is the adjustment coefficient; is the load change rate; is the adjustment amount error term.

2. The method for generating a layer by cropping online satellite images according to claim 1, characterized in that The specific method for establishing the satellite image spatial index database is: partitioning and storing the satellite image data to be processed, where the satellite image data to be processed includes original data of solar zenith angle, extraterrestrial irradiance, atmospheric optical thickness, scattering phase function, and surface reflectance.

3. The method for generating a layer by cropping online satellite images according to claim 2, wherein, The parameters based on the layer clipping request are specifically: receiving the layer clipping request, and locating the target satellite image data from the satellite image spatial index database according to the spatial range parameters and target resolution parameters included in the layer clipping request.

4. A method for generating a layer by cropping online satellite images according to claim 3, characterized in that, The specific method for obtaining the image slice data to be processed from the target pyramid level is: selecting the corresponding target pyramid level in the multi-level image pyramid according to the target resolution parameter, obtaining the image slice data to be processed in the target pyramid level, and assigning the image slice data to be processed to the corresponding target memory pool level in the multi-level image memory pool that corresponds to the target pyramid level.

5. The method for generating a layer by cropping online satellite images according to claim 4, characterized in that, The radiative transfer equation set includes a solar radiation calculation equation, an atmospheric scattering calculation equation, a surface reflection calculation equation, a sensor reception calculation equation, and a radiation component decomposition calculation matrix, which are used to perform atmospheric correction processing on the image slice data to be processed in the target memory pool level to obtain the corrected image slice data.

6. A method for generating a layer by cropping online satellite images according to claim 5, characterized in that, The weight calculation equation set includes a distance weight calculation equation, a gradient weight calculation equation, a brightness weight calculation equation, a comprehensive weight calculation equation, and a weight component decomposition calculation matrix, which are used to perform dynamic stitching processing and execute edge feathering processing on the corrected image slice data in the target memory pool level.

7. A method for generating a layer by cropping online satellite images according to claim 6, characterized in that, Specifically, the generation of the target layer data is as follows: the spliced image data is subjected to a projection transformation process according to the final pixel mixing weight value output by the weight calculation equation set and the apparent radiance value output by the radiative transfer equation set, the image data after the projection transformation is subjected to a cropping process according to the spatial range parameter, and the image data after the cropping process is output as the target layer data.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions that, when running on a computer, are used to execute the method for online satellite image cropping to generate a layer according to any one of claims 1-7.

9. An online satellite image cropping and generating layer system, characterized in that, It includes the computer-readable storage medium described in claim 8.

Images