Coal surface reflectivity enhancement method based on multi-angle spectrum consistency constraint

Abstract

The invention relates to the technical field of spectral reflectivity, in particular to a coal surface reflectivity enhancement method based on multi-angle spectral consistency constraint, which comprises the following steps of: synchronously acquiring multi-angle spectral data by a plurality of unmanned aerial vehicles, constructing a spatial registration image group, extracting consistent pixels to construct an angle reflection structure, and identifying reflection difference wave bands and directions. And executing directivity correction, and fusing to generate an enhanced reflectivity image result. The method comprises the following steps: acquiring a multi-angle spectral reflectivity image group through a plurality of unmanned aerial vehicles, establishing a pixel corresponding relation in combination with time stamps and spatial registration, aggregating reflectivity values at the same position according to angles to form a directional structure, identifying an abnormal wave band region through difference calculation, and processing angle differences in combination with spectral sequence rearrangement and direction correction, so as to identify the abnormal wave band region. A unified reflectivity value set is constructed, a two-dimensional image matrix is generated through spatial splicing and fusion, and the spatial consistency and direction adaptability of coal area multi-angle spectral data are enhanced.

Description

Technical Field

[0001] This invention relates to the field of spectral reflectance technology, and in particular to a method for enhancing the reflectance of coal surfaces based on multi-angle spectral consistency constraints. Background Technology

[0002] The field of spectral reflectance technology encompasses imaging detection methods for acquiring, modeling, and analyzing the reflectance characteristics of target surfaces under different lighting and observation angle conditions. It is widely used in geological exploration, environmental remote sensing, agricultural monitoring, and energy resource assessment. The core content is based on hyperspectral or multispectral imaging equipment, combined with techniques such as radiometric calibration, geometric correction, reflectance inversion, and consistency modeling to achieve accurate estimation and analysis of the reflectance of land surfaces or object surfaces. In coal detection scenarios, spectral reflectance measurements must address the influence of external environmental lighting interference, changes in observation angle, and target surface roughness. Therefore, this field typically combines multi-angle imaging, reflectance interpolation modeling, and angle normalization reconstruction to improve the stability and applicability of reflectance data.

[0003] Among them, the coal surface reflectance enhancement method based on multi-angle spectral consistency constraints refers to the problem of reflectance data distortion under multi-angle observation conditions in coal surface spectral detection. This method enhances reflectance by introducing a spectral consistency constraint model. It utilizes spectral data obtained from repeated observations at multiple angles to construct an angle-dependent differential expression model. Combined with the reflectance variation trend, it fits and corrects the spectral differences between observation angles through numerical optimization, and completes the inter-frame reflectance mapping using a baseline viewpoint reflectance reconstruction mechanism. Core techniques include angle difference modeling, joint analysis of multi-frame spectral sequences, spectral angle-dependent weight fitting, and consistency reconstruction calculation, used to improve the stability of coal surface reflectance data in complex outdoor environments.

[0004] Existing technologies rely on single-point observation and image sequence processing to model reflectance, which makes it difficult to obtain continuous spatiotemporal correlations of multi-angle observation data over a large area. The lack of time stamps and spatial registration between images leads to inaccurate data matching, and the expression of angular reflectance characteristics is limited. This results in reflectance data at the same spatial location at different angles being prone to deviation. It also lacks the ability to locate and correct local distortion areas, making it difficult to meet the requirement of stable reflectance extraction under conditions of strong angular interference or drastic reflectance changes. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for enhancing the reflectivity of coal surfaces based on multi-angle spectral consistency constraints. The technical solution is as follows: A method for enhancing the reflectivity of coal surfaces based on multi-angle spectral consistency constraints includes the following steps: S1: By controlling multiple unmanned aerial vehicles to scan the surface of the coal pile along different flight paths, spectral reflectance data of the covered area is collected. At the same time, the shooting orientation and flight coordinates are recorded. The images are registered and track matched to construct a coal area reflectance image group under multi-angle correspondence. S2: Based on the coal area reflectance image group under the multi-angle correspondence, extract coal surface area points with positional consistency, group image pixels in the same spatial position according to the observation angle, establish the pairing relationship between direction and reflectance value, and construct the coal surface pixel and angle reflectance data structure. S3: Based on the aggregation structure of the coal surface pixels and angular reflectance data, the reflectance values ​​at different angles are sequence-expanded and the difference is calculated to identify the band regions with large variation amplitudes and mark the corresponding observation directions, thereby constructing the identification results of the directional difference region of coal surface reflectance. S4: Based on the abnormal observation direction in the identification results of the directional difference region of coal surface reflectance, extract the relevant reflectance sequence in the original image, perform offset correction and trend sorting on the difference bands, and generate an angle-corrected reflectance numerical set.

