Regional hyperspectral remote sensing image mosaic method, system, equipment and medium

Abstract

The invention relates to a regional hyperspectral remote sensing image mosaic method, system and device and a medium, and belongs to the technical field of remote sensing image processing. The method comprises the following steps: acquiring a geographic range and an image spatial resolution of a target area, and calculating a row number and a column number of an image of the target area and a geographic reference transformation parameter; calculating a geographic center coordinate of each pixel of the target area image to obtain a center coordinate set; retrieving a hyperspectral image file covering the central coordinate set from a preset data pool, and extracting candidate spectral data of corresponding coordinates to form a spectral set; calculating the goodness of fit of the spectrum dominant mineral corresponding to each spectrum in the spectrum set to obtain a goodness of fit array; selecting candidate spectral data corresponding to the maximum goodness of fit, and assigning the candidate spectral data as current pixel data of the target area image; and generating a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameters and the assigned target region image. The method can improve the overall quality of the remote sensing image and the mineral recognition precision.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image processing, in particular to a regional hyperspectral remote sensing image mosaicking method, system, device and medium. BACKGROUND

[0002] Hyperspectral remote sensing image has become a core technology for modern geological survey and mineral resource assessment due to its nanometer-level spectral resolution capability. This technology captures the reflection characteristics of surface materials in hundreds of continuous narrow bands, constructs a unique spectral "fingerprint", and makes the accuracy of mineral species identification and spatial distribution mapping reach an unprecedented level. With the development of hyperspectral remote sensing technology, the amount of image data obtained is increasing, and how to effectively splice these data into an image with high spectral consistency and high geographical accuracy has become an important technical challenge.

[0003] At present, most of the hyperspectral image splicing methods adopt simple techniques based on geometric registration and resampling, and usually rely on spatial matching of images to complete image splicing. However, in actual application, the spectral feature consistency of hyperspectral data is mostly ignored, and the spliced image may have inconsistent spectra. Moreover, the existing image mosaicking method does not fully utilize mineral spectral features and existing spectral databases, resulting in a large room for improvement in mineral identification accuracy and application effect of the final spliced image data. SUMMARY

[0004] In order to improve the overall image quality and mineral identification accuracy, the present application provides a regional hyperspectral remote sensing image mosaicking method, system, device and medium.

[0005] In a first aspect, the present application provides a regional hyperspectral remote sensing image mosaicking method, which adopts the following technical solution: A regional hyperspectral remote sensing image mosaicking method, the method comprising: obtaining the geographical range and image spatial resolution of a target region; calculating the row and column numbers and geographical reference transformation parameters of the target region image according to the geographical range and image spatial resolution; traversing each pixel position of the target region image based on the row and column numbers, and calculating the geographical center coordinates of each pixel according to the geographical reference transformation parameters to obtain a center coordinate set; retrieving hyperspectral image files covering the center coordinate set from a preset data pool, and extracting candidate spectral data of the corresponding coordinates based on the nearest neighbor method to form a spectral set; calculating the fitting goodness of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain a fitting goodness array; Traverse the goodness-of-fit array, select the candidate spectral data corresponding to the maximum goodness-of-fit, and assign it as the current pixel data of the target region image; Based on the geographic reference transformation parameters and the assigned target region image, generate a regional hyperspectral remote sensing mosaic image file.

[0006] By adopting the above technical solutions, an efficient and accurate hyperspectral remote sensing image mosaic method is constructed. The method fully considers the matching of geographic information and spectral characteristics in the image stitching process, optimizes the selection of spectral data, and improves the accuracy of mineral identification through mineral spectral fitting. The final generated hyperspectral remote sensing image has higher spectral consistency and geographic accuracy, and has important application value for geological survey, resource management and other fields.

[0007] Optionally, the step of calculating the row and column numbers of the target region image and the geographic reference transformation parameters according to the geographic range and the image spatial resolution comprises: Obtain the longitude range, latitude range and image spatial resolution of the target region; Based on the latitude range and the image spatial resolution, calculate the number of rows of the target region image, and based on the longitude range and the image spatial resolution, calculate the number of columns of the target region image; According to the minimum longitude, maximum latitude and image spatial resolution of the target region, generate an affine transformation coefficient array; the affine transformation coefficient array includes the minimum longitude, image spatial resolution, zero value, maximum latitude, zero value and negative spatial resolution.

[0008] By adopting the above technical solutions, the accurate positioning of hyperspectral remote sensing image in geographic space and the accuracy of pixel position are ensured. Through the comprehensive application of spatial resolution, latitude and longitude range and affine transformation coefficient, a stable foundation is provided for subsequent image stitching and spectral data extraction. Finally, the generated image data can effectively guarantee the spatial accuracy and spectral consistency, improve the overall quality of the image, and provide high-precision and high-reliability data support for mineral identification, geological analysis and other applications.

[0009] Optionally, the step of traversing each pixel position of the target region image based on the row and column numbers and calculating the geographic center coordinates of each pixel according to the geographic reference transformation parameters comprises: Based on the row and column numbers, traverse each pixel position of the target region image in row number and column number order; For each traversed pixel position (row, col), calculate the longitude coordinate and latitude coordinate of the geographic center of each pixel according to the affine transformation coefficient array, and the calculation formula is: lon_geo = lon_min + col * r; lat_geo = lat_max - row * r; In the above formula, lon_geo is the longitude coordinate, lon_min is the minimum longitude value, lat_geo is the latitude coordinate, lat_max is the maximum latitude value, and r is the spatial resolution.

[0010] By adopting the technical scheme, the geographical center coordinates of each pixel in the target area image are accurately calculated, and the problem that the pixel position in the remote sensing image is inconsistent with the geographical coordinate system is effectively solved. The pixel row and column coordinates of the image are converted into actual geographical coordinates, and the calculation process improves the image stitching accuracy, especially for remote sensing images that need high-precision geographical positioning, and ensures the geographical consistency of image stitching and the accuracy of subsequent data analysis.