[0006] As a further aspect of the present invention, the coal area reflectance image group includes multi-view image data, reflectance distribution matrix, and spatiotemporal identification information; the aggregation structure of coal surface pixels and angular reflectance data includes angle numbering information, spatially corresponding pixels, and directional reflectance set; the identification result of directional difference regions of coal surface reflectance includes high difference band intervals, abnormal observation direction markers, and directional change characteristics; and the angle-corrected coal area reflectance numerical set includes adjusted spectral sequence, difference band correction values, and unified directional reflectance data.

[0007] As a further aspect of the present invention, the step of obtaining S1 is as follows: S101: Based on the spectral image frame sequence acquired by multiple unmanned aerial vehicles during flight along different routes and attitude angles, the continuous spectral reflectance data within the coverage area is extracted frame by frame. Combined with the flight path coordinate sequence and shooting azimuth angle data acquired synchronously by the aircraft, the image frames are matched with the corresponding time series numbers during the processing. Through the matching processing of image frames with their position information, the spatiotemporal mapping dataset corresponding to the spectral image frames is obtained. S102: Based on the image time sequence number and path coordinate sequence in the spatiotemporal mapping dataset corresponding to the spectral image frame, the pairing process is executed according to the order in which the image frames appear, and the path coordinate changes between adjacent image frames are processed in correspondence with the shooting azimuth angle data. The displacement correlation information group of multi-view image frames is obtained by comparing the continuity difference of the image frames with the attitude angle data. S103: Based on the coordinate relationship content in the displacement association information group of the multi-view image frames, and according to the path value and attitude angle value relationship between adjacent image frames, the image frame content is readjusted under unified spatial parameters so that the reflection data of each image frame is processed in the same coordinate system, and finally the coal area reflectivity image group under the multi-angle correspondence is obtained.

[0008] As a further aspect of the present invention, the coordinate relationship content in the displacement association information group of the multi-view image frames is used to filter image frame pairs that satisfy both the path coordinate change range and the shooting azimuth angle change range within a preset threshold range to establish an effective matching relationship; the process of readjusting the image frame content under unified spatial parameters includes: normalizing the spatial position of the image frames based on the effective matching relationship, and performing image reprojection operation according to the attitude angle change between image frames to achieve spatial alignment; the coal area reflectivity image group under the multi-angle correspondence relationship is obtained by reconstructing image frames under a unified coordinate system, and includes multiple image frames with different shooting azimuth angles that satisfy the effective matching relationship.

[0009] As a further aspect of the present invention, the step of obtaining S2 is as follows: S201: Based on the multi-angle correspondence, the position coordinates of the same pixel in the coal area reflectance image group under different angles are cross-filtered, the spatial reference coordinate values ​​of each image frame are called, and the spatial position index values ​​of all pixels are compared to obtain the spatial consistency index set of position points. S202: Based on the coordinate numbers of each location point in the spatial consistency index set, perform classification operations on pixels with the same coordinate numbers according to the acquisition order of the spectral image frames, call different angle number values ​​to assign corresponding values ​​to the reflectance data corresponding to each angle, and obtain the reflectance data group under the angle order. S203: For each reflectance value and direction number data in the reflectance data group under the angular order, based on the corresponding spatial coordinate number of each group, the direction number and the corresponding reflectance value of each group are structurally merged, and all spatial position index numbers are uniformly mapped to obtain the coal surface pixel angle reflectance data structure.

[0010] As a further aspect of the present invention, the step of obtaining S3 is as follows: S301: Based on the reflectance value sequence corresponding to each position index in the coal surface pixel angle reflectance data structure, for each set of reflectance value sequences under the observation direction, extract its numerical order at all band positions, call the sequence order data corresponding to each direction for position comparison processing, and obtain the band-corresponding reflectance sequence set; S302: Based on the same band position in the reflectance sequence set corresponding to the band, extract the reflectance values ​​under each observation direction, call the direction index order to compare the values ​​under adjacent observation directions pairwise, and number the band position sequence after each comparison according to the direction order to obtain the direction sequence band comparison result set. S303: For the numerical variation of each band in the directional sequence band comparison result set under different observation directions, monitor the degree of numerical difference in the directional sequence under each group of bands, extract and process according to the directional index corresponding to the range of difference, classify all bands with prominent range of variation and their corresponding directional index contents, and obtain the identification results of the directional difference region of coal surface reflectivity.

[0011] As a further aspect of the present invention, the step of obtaining S4 is as follows: S401: Based on the observation direction index contained in the identification result of the directional difference region of coal surface reflectance, extract the spectral reflectance sequence under the corresponding direction in the original reflectance image frame, call the spectral values ​​of each band position in the difference region, combine the spatial index data to locate the content of the corresponding image region, and obtain the abnormal direction reflectance band sequence. S402: For the spectral value of each band position in the abnormal directional reflectance band sequence, call the spectral value sequence of the same position in other directional sequences, compare their numerical relationships, and perform position adjustment operation on the spectral value sequence according to the order offset to obtain the band position comparison adjustment result; S403: Based on the order of the adjusted spectral values ​​in the band position comparison adjustment results, match them one-to-one with the spatial indexes in the original sequence, classify and organize the reflectance data frame content according to each spatial position number, re-classify the values ​​under all abnormal observation directions into the corresponding spatial mapping index, and obtain the angle-corrected reflectance value set.

[0012] As a further aspect of the present invention, when the spectral values ​​of each band position in the abnormal directional reflectance band sequence are compared with the spectral values ​​of the same position in other directional sequences, the position adjustment operation is only performed on spectral values ​​that meet the condition of setting a limit on the order offset range, and the order difference does not exceed the preset adjustment tolerance. In the process of matching the adjusted spectral value order with the spatial index in the original sequence in the band position comparison adjustment result, the image region content with continuous spatial number is selected first for sorting based on the continuity of the spatial index as the matching premise. When each spectral value in the angle-corrected reflectance numerical set is assigned to the corresponding spatial mapping index, normalization processing is performed according to the distribution trend of spectral values ​​of different directional sequences under the same spatial position, so as to eliminate the interference of observation direction differences on reflectance data.

[0013] As a further aspect of the present invention, the method further includes: S5: Based on the set of coal area reflectance values ​​after angle correction, the reflectance values ​​at each location are stitched together into an image matrix, all directional data are fused and mapped to spatial coordinates to construct an enhanced coal surface reflectance image result. The enhanced reflectance image of the coal surface includes a fused reflectance image, a corrected image matrix, and a spatial coordinate mapping result.