[0011] Optionally, the step of retrieving a hyperspectral image file covering the center coordinate set from a preset data pool and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set includes: Creating a database table to store image file paths and corresponding geographical boundary geometric data; Traversing the hyperspectral image files in the preset data pool, parsing the geographical coordinate system and the geographical coordinates of the four corner points for each image, and storing the file path and the polygon geographical boundary generated based on the four corner point coordinates in the database table; Based on the input center coordinate set, performing a spatial relationship query in the database table to filter out a list of hyperspectral image files covering the geographical center coordinates of each pixel; Based on each image in the hyperspectral image file list, calculating the nearest neighbor pixel position in the image according to the geographical center coordinates of each pixel, and extracting the spectral data of the nearest neighbor pixel as candidate spectral data; Summarizing all extracted candidate spectral data to obtain a spectral set.

[0012] By adopting the technical scheme, a database is established to manage the paths and geographical boundary information of the image files, and the image files covering the target pixel position are accurately filtered out through spatial query, ensuring the accuracy and consistency of the image data. On this basis, the nearest neighbor pixel position of each target pixel is calculated and the spectral data is extracted, and finally a rich spectral set is formed. This method improves the accuracy and efficiency of spectral data extraction, and can provide strong support for subsequent high-precision mineral identification and resource assessment.

[0013] Optionally, the step of calculating the fitting goodness of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain a fitting goodness array includes: Traversing each spectrum in the spectral set; calculating a dominant mineral fitting degree of each spectrum based on a mineral spectral feature analysis algorithm; calculating a characteristic absorption depth of the dominant mineral; generating a fitting goodness value of each spectrum through a weighted product operation of the fitting degree and the characteristic absorption depth; combining the fitting goodness values of each spectrum in traversal order to form a fitting goodness array.

[0014] By adopting the above technical solution, the mineral fitting degree and absorption depth are combined to provide a comprehensive evaluation index for each spectrum. This method not only improves the matching degree of the spectrum and the mineral, but also reflects the absorption characteristics of the mineral characteristics, thereby optimizing the accuracy of mineral identification. By forming the fitting goodness array, subsequent spectrum screening and analysis become more efficient, providing precise support for geological exploration, mineral resource evaluation, and other fields.

[0015] Optionally, the step of traversing the fitting goodness array, selecting the spectrum corresponding to the maximum fitting goodness, and assigning it as the current pixel data of the target region image includes: traversing the fitting goodness array to identify and extract the maximum fitting goodness value therein; extracting the corresponding candidate spectrum data as optimal spectrum data from the spectrum set according to the maximum fitting goodness value; assigning the extracted optimal spectrum data to the specified position of the current pixel in the target region image data storage structure.

[0016] By adopting the above technical solution, the system traverses the fitting goodness array to identify and extract the maximum fitting goodness value, ensuring that the most matching spectrum data is selected for the target region. Then, the system extracts the corresponding optimal spectrum from the candidate spectrum set according to the fitting goodness value, and finally accurately assigns it to the corresponding position in the target image. This technical solution improves the accuracy of spectrum data selection and the quality of image stitching, ensuring that the generated regional hyperspectral remote sensing mosaic image has higher spectral consistency and spatial precision, providing reliable data support for subsequent mineral identification and resource evaluation applications.

[0017] Optionally, based on the geographic reference transformation parameters and the assigned target region image, the step of generating a regional hyperspectral remote sensing mosaic image file includes: receiving geographic reference transformation parameters and assigned target region image data; setting the affine transformation coefficients of the image file based on the geographic reference transformation parameters, and specifying the coordinate system as the pre-set geographic coordinate system; writing the target region image data into the file storage structure to generate a regional hyperspectral remote sensing mosaic image file.

[0018] By adopting the above technical solution, the geographic reference transformation parameters and the target image data are received and processed, ensuring that the image data is ready for geographic mapping. Then, the affine transformation coefficients are set based on the geographic reference transformation parameters, and a standard geographic coordinate system is assigned to the image, ensuring that the image can be accurately mapped to the actual geographic location. Finally, the system writes the image data into a file storage structure, generating a remote sensing image file with accurate geographic location information. This process ensures the spatial consistency of the image data and provides reliable data support for subsequent remote sensing analysis, mineral identification and geological exploration.

[0019] In a second aspect, the present application provides a regional hyperspectral remote sensing image mosaic system, which adopts the following technical solution: A regional hyperspectral remote sensing image mosaic system, the system comprises: An acquisition module for acquiring the geographic range and image spatial resolution of a target region; A parameter calculation module for calculating the number of rows and columns of the target region image and the geographic reference transformation parameters based on the geographic range and image spatial resolution; A center coordinate calculation module for traversing each pixel position of the target region image based on the number of rows and columns, and calculating the geographic center coordinates of each pixel based on the geographic reference transformation parameters to obtain a center coordinate set; A spectrum extraction module for retrieving a hyperspectral image file covering the center coordinate set from a preset data pool, and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set; A goodness of fit calculation module for calculating the goodness of fit of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain a goodness of fit array; A pixel assignment module for traversing the goodness of fit array, selecting the candidate spectral data corresponding to the maximum goodness of fit, and assigning it as the current pixel data of the target region image; An image file generation module for generating a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameters and the target region image after assignment.

[0020] In a third aspect, the present application provides a computer device, which adopts the following technical solution: A computer device comprising a memory, a processor and a computer program stored in the memory, the processor executing the computer program to implement the steps of the method according to the first aspect.

[0021] In a fourth aspect, the present application provides a computer readable storage medium, which adopts the following technical solution: A computer readable storage medium storing a computer program capable of being loaded and executed by a processor to perform any one of the methods according to the first aspect.