[0014] As a further aspect of the present invention, the step of obtaining S5 is as follows: S501: Based on the reflectance values ​​corresponding to each spatial index in the set of angle-corrected reflectance values, extract the spectral values ​​of each location point according to their coordinate order, and call the spatial coordinate sequence to stitch the reflectance of each point. Arrange all reflectance values ​​in a two-dimensional structure to obtain a spatial stitched image matrix. S502: Based on the image pixel content of each spatial position in the spatial stitching image matrix, synchronously process the image data of the same area under different observation directions, sequentially extract the data frame content of the corresponding position in the image matrix under each direction, call the reflectance value of each direction to perform channel-level corresponding processing, and obtain a multi-directional reflectance stitching matrix. S503: For the image data frames in the multi-directional reflectivity stitching matrix, extract the pixel positions corresponding to each spatial index, and map each pixel position in the image to the index number of the coal pile surface area through spatial parameter mapping rules. Fill in the image data content according to the correspondence to obtain the coal surface reflectivity enhancement image result.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: In this invention, a wide-coverage set of spectral reflectance images is acquired using multiple unmanned aerial vehicles. The images are registered and matched using time stamps and spatial coordinates to establish pixel correspondence under multi-angle observation. By aggregating reflectance values ​​at the same location by angle to form a directional data structure, the combined expressive power between angle and reflectance is enhanced. Anomalous reflectance regions are identified through difference calculation and band difference extraction. The reflectance differences between multiple angles are processed by combining spectral sequence rearrangement and directional correction to construct a unified set of reflectance values. A complete two-dimensional image matrix is ​​generated through spatial stitching and fusion operations, thereby improving the spatial coordination and directional correlation of multi-angle spectral information in coalfields. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart illustrating the acquisition process of S1 in this invention; Figure 3This is a flowchart illustrating the acquisition process of S2 in this invention; Figure 4 This is a flowchart illustrating the acquisition process of S3 in this invention; Figure 5 This is a flowchart illustrating the acquisition process of S4 in this invention; Figure 6 This is a flowchart of the acquisition process for S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1 This invention provides a technical solution: a method for enhancing the surface reflectance of coal based on multi-angle spectral consistency constraints, comprising the following steps: S1: By controlling multiple unmanned aerial vehicles to fly synchronously over the surface of the coal pile along different routes and attitude angles, continuous spectral reflectance data of the covered area is collected. The shooting orientation and flight path coordinate information are recorded in each flight, and each frame of the image is time-stamped. Based on the recorded position information and time sequence, the spectral images are spatially registered and track matched to construct a coal area reflectance image group under multi-angle correspondence. S2: Based on the coal area reflectance image group under the multi-angle correspondence, extract the location points of the coal pile surface area with spatial location consistency relationship, group the image pixels in the same spatial location according to the acquisition order of different angles, number and mark the reflectance values ​​under different angles in each group, establish the pairing relationship between each direction and the corresponding reflectance value, and construct the coal surface pixel and angle reflectance data structure. S3: Based on the aggregation structure of coal surface pixel and angular reflectance data, position alignment and band sequence expansion are performed on the reflectance values ​​under different observation directions in each set of reflectance data. The difference is calculated according to the same band position in the sequence. Band-by-band comparison is performed on the obtained difference sequence to identify the band interval with large reflectance variation and record its corresponding observation direction. The result of identifying the directional difference region of coal surface reflectance is constructed. S4: Based on the abnormal observation direction recorded in the identification results of the directional difference region of coal surface reflectance, extract the reflectance sequence corresponding to the direction in the original image, perform offset correction operation on its band position in the difference region, rearrange the spectral values ​​of the difference band according to its reflectance change trend in other directions, and perform order adjustment processing to generate an angle-corrected reflectance value set. S5: Based on the set of reflectance values ​​of coal areas after angle correction, the reflectance values ​​of each area are stitched together according to spatial coordinates to form a two-dimensional image matrix. The reflectance data of all observation angles after processing are uniformly fused, and the fusion result is mapped to the coordinate space of the coal pile surface area to construct the coal surface reflectance enhancement image result.

[0023] The coal area reflectance image set includes multi-view image data, reflectance distribution matrix, and spatiotemporal identification information. The aggregated structure of coal surface pixel and angular reflectance data includes angle numbering information, spatially corresponding pixels, and directional reflectance set. The identification results of directional difference regions of coal surface reflectance include high difference band intervals, abnormal observation direction markers, and directional change characteristics. The set of coal area reflectance values ​​after angle correction includes adjusted spectral sequences, difference band correction values, and unified directional reflectance data. The coal surface reflectance enhancement image results include fused reflectance images, corrected image matrices, and spatial coordinate mapping results.

[0024] Please see Figure 2 The steps to obtain S1 are as follows: S101: Based on the spectral image frame sequence acquired by multiple unmanned aerial vehicles during flight along different routes and attitude angles, the continuous spectral reflectance data within the coverage area is extracted frame by frame. Combined with the flight path coordinate sequence and shooting azimuth angle data acquired synchronously by the aircraft, the image frames are matched with the corresponding time series numbers during the processing. Through the matching processing of image frames with their position information, the spatiotemporal mapping dataset corresponding to the spectral image frames is obtained. First, multiple unmanned aerial vehicles (UAVs) are activated to collect spectral image frame sequences at a fixed frequency along a set flight path. Simultaneously, the flight path coordinate sequence is recorded synchronously through the airborne high-precision positioning module. And the azimuth data recorded by the inertial measurement unit. In the processing, for any frame in the spectral image frame sequence... Accurately extract the shooting trigger timestamp from its metadata. To address the time discrepancy caused by inconsistent sensor sampling frequencies, a search was conducted in the flight path coordinate sequence for... Two GPS recording points that are adjacent in time are recorded as the preceding time. and their corresponding three-dimensional spatial coordinates and subsequent time series and its coordinates An interpolation model is constructed based on the time ratio, and the time interpolation coefficients are calculated. Its calculation formula is Subsequently, linear interpolation logic was used to calculate the precise spatial coordinates corresponding to the trigger time of the image frame. The operational logic is expressed as follows That is, to perform operations on longitude, latitude and altitude respectively. Equal component calculation. Similarly, combining the shooting azimuth data, using the same interpolation coefficients... pitch angle Roll angle and yaw angle Perform synchronous calculations to obtain fine pose data corresponding to the image frames. For example, when , , At that time, the calculation yielded ,like , ,but Finally, the calculated results and With image frame By performing unique binding and processing the correspondence between image frames and their location information, a spatiotemporal mapping dataset corresponding to the spectral image frame is obtained.

[0025] S102: Based on the image time sequence number and path coordinate sequence in the spatiotemporal mapping dataset corresponding to the spectral image frame, the pairing process is executed according to the order in which the image frames appear. The path coordinate changes between adjacent image frames are processed in correspondence with the shooting azimuth data. The displacement correlation information group of multi-view image frames is obtained by comparing the continuity difference of the image frames with the attitude angle data. Call the spatiotemporal mapping dataset corresponding to the spectral image frames, according to the image time series numbering. Extract two temporally adjacent frames and To accurately determine the geometric relationships between images and obtain the relative flight altitude of the aircraft. and the field of view parameters of the spectral camera Through geometric optical projection calculations Calculate the theoretical width of a single frame image covering the ground. Simultaneously, the baseline length between the center points of adjacent image frames is calculated. ,Right now and The Euclidean distance between them. The overlap rate is calculated based on these two parameters. The formula is This is used to assess the continuity between images. Based on this, attitude angle data is then incorporated. Calculate the rotation matrix and change the path coordinates between adjacent image frames. With attitude angle change Perform the corresponding processing. Utilize the rotation matrix. Transform the image coordinate system and establish a pixel displacement correlation model between adjacent image frames by combining the displacement vector. As shown in Table 1, if , Then calculate ;like Then the overlap rate By comparing the continuity differences of image frames with the attitude angle data, sufficient overlapping field of view is ensured for multi-view analysis, ultimately yielding a group of displacement correlation information for multi-view image frames.

[0026] Table 1: Flight Data Acquisition Parameters and Overlap Calculation Table

[0027] S103: Based on the coordinate relationship content in the displacement association information group of multi-view image frames, and according to the path value and attitude angle value relationship between adjacent image frames, the image frame content is readjusted under unified spatial parameters so that the reflection data of each image frame is processed in the same coordinate system, and finally the coal area reflectivity image group under multi-angle correspondence is obtained. Based on the coordinate relationships in the displacement correlation information group of multi-view image frames, a unified geographic reference coordinate system is established. For each frame of the image, its position coordinates are used. and attitude data Construct the exterior orientation element matrix Based on the relationship between path values ​​and attitude angle values ​​between adjacent image frames, for each pixel in the image frame... Perform inverse kinematics calculations on the collinearity equations and project them precisely onto plane coordinates in the geographic reference coordinate system. In this process, the image frame content is readjusted under unified spatial parameters to eliminate geometric distortions caused by tilted shooting angles and terrain undulations. Based on resampling rules, the reflection data of each image frame are processed correspondingly within the same grid coordinate system. Specifically, for the same ground grid unit... Indexing its different image frames The reflectivity value in and the corresponding observation angle vector This ensures that reflectance data from each coalfield, captured at different times and angles, are indexed to the same geographical location, achieving spatial alignment of multi-source data. Ultimately, a set of coalfield reflectance images with multi-angle correspondences is obtained. This image set, indexed by geographic coordinates, stores spectral reflectance information for each location at different observation angles.