[0022] In summary, the present application includes at least one of the following beneficial technical effects: by accurately calculating the geographical range and spatial resolution of the target area, the accurate positioning of the image data is ensured. The system traverses each pixel of the target image, calculates the center coordinates using the georeferencing transformation, and extracts the spectral data that best matches each pixel position from the data pool. The goodness of fit of the spectrum is calculated by the Tetracorder algorithm, which optimizes the selection of the spectrum and ensures the optimal matching of the spectrum to the mineral. Finally, the spectrum with the maximum goodness of fit is assigned to the target image, ensuring that the image has high accuracy in spatial and spectral consistency. The technical solution of the present application provides a reliable data processing scheme for mineral identification and geological exploration in practical applications. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is the first flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0024] Figure 2 is the second flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0025] Figure 3 is the third flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0026] Figure 4 is the fourth flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0027] Figure 5 is the fifth flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0028] Figure 6 is the sixth flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application.

[0029] Figure 7 is the seventh flowchart of the regional hyperspectral remote sensing image mosaicking method of one of the embodiments of the present application. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical scheme and advantages of the present application clearer, the following will further describe the present application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. Figures 1-7

[0031] The embodiments of the present application disclose a regional hyperspectral remote sensing image mosaicking method. ​

[0032] Referring to Figure 1 A regional hyperspectral remote sensing image mosaic method, the method comprising: Step S101, obtaining the geographical range and image spatial resolution of the target region; Specifically, in the processing of hyperspectral remote sensing images, first, the geographical range of the target region and the spatial resolution of the image need to be obtained. The geographical range refers to the latitude and longitude range of the region covered by the image, while the spatial resolution is the actual size of the ground represented by each pixel in the image. Obtaining these two pieces of information is crucial for subsequent image stitching and analysis, as they directly affect the size, accuracy, and location of the image on the map.

[0033] In this scheme, the spatial resolution is defined as the "degree" unit in the geographic coordinate system, i.e., the range covered by a pixel in the latitude and longitude coordinate system. For example, if the spatial resolution of the image is 0.01 degrees, each pixel represents a ground range of about 1 km x 1 km (because 1 degree on the Earth's equator is approximately 111 km). The geographical range is usually represented by the latitude and longitude coordinates of the target region, assuming that the longitude range of the target region is [lon_min, lon_max] and the latitude range is [lat_min, lat_max]. These pieces of information provide a basis for subsequent coordinate calculations and image stitching.

[0034] Step S102, calculating the number of rows and columns of the target region image and the geographic reference transformation parameters according to the geographical range and the image spatial resolution; The number of rows and columns of the image refers to the number of pixels in the image, which determines the resolution and coverage area of the image. By giving the latitude and longitude range of the target region and the spatial resolution, the number of rows and columns of the image can be accurately calculated. For example, if the latitude range of the target region is 2 degrees and the spatial resolution is 0.01 degrees, the number of rows of the image should be 200. Geographical reference transformation is a mathematical mapping relationship that converts the pixel coordinates (row and column numbers) of the image into actual geographic coordinates (latitude and longitude). This mapping is achieved through an affine transformation, which usually consists of 6 parameters, including the upper left corner coordinates, pixel size, and rotation angle. In this scheme, the transformation parameter form is [lon_min, r, 0, lat_max, 0, -r], which means that the upper left corner coordinates of the image are (lon_min, lat_max), and the horizontal and vertical resolutions of each pixel are r (i.e., the spatial resolution).

[0035] Step S103, traversing each pixel position of the target region image based on the number of rows and columns, and calculating the geographic center coordinates of each pixel according to the geographical reference transformation parameters to obtain a set of center coordinates; In this step, the row and column numbers of each pixel in the image are mapped to the corresponding latitude and longitude coordinates through a geographic reference conversion formula. This process involves converting each pixel position (e.g., row and column numbers) of the target image into actual geographic coordinates, ensuring the precise location of each pixel in the image.

[0036] In this step, for each pixel to be traversed, assuming its row and column numbers are (row, col), its corresponding latitude and longitude coordinates can be calculated by the following formulas: longitude = lon_min + col * r, latitude = lat_max - row * r; Through such conversion, the center coordinates of each pixel can be obtained.

[0037] Step S104, retrieve the hyperspectral image file covering the center coordinate set from the preset data pool, and extract the candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set; In this step, the hyperspectral image file covering the coordinate range is retrieved from the preset data pool according to the pixel geographic coordinate set calculated in the previous step. The retrieval process can be assisted by a spatial database (e.g., PostGIS) to manage image data and its corresponding geographic information. For each target pixel, the nearest neighbor method is used to retrieve the hyperspectral data closest to the pixel position from the data pool, and the corresponding spectral data is extracted.

[0038] Specifically, the nearest neighbor method is a common interpolation method that determines the spectral value of a target pixel by finding the image data point closest to the pixel's coordinates. Through this method, the extracted spectral data can accurately reflect the spectral characteristics of the target pixel's location, providing accurate spectral data for subsequent mineral identification.

[0039] Step S105, calculate the fitting goodness of each spectrum in the spectral set corresponding to the spectral dominant mineral to obtain a fitting goodness array; In some embodiments, the Tetracorder algorithm can be used to calculate the fitting goodness between each spectrum and the known mineral spectral model. Fitting goodness refers to the similarity between the spectrum to be analyzed and the reference mineral spectrum, usually represented by a numerical value, and the larger the value, the higher the matching degree. The Tetracorder algorithm is a commonly used method for mineral identification, which calculates the matching degree of each spectrum with different mineral models based on the spectral characteristics of known minerals. The calculation of fitting goodness depends on the absorption characteristics of each spectrum and its corresponding mineral spectral model. When the spectrum matches the model of the mineral to a high degree, the fitting goodness value is larger. Through this method, the spectrum that best matches the target mineral characteristics can be selected from multiple candidate spectra.

[0040] Step S106, traverse the fitting goodness array, select the candidate spectral data corresponding to the maximum fitting goodness, and assign it as the current pixel data of the target area image; By traversing the goodness-of-fit array, the spectral data with the highest goodness-of-fit is selected as the spectral value of the current pixel of the target image, the purpose is to ensure that the data of each pixel is consistent with its geographical location and mineral characteristics.

[0041] It can be understood that by assigning each pixel with spectral data that best fits its geographical location and mineral characteristics, the overall quality and accuracy of the image is improved, ensuring that the mosaicked image has high spectral consistency and precision.

[0042] Step S107, based on the geographical reference transformation parameters and the assigned target area image, generate a regional hyperspectral remote sensing mosaic image file.

[0043] Wherein, after all the pixel data is assigned, the system writes the data into the image file based on the geographical reference transformation parameters to generate the final regional hyperspectral remote sensing mosaic image, this step completes the image writing process through GIS tools (such as GDAL), ensuring that the final generated image file has correct geographical information. The final generated hyperspectral remote sensing image not only accurately connects with the actual region in space, but also reflects the mineral distribution and other geological characteristics, providing reliable data support for mineral mapping, resource assessment and other fields.

[0044] In the above embodiment, an efficient and accurate hyperspectral remote sensing image mosaicking method is constructed, which fully considers the matching of geographical information and spectral characteristics in the image stitching process, optimizes the selection of spectral data, and improves the accuracy of mineral identification through mineral spectral fitting. The final generated hyperspectral remote sensing image has higher spectral consistency and geographical precision, and has important application value for geological survey, resource management and other fields.

[0045] Reference Figure 2 As an embodiment of step S102, according to the geographical range and image spatial resolution, the steps of calculating the row and column number of the target area image and the geographical reference transformation parameters include: Step S201, obtaining the longitude range, latitude range and image spatial resolution of the target area; Wherein, the longitude range is the geographical coordinate range of the target area from west to east, represented as [lon_min, lon_max], the latitude range is the geographical coordinate range of the target area from south to north, represented as [lat_min, lat_max], and the spatial resolution refers to the actual ground area covered by each pixel of the image.

[0046] Step S202, based on the latitude range and image spatial resolution, calculate the number of rows of the target area image, based on the longitude range and image spatial resolution, calculate the number of columns of the target area image; Specifically, assuming that the longitude range of the target region is [lon_min, lon_max], the latitude range is [lat_min, lat_max], and the image spatial resolution is r (in the geographic coordinate system "degree" unit), the calculation formula of the row number is row_num=int((lat_max-lat_min) / r), and the calculation formula of the column number is col_num=int((lon_max-lon_min) / r).

[0047] In step S203, an affine transformation coefficient array is generated according to the longitude minimum value, the latitude maximum value, and the image spatial resolution of the target region; the affine transformation coefficient array includes the longitude minimum value, the image spatial resolution, a zero value, the latitude maximum value, a zero value, and a negative spatial resolution.

[0048] The geographic reference transformation parameter is an affine transformation coefficient array, which is composed of 6 parameters, namely [lon_min, r, 0, lat_max, 0, -r], lon_min is the longitude coordinate of the upper left corner of the image; r is the spatial resolution of the image, which determines the geographic size of each pixel; the first 0 indicates that the horizontal rotation angle is zero because the horizontal and vertical coordinate axes of the image do not rotate; lat_max is the latitude coordinate of the upper left corner of the image; the second 0 indicates that the vertical rotation angle is zero; -r indicates the vertical spatial resolution of the image, which is usually negative because the latitude decreases with the increase of the row number.

[0049] In the above embodiment, the accuracy of the positioning of the hyperspectral remote sensing image in the geographic space and the accuracy of the pixel position are ensured, and a stable foundation is provided for subsequent image stitching and spectral data extraction by comprehensive application of the spatial resolution, the longitude and latitude range, and the affine transformation coefficient. Finally, the generated image data can effectively guarantee the spatial accuracy and spectral consistency, improve the overall quality of the image, and provide high-precision and high-reliability data support for mineral identification, geological analysis, and other applications.

[0050] Reference Figure 3 As an embodiment of step S103, the step of traversing each pixel position of the target region image based on the row and column numbers and calculating the geographic center coordinates of each pixel based on the geographic reference transformation parameter includes: In step S301, each pixel position of the target region image is traversed based on the row and column numbers in the order of row number and column number. The pixel position of the image is identified by the row number and the column number, the row number (row) represents the vertical position of the image, and the column number (col) represents the horizontal position of the image. In the case where the row and column numbers of the image are known, each pixel in the image is traversed in the order from top to bottom and from left to right.

[0051] Exemplarily, the row number and the column number are traversed from 0 to row_num-1 and from 0 to column_num-1 respectively. For example, if the image has 500 rows and 600 columns, the row number is traversed from 0 to 499, and the column number is traversed from 0 to 599.

[0052] In step S302, for each pixel position (row, col) traversed, the longitude coordinate and the latitude coordinate of the geographic center of each pixel are calculated according to the affine transformation coefficient array, and the calculation formula is: lon_geo = lon_min + col * r; lat_geo = lat_max - row * r; In the above formula, lon_geo is the longitude coordinate, lon_min is the minimum longitude value, lat_geo is the latitude coordinate, lat_max is the maximum latitude value, and r is the spatial resolution.

[0053] Specifically, the geographic center coordinates of each pixel are accurately calculated by affine transformation, ensuring that the spatial position of the image is consistent with the geographic coordinate system.

[0054] In the above embodiment, the geographic center coordinates of each pixel in the target area image are accurately calculated, effectively solving the problem that the pixel position in the remote sensing image is not consistent with the geographic coordinate system. The pixel row and column coordinates of the image are converted into actual geographic coordinates, and this calculation process improves the image stitching accuracy, especially for remote sensing images that require high-precision geographic positioning, ensuring the geographic consistency of image stitching and the accuracy of subsequent data analysis.