[0028] Please see Figure 3 The steps to obtain S2 are as follows: S201: Based on the multi-angle correspondence of the coal area reflectance image group, the position coordinates of the pixels at the same position in the image frame under different angles are cross-filtered, the spatial reference coordinate values ​​of each image frame are called, and the spatial position index values ​​of all pixels are compared to obtain the spatial consistency index set of the position points. Based on the multi-angle correspondence of coal area reflectance image frames, the position coordinates of pixels at the same location in each image frame under different angles are cross-filtered. To ensure accurate correspondence of physical locations, a spatial consistency judgment threshold is set. (For example, set to 0.2 meters). Feature points in two image frames from different angles to be matched. and The spatial reference coordinates of each image frame are retrieved to obtain its reprojected ground coordinates. and Calculate the Euclidean distance deviation between two points using the following formula: Perform a positional alignment operation on the spatial index values ​​of all pixels: if the calculated... Then determine and Observations pointing to the same physical location are considered valid and included in the spatial consistency index set of location points; if If the coordinates of point A are (500.1, 600.1) and the coordinates of point B are (500.2, 600.15), then the calculated values ​​are: If the distance is less than the 0.2m threshold, the consistency is considered passed. This step uses strict geometric distance verification to obtain a spatial consistency index set of location points, ensuring that the data processed subsequently originates from the same physical reflecting surface.

[0029] S202: Based on the coordinate numbers of each location point in the spatial consistency index set, perform classification operations on pixels with the same coordinate numbers according to the acquisition order of the spectral image frames, call different angle number values ​​to assign corresponding values ​​to the reflectance data corresponding to each angle, and obtain the reflectance data group under the angle order. Based on the coordinate numbers of each location point in the spatial consistency index set In the spatiotemporal mapping dataset, all validated image frames are retrieved in reverse order. For pixels with the same coordinate number, the acquisition time of the spectral image frame is used as the basis for the search. The classification operations are performed in a specific order to create a timeline index. Based on this, different angle-based numbering values ​​are called. (This number corresponds to specific observation geometry, such as elevation or azimuth angle.) Assign values ​​to the reflectance data corresponding to each angle. Extract the reflectance values ​​of the corresponding pixels in each image frame. Generate a sequence arranged in order of angular change. This process transforms pixel values, originally discretely distributed across different image files, into a continuous stream of observation data for a specific physical location. For example, for coordinate point P1, its values ​​are sequentially extracted from... Time (10-degree perspective) The reflectance values ​​at each time point (20-degree viewing angle) are grouped into ordered sets. Ultimately, the reflectance data sets in angular order are obtained, laying the data foundation for subsequent analysis of the variation of reflectance with angle.

[0030] S203: For each reflectance value and direction number data in the reflectance data group under the angle order, based on the corresponding spatial coordinate number of each group, the direction number and the corresponding reflectance value of each group are structurally merged, and all spatial position index numbers are uniformly mapped to obtain the coal surface pixel angle reflectance data structure. For each reflectance value in the reflectance data set under angle order With direction number data A multidimensional data container containing spatial, spectral, and angular dimensions is constructed. This is based on the corresponding spatial coordinate numbers for each group. Calculate its corresponding center coordinates As a unified spatial index, it eliminates minor positioning errors. Each group of direction numbers and corresponding reflectivity values ​​are structurally merged, mapping abstract direction numbers to specific physical parameters, such as local incident angles. and observation azimuth To form a mapping relationship A unified mapping process is performed on all spatial location index numbers to construct a global lookup table. As shown in Table 2, this step enables quick access to the full-angle reflectance data corresponding to a location using a unique index ID. Finally, the pixel angular reflectance data structure of the coal surface is obtained, achieving a complete conversion from image pixel space to physical attribute space.