[0055] Reference Figure 4 As an embodiment of step S104, the step of retrieving a hyperspectral image file covering the central coordinate set from the preset data pool and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set includes: In step S401, a database table is created to store the image file path and the corresponding geographic boundary geometry data. Specifically, the database table is used to store the path of the image file and the geographic boundary geometry data associated with each image. The geographic boundary is a polygon (usually a rectangle or quadrilateral) formed by connecting the geographic coordinates of the four corner points of a single scene image. The geographic boundary of each image defines the area range covered by the image, which is crucial for subsequent spatial queries. The image file path points to the storage location of each hyperspectral image, ensuring that the image file can be quickly retrieved.

[0056] In some embodiments, a special data table can be created using the PostGIS spatial database, which contains two core fields: a file_location field and a geom field. Among them, the file_location field stores the complete physical path of the hyperspectral image file (string type); the geom field stores the image geographic boundary vector data (EPSG: 4326 coordinate system), and the geometric object is a closed polygon formed by connecting the four corner point coordinates of the image.

[0057] Step S402, traverse the hyperspectral image files in the preset data pool, parse the geographic coordinate system and the four corner point geographic coordinates for each image, and store the file path and the polygon geographic boundary generated based on the four corner point coordinates into the database table; Specifically, each hyperspectral image file in the data pool is traversed to extract the geographic coordinate system information and the geographic coordinates of the four corner points. Through these coordinates, the geographic boundary polygon of the image can be generated, and the file path and the geographic boundary are stored in the previously created database table. This process ensures that each image file is correctly associated with its geographic information (i.e. four corner point coordinates and geographic coordinate system).

[0058] In some embodiments, all hyperspectral image files (such as ENVI format files) in the preset data pool are traversed, and the following steps are performed for each image: call GDAL library to read image metadata and parse its geographic coordinate system information; extract the latitude and longitude coordinates of the four corner points of the image (format [top left, top right, bottom right, bottom left]); bind the image file path and the polygon geographic boundary generated by the four corner point coordinates, and insert them into the PostGIS data table as new records.

[0059] Step S403, based on the input central coordinate set, perform spatial relationship query in the database table to filter out the list of hyperspectral image files covering the geographic center coordinates of each pixel; Among them, using the query function in the spatial database, according to the geographic center coordinates of each pixel input, the image file containing the coordinates is retrieved from the database. Spatial relationship query can efficiently find out those image files matching the input coordinate set, ensuring that the image corresponding to each pixel can be accurately filtered out. This step is the key to ensure accurate splicing between different images, avoiding interference of irrelevant data.

[0060] For example, assuming that the geographic center coordinates of the target pixel are (lon_geo=100.5, lat_geo=49.5). The system will perform a spatial inclusion relationship query in PostGIS to obtain all image file paths covering the coordinates. For example, / images / image1.tif may be retrieved because its geographic boundary contains the coordinates.

[0061] Step S404, based on each image in the hyperspectral image file list, calculate the nearest neighbor pixel position in the image according to the geographic center coordinates of each pixel, and extract the spectral data of the nearest neighbor pixel as candidate spectral data; Wherein, for each image in the image file list, based on the geographic center coordinates of the target pixel, calculate the nearest neighbor pixel position of the coordinates in the image. Use this position to extract spectral data as candidate spectral data to ensure that the extracted spectral data is consistent with the geographic location of the target pixel.

[0062] Specifically, regarding the nearest neighbor pixel position calculation, through the input geographic center coordinates, combined with the geographic reference transformation of the image, the position of the coordinates in the image is calculated (usually the nearest pixel position). Then, the spectral data of this position is extracted from the image.

[0063] For example, for each image file in the images list, calculate its nearest neighbor pixel position in the current image according to the target pixel geographic coordinates (lon_geo, lat_geo), read the geographic transformation parameters of the image using GDAL, convert the geographic coordinates to image pixel coordinates (pixel_x, pixel_y) through inverse transformation, take the integer of the coordinates to get the nearest neighbor pixel row and column number, extract the spectral values of all bands of this pixel to form a one-dimensional array: sp_i = [band_1_val, band_2_val, …, band_b_val].

[0064] Step S405, aggregate all extracted candidate spectral data to obtain a spectral set.

[0065] Wherein, the spectral data extracted from all image files is aggregated to form the final spectral set. The spectral set is composed of multiple candidate spectral data, which contains all candidate spectral information of the target pixel, and is used for subsequent spectral analysis, fitting and mineral identification tasks. For example, an empty spectral container spectrums=[] can be created, the images list is traversed, and sp_i extracted from each file is added to the container in order, and the final spectral set spectrums=[sp_1, sp_2, …, sp_n] is output, where n is the number of images covering the target pixel.

[0066] In the above embodiments, a database is established to manage the path and geographical boundary information of the image files, and the image files covering the target pixel positions are accurately screened out through spatial query, thereby ensuring the accuracy and consistency of the image data. On this basis, the nearest neighbor pixel position of each target pixel is calculated and the spectral data is extracted, and finally a rich spectral set is formed. This method improves the accuracy and efficiency of spectral data extraction, and can provide strong support for subsequent high-precision mineral identification and resource assessment.

[0067] Referring to Figure 5 As an embodiment of step S105, the step of calculating the fitting degree of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain the fitting degree array includes: Step S501, traversing each spectrum in the spectral set; Each spectrum is a sequence composed of a series of band values, representing the spectral characteristics such as reflectivity or radiance of the target area or sample. In actual application, the spectral set is usually extracted from pixels in different hyperspectral images, and each spectrum represents different ground features or material characteristics. By traversing each spectrum, its matching degree with different minerals can be analyzed one by one, and its fitting degree can be calculated, which is the basis for generating spectral analysis results.

[0068] Step S502, calculating the fitting degree of the dominant mineral of each spectrum based on the mineral spectral feature analysis algorithm; The mineral spectral feature analysis algorithm is used to identify the mineral of each spectrum and calculate the fitting degree of the dominant mineral corresponding to each spectrum. The fitting degree refers to the similarity between the spectrum and the known mineral spectrum, which is usually represented by a value between 0 and 1, and the higher the value, the higher the matching degree. Through the mineral spectral feature analysis algorithm (such as a software tool implementing the Tetracorder algorithm framework), the system can accurately identify the mineral corresponding to each spectrum and calculate its matching degree.

[0069] Specifically, Tetracorder is a commonly used mineral spectral matching algorithm, and the tool kit GeoAHSI can be used to analyze each spectrum sp_i in the above spectral set. By comparing the unknown spectrum with the data in the known mineral spectrum library, the best match is found and the fitting degree is calculated. This algorithm is widely used in mineral analysis of hyperspectral remote sensing data.

[0070] Step S503, calculating the characteristic absorption depth of the dominant mineral; The absorption depth refers to the depth of the mineral absorption spectrum in a specific band (such as 2.3 μm, 2.35 μm, etc.). The characteristic absorption depth is usually related to the spectral characteristics of the mineral and is an important parameter reflecting the composition of the mineral.

[0071] It can be understood that the absorption depth, as an important component of the mineral spectral feature, can provide data closely related to the characteristics of the mineral. This step, combined with the goodness of fit, helps to improve the accuracy of mineral identification and spectral analysis by comprehensively reflecting the characteristics of the mineral.

[0072] In step S504, the goodness of fit value of each spectrum is generated by weighted product operation of the goodness of fit and the characteristic absorption depth; The formula of the weighted product operation is: goodness of fit value = dominant mineral goodness of fit × characteristic absorption depth × weight coefficient. The dominant mineral goodness of fit represents the matching degree of the spectrum and the mineral, the characteristic absorption depth represents the absorption intensity of the spectrum, and the weight coefficient is used to adjust the importance of the goodness of fit and the absorption depth, which is usually set to 1.0 unless it needs to be adjusted according to actual needs.

[0073] Specifically, the goodness of fit of each spectrum is combined with its characteristic absorption depth by weighted product operation to generate the goodness of fit value of each spectrum. The goodness of fit value is a comprehensive quantitative index reflecting the matching degree of the spectrum and the mineral and its absorption characteristics.

[0074] In step S505, the goodness of fit values of each spectrum are combined in the traversal order to form a goodness of fit array.

[0075] By combining the goodness of fit values of all spectra into an array in order, it is convenient to uniformly process all spectra. The goodness of fit array is usually used as the basis for subsequent decision-making, such as selecting the spectrum with the highest goodness of fit for further analysis.

[0076] In the above embodiments, the mineral goodness of fit and the absorption depth are combined to provide a comprehensive evaluation index for each spectrum. This method not only improves the matching degree of the spectrum and the mineral, but also reflects the absorption characteristics of the mineral characteristics, thereby optimizing the accuracy of mineral identification. By forming the goodness of fit array, the subsequent spectrum screening and analysis become more efficient, providing accurate support for geological exploration, mineral resource evaluation and other fields.

[0077] Reference Figure 6 As an embodiment of step S106, the step of traversing the goodness of fit array, selecting the spectrum corresponding to the maximum goodness of fit, and assigning it as the current pixel data of the target region image includes: In step S601, the maximum goodness of fit value in the goodness of fit array is identified and extracted; The goodness-of-fit array contains the goodness-of-fit values of each spectrum, which represent the matching degree of the spectrum to the target mineral. Each element in the array represents the matching degree between a spectrum and the target mineral, and the larger the value, the higher the matching degree. The goal is to determine the spectral data that best matches the characteristics of the target area by the maximum goodness-of-fit value, so the current optimal spectrum corresponding to the mineral characteristics can be determined by traversing the array and identifying the maximum value.

[0078] Step S602, according to the maximum goodness-of-fit value, extract the corresponding candidate spectrum data from the spectrum set as the optimal spectrum data; Each spectrum data contains a series of band values, which represent the reflection intensity of the mineral at different wavelengths, representing the spectral response of a mineral, which is usually matched through a spectral data set (such as a spectral feature library of different minerals). By extracting the spectrum with the highest goodness-of-fit, it is ensured that the spectral information of the target image can accurately reflect the actual mineral characteristics of the target area.

[0079] Step S603, assign the extracted optimal spectrum data to the specified position of the current pixel in the target area image data storage structure.

[0080] The extracted optimal spectrum data is assigned to the corresponding pixel position in the data storage structure of the target area image. The target image data storage structure is usually a three-dimensional array, where the position of each pixel is determined by the row, column and band index. By assigning the optimal spectrum data to the corresponding position, it is ensured that the spectral data of the target image accurately corresponds to the actual geographical position.

[0081] Specifically, the data storage structure usually uses a three-dimensional array to store image data, and the dimensions are defined by the number of rows, columns and bands of the image. For example, the size of the image is 500 rows x 600 columns, and each pixel has 3 band data, so the data structure is a 500 x 600 x 3 array. For the pixel position amplitude, the position of the current pixel in the image data structure can be determined according to the row number and column number, and then the extracted spectrum data is assigned to the position.

[0082] In the above embodiment, the system traverses the goodness-of-fit array to identify and extract the maximum goodness-of-fit value, ensuring that the spectrum data that best matches the target area is selected. Then, the system extracts the corresponding optimal spectrum from the candidate spectrum set according to the goodness-of-fit value, and finally accurately assigns it to the corresponding position in the target image. This technical solution improves the accuracy of spectral data selection and the quality of image stitching, ensuring that the generated regional hyperspectral remote sensing image has higher spectral consistency and spatial accuracy, providing reliable data support for subsequent mineral identification and resource assessment applications.