[0031] Table 2: Data Structure Table of Coal Pixel Angle Reflectance

[0032] Please see Figure 4 The steps to obtain S3 are as follows: S301: Based on the reflectance value sequence corresponding to each position index in the coal surface pixel angle reflectance data structure, for each set of reflectance value sequences under the observation direction, extract its numerical order at all band positions, call the sequence order data corresponding to each direction for position comparison processing, and obtain the band-corresponding reflectance sequence set; Based on the location index of the coal surface pixel angle reflection data structure The corresponding reflectance value sequence traverses all spectral band positions supported by the spectroscopic camera. For each set of reflectance values ​​under each observation direction, their numerical order across all spectral bands is extracted. Specifically, for a specific spectral band... (e.g., near-infrared band), extract the sequence of this location in all observation directions. Based on the reflectance values ​​below, construct the reflectance sequence corresponding to the band. The sequence data corresponding to each direction is retrieved to ensure that the reflectance values ​​are arranged strictly in logical order from the maximum negative value to the maximum positive value according to the observation angle (e.g., from -45° to +45°), and positional comparison processing is performed. Through this operation, the complex data cube is reduced to a series of one-dimensional band curves arranged by angle, obtaining the reflectance sequence set corresponding to each band, thus highlighting the anisotropic characteristics under a single band.

[0033] S302: Based on the same band position in the reflectance sequence set corresponding to the band, extract the reflectance values ​​under each observation direction, call the direction index order to compare the values ​​under adjacent observation directions pairwise, and number the band position sequence after each comparison according to the direction order to obtain the direction sequence band comparison result set. Based on the same band position in the reflectance sequence set corresponding to the band, extract the reflectance values ​​for each observation direction. Then, using the direction index order, perform the following operations on the sequence. Adjacent observation directions and Extract the corresponding reflectance value and The values ​​from adjacent observation directions are compared pairwise, and the reflectance gradient between adjacent directions is calculated. The calculation logic is as follows: This gradient value reflects the degree of reflectivity variation with angle. The band position sequences after each set of controls are numbered sequentially according to direction to generate gradient sequences. This detailed record of the local slope information of the reflectivity curve indicates that an abnormally large gradient value at a certain point often suggests the presence of specular reflection or shadow occlusion. Ultimately, a set of directional sequence band comparison results is obtained, providing a quantitative basis for subsequent difference identification.

[0034] S303: For the numerical variation of each band under different observation directions in the band comparison results set of the directional sequence, monitor the degree of numerical difference in the directional sequence under each group of bands, extract and process according to the directional index corresponding to the range of difference, classify all bands with prominent range of change and their corresponding directional index contents, and obtain the identification results of the directional difference region of coal surface reflectivity. This study monitors the numerical variations of each band in the directional sequence band comparison dataset under different observation directions, and assesses the degree of numerical differences in the directional sequences within each band group. The statistical indicator, the coefficient of variation, is introduced. To quantify this difference, first, the average reflectance of the sequence is calculated. Next, calculate the standard deviation of the sequence. Finally, through the formula Obtain the coefficient of variation. Set a threshold for determining the difference. (For example, set to 0.2). Extraction is performed according to the direction index corresponding to the range of difference variation: if a certain band sequence If so, it is determined that there is a significant directional difference in the region, and the corresponding directional index is extracted. All bands with prominent variations and their corresponding directional indices were categorized, as shown in Table 3. Point P_A002, with a CV as high as 0.6, was identified as an anomalous region. Finally, the identification results of the directional difference region of coal surface reflectivity were obtained, and the spatial points requiring correction were accurately located.