[0083] Reference Figure 7As an embodiment of step S107, the step of generating a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameters and the assigned target region image data includes: Step S701: receiving geographic reference transformation parameters and assigned target region image data; The assigned target region image data is usually a three-dimensional array representing spatial data of the image, and the dimensions are defined by the number of rows, columns, and bands of the image. The geographic reference transformation parameters include the minimum longitude, maximum latitude, spatial resolution, and zero value of the image, which will help the subsequent steps to calculate the spatial position information of the image.

[0084] Step S702: setting the affine transformation coefficients of the image file based on the geographic reference transformation parameters, and specifying the coordinate system as a preset geographic coordinate system; The affine transformation coefficients of the image file are set through the geographic reference transformation parameters to ensure that the pixel coordinates of the image can be accurately converted into coordinates in the geographic coordinate system. The core of this step is to generate affine transformation coefficients using the starting longitude and latitude of the image, as well as the spatial resolution. These coefficients usually include the coordinate offset of the image (such as the minimum longitude and maximum latitude), as well as the resolution (such as r) of the image, to realize the conversion from the pixel coordinate system to the geographic coordinate system.

[0085] Step S703: writing the target region image data into the file storage structure to generate a regional hyperspectral remote sensing mosaic image file.

[0086] The system writes the target region image data into the file storage structure to generate a regional hyperspectral remote sensing mosaic image file. The target image data storage structure is usually a three-dimensional array representing the spatial resolution, number of rows, number of columns, and number of bands of the image. This step saves these data as a standard remote sensing image file format, such as TIFF, ENVI, etc., according to the previously set geographic reference parameters.

[0087] In the embodiments of the present application, the image data is stored in the file structure, usually using mainstream image formats such as GeoTIFF or other standard image formats, which support the storage of three-dimensional image data (row x column x band). Further, according to the affine transformation coefficients and the specified coordinate system, the system writes the data into the file system and outputs the finally generated regional hyperspectral remote sensing mosaic image file.

[0088] Exemplarily, first, a three-dimensional data cube is pre-constructed, with dimensions of (row_num, column_num, b); where row_num represents the total number of rows of the target image (the number of pixels in the latitude direction), column_num represents the total number of columns of the target image (the number of pixels in the longitude direction), and b represents the number of hyperspectral bands. The memory allocation can initialize the array using the float32 data type.

[0089] Next, all pixel positions (row index row and column index col) of the target image are traversed, the optimal spectral data sp_opt (one-dimensional array with length b) is obtained from the goodness-of-fit screening result, and then the assignment operation is performed. After the assignment is completed, the cube becomes a complete regional hyperspectral data cube.

[0090] Further, the image space attribute is set based on the geographic reference transformation parameter, the coordinate system is EPSG: 4326 (WGS84 geographic coordinate system), and the affine transformation coefficient is [lon_min, r, 0, lat_max, 0, -r]. Finally, the GDAL library is called to generate an image file, and a hyperspectral remote sensing mosaic image containing geographic coordinate information is obtained.

[0091] In the above embodiment, the geographic reference transformation parameter and the target image data are received and processed to ensure that the image data is ready for geographic mapping. Then, the affine transformation coefficient is set based on the geographic reference transformation parameter, and a standard geographic coordinate system is specified for the image to ensure that the image can be accurately mapped to the actual geographic location. Finally, the system writes the image data into a file storage structure to generate a remote sensing image file with accurate geographic location information. This process ensures the spatial consistency of the image data and provides reliable data support for subsequent remote sensing analysis, mineral identification and geological exploration.

[0092] The embodiment of the present application also discloses a regional hyperspectral remote sensing image mosaic system.

[0093] A regional hyperspectral remote sensing image mosaic system, the system comprising: An acquisition module configured to acquire the geographic range and image spatial resolution of a target region; A parameter calculation module configured to calculate the row and column numbers and the geographic reference transformation parameter of the image of the target region based on the geographic range and image spatial resolution; A center coordinate calculation module configured to traverse each pixel position of the image of the target region based on the row and column numbers, and calculate the geographic center coordinates of each pixel based on the geographic reference transformation parameter to obtain a center coordinate set; A spectral extraction module configured to retrieve a hyperspectral image file covering the center coordinate set from a pre-set data pool, and extract candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set; a goodness-of-fit calculation module, configured to calculate a goodness of fit of a spectrum-dominant mineral corresponding to each spectrum in the spectrum set, to obtain a goodness-of-fit array; a pixel assignment module, configured to traverse the goodness-of-fit array, select candidate spectrum data corresponding to a maximum goodness of fit, and assign the candidate spectrum data as current pixel data of the target region image; an image file generation module, configured to generate a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameter and the target region image after the assignment.

[0094] The regional hyperspectral remote sensing image mosaic system according to the embodiments of the present application can implement any of the above-mentioned regional hyperspectral remote sensing image mosaic methods, and the specific working processes of the modules in the regional hyperspectral remote sensing image mosaic system can refer to the corresponding processes in the above-mentioned method embodiments.

[0095] In several embodiments provided in the present application, it should be understood that the provided methods and systems can be implemented in other manners. For example, the above-mentioned system embodiments are merely schematic; for example, the division of a certain module is merely a logical function division, and actual implementation can have another division manner, for example, a plurality of modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0096] The embodiments of the present application further disclose a computer device.

[0097] The computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements a regional hyperspectral remote sensing image mosaic method as described above when executing the computer program.

[0098] The embodiments of the present application further disclose a computer readable storage medium.

[0099] The computer readable storage medium stores a computer program capable of being loaded and executed by a processor to implement any of the above-mentioned regional hyperspectral remote sensing image mosaic methods.

[0100] The computer readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, device or apparatus; the program code contained in the computer readable medium can be transmitted by any appropriate medium, including but not limited to wireless, wire, optical cable, RF, etc., or any appropriate combination of the above.

[0101] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0102] The above are only preferred embodiments of the present application, not intended to limit the protection scope of the present application, any one feature disclosed in the specification (including the abstract and the drawings) can be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A method for mosaicking regional hyperspectral remote sensing images, characterized in that, The method comprises: acquiring a geographical range and an image spatial resolution of a target region; calculating the number of rows and columns of the target region image and the geographical reference transformation parameters according to the geographical range and the image spatial resolution; traversing each pixel position of the target region image based on the number of rows and columns, and calculating the geographical center coordinates of each pixel according to the geographical reference transformation parameters to obtain a center coordinate set; retrieving a hyperspectral image file covering the center coordinate set from a preset data pool, and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set; calculating the fitting goodness of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain a fitting goodness array; traversing the fitting goodness array, selecting the candidate spectral data corresponding to the maximum fitting goodness, and assigning it as the current pixel data of the target region image; generating a regional hyperspectral remote sensing mosaic image file based on the geographical reference transformation parameters and the target region image after assignment.

2. The method of claim 1, wherein, The step of calculating the number of rows and columns of the target region image and the geographical reference transformation parameters according to the geographical range and the image spatial resolution comprises: acquiring a longitude range, a latitude range and an image spatial resolution of a target region; calculating the number of rows of the target region image based on the latitude range and the image spatial resolution, and calculating the number of columns of the target region image based on the longitude range and the image spatial resolution; generating an affine transformation coefficient array according to the minimum longitude value, the maximum latitude value and the image spatial resolution of the target region; the affine transformation coefficient array comprises the minimum longitude value, the image spatial resolution, zero, the maximum latitude value, zero and a negative spatial resolution.

3. The method of claim 2, wherein, The step of traversing each pixel position of the target region image based on the number of rows and columns, and calculating the geographical center coordinates of each pixel according to the geographical reference transformation parameters comprises: traversing each pixel position of the target region image in row number and column number order based on the number of rows and columns; for each traversed pixel position (row, col), calculating the longitude coordinate and the latitude coordinate of the geographical center of each pixel according to the affine transformation coefficient array, and the calculation formula is: lon_geo = lon_min + col * r; lat_geo = lat_max - row * r; In the above formula, lon_geo is the longitude coordinate, lon_min is the minimum longitude value, lat_geo is the latitude coordinate, lat_max is the maximum latitude value, and r is the spatial resolution.

4. The method of claim 1, wherein, The step of retrieving a hyperspectral image file covering the center coordinate set from a preset data pool, and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set comprises: creating a database table to store image file paths and corresponding geographical boundary geometric data; traversing the hyperspectral image files in the preset data pool, parsing the geographical coordinate system and the geographical coordinates of the four corner points for each image, and storing the file path and the polygon geographical boundary generated based on the four corner point coordinates in the database table; based on the input center coordinate set, performing a spatial relationship query in the database table to filter out a list of hyperspectral image files covering the geographical center coordinates of each pixel; According to the geographic center coordinates of each pixel, a nearest neighbor pixel position in the image is calculated based on each image in the hyperspectral image file list, and spectral data of the nearest neighbor pixel is extracted as candidate spectral data; All extracted candidate spectral data is summarized to obtain a spectral set.

5. The method of claim 4, wherein, The step of calculating the fitting degree of the spectral dominant mineral corresponding to each spectrum in the spectral set includes: Each spectrum in the spectral set is traversed; The fitting degree of the dominant mineral of each spectrum is calculated based on a mineral spectral feature analysis algorithm; The characteristic absorption depth of the dominant mineral is calculated; The fitting degree value of each spectrum is generated through weighted multiplication operation of the fitting degree and the characteristic absorption depth; The fitting degree values of each spectrum are combined in traversal order to form a fitting degree array.

6. The method of claim 5, wherein, The step of traversing the fitting degree array, selecting the spectrum corresponding to the maximum fitting degree, and assigning it as the current pixel data of the target region image includes: The maximum fitting degree value in the fitting degree array is identified and extracted; According to the maximum fitting degree value, the corresponding candidate spectral data in the spectral set is extracted as optimal spectral data; The extracted optimal spectral data is assigned to the specified position of the current pixel in the target region image data storage structure.

7. The method for mosaicking a regional hyperspectral remote sensing image according to any one of claims 1 to 6, characterized in that, The step of generating a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameters and the assigned target region image includes: Receiving geographic reference transformation parameters and assigned target region image data; Setting the affine transformation coefficients of the image file based on the geographic reference transformation parameters, and specifying the coordinate system as a preset geographic coordinate system; Writing the target region image data into the file storage structure to generate a regional hyperspectral remote sensing mosaic image file.

8. A regional hyperspectral remote sensing image mosaic system, characterized in that, The system includes: An acquisition module for acquiring the geographic range and image spatial resolution of the target region; A parameter calculation module for calculating the row and column numbers and geographic reference transformation parameters of the target region image based on the geographic range and image spatial resolution; A center coordinate calculation module for traversing each pixel position of the target region image based on the row and column numbers, and calculating the geographic center coordinates of each pixel based on the geographic reference transformation parameters to obtain a center coordinate set; A spectral extraction module for retrieving hyperspectral image files covering the center coordinate set from a preset data pool, and extracting candidate spectral data corresponding to the coordinates based on the nearest neighbor method to form a spectral set; A fitting degree calculation module for calculating the fitting degree of the spectral dominant mineral corresponding to each spectrum in the spectral set to obtain a fitting degree array; A pixel assignment module for traversing the fitting degree array, selecting the candidate spectral data corresponding to the maximum fitting degree, and assigning it as the current pixel data of the target region image; An image file generation module for generating a regional hyperspectral remote sensing mosaic image file based on the geographic reference transformation parameters and the assigned target region image.

9. A computer device, comprising: A computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and executed to perform the method of any one of claims 1 to 7.

Images