[0035] Table 3: Calculation Table for Identifying Differential Regions

[0036] Please see Figure 5 The steps to obtain S4 are as follows: S401: Based on the observation direction index contained in the identification results of the directional difference region of coal surface reflectance, extract the spectral reflectance sequence under the corresponding direction in the original reflectance image frame, call the spectral values ​​of each band position in the difference region, combine the spatial index data to locate the content of the corresponding image region, and obtain the abnormal direction reflectance band sequence. Based on the observation direction index included in the identification results of directional differences in coal surface reflectance, all spatial locations marked as "significantly different" were analyzed. and the corresponding abnormal direction index A backtracking operation is performed on the original reflectance image frame to extract the spectral reflectance sequence in the corresponding direction. Spectral values ​​of each band within the difference region are retrieved and combined with spatial index data to precisely locate the content of the corresponding image region. Specifically, the abnormal direction... Spectral reflectance values ​​below Compared with the spectral value sequence of this location under other normal observation directions An anomaly analysis dataset was constructed for this specific point. This operation successfully isolated the problematic data from the massive amount of observation data, obtained the reflectivity band sequence of the anomaly direction, and ensured that subsequent correction operations were only performed on the specific outlier without affecting the integrity of the normal data.

[0037] S402: For the spectral value of each band position in the abnormal directional reflectance band sequence, call the spectral value sequence of the same position in other directional sequences, compare their numerical relationships, and perform position adjustment operation on the spectral value sequence according to the order offset to obtain the band position comparison adjustment result; For each band position in the reflectance band sequence of anomaly directions, the spectral values ​​of the same position in other directional sequences are used as a reference. Anomaly direction is selected. Adjacent Preorder Normal Direction and subsequent normal direction and the corresponding reflectivity values and The numerical relationships were compared and processed to calculate the angular distance weights. Based on the sequence offset, the baseline correction value is calculated using linear interpolation logic. To preserve the physical accuracy of the data, a modified intensity coefficient is introduced. (For example, using 0.8), perform a position adjustment operation on the spectral value sequence. The formula for calculating the final correction value is as follows: For example, the current order value Post-order value outliers Weight At that time, calculate Final correction value This step yields the results of band position adjustment, effectively suppressing strong reflection anomalies.

[0038] S403: Based on the order of the adjusted spectral values ​​in the band position comparison adjustment results, match them one by one with the spatial indexes in the original sequence, classify and organize the reflectance data frame content according to each spatial position number, re-incorporate the values ​​under all abnormal observation directions into the corresponding spatial mapping index, and obtain the angle-corrected reflectance value set. Based on the order of the adjusted spectral values ​​in the band position comparison adjustment results, read the calculated correction values. Match each index with its spatial index in the original sequence and number them according to their spatial location. The reflectivity data frames were categorized and organized. The corrected values ​​were then remapped back to the original data structure, replacing any outdated values. The corresponding database records were updated. Values ​​from all anomalous observation directions were re-indexed into their corresponding spatial mappings to ensure strict consistency between the corrected data frame content and spatial location. By traversing all bands and angles, the entire dataset was cleaned and updated, ultimately yielding an angle-corrected reflectance value set. This set, while preserving the true texture information of the coal surface, eliminated radiation distortion caused by poor observation angles, providing a high-quality data source for subsequent image generation.

[0039] Please see Figure 6 The steps to obtain S5 are as follows: S501: Based on the reflectance values ​​corresponding to each spatial index in the set of angle-corrected reflectance values, extract the spectral values ​​of each location point according to their coordinate order, and call the spatial coordinate sequence to stitch the reflectance of each point. Arrange all reflectance values ​​in a two-dimensional structure to obtain a spatial stitched image matrix. A two-dimensional image matrix is ​​defined based on the reflectance values ​​corresponding to each spatial index in the set of angle-corrected reflectance values. Its number of rows Number of columns The determination is based on the survey area and spatial resolution. The spectral values ​​of each location are extracted according to their coordinate order, and their indices in the matrix are calculated. The spatial coordinate sequence is used to stitch together the reflectance values ​​at each point, and the corrected reflectance values ​​at each point are then combined. Fill to In the process, missing elements in the matrix (such as holes caused by scanning gaps) are filled using bilinear interpolation; for overlapping areas, mean fusion is used. All reflectance values ​​are arranged in a two-dimensional structure, generating independent grayscale matrices for each band and each observation angle. Finally, a spatially stitched image matrix is ​​obtained, which logically restores the continuous image shape of the coal surface, and each pixel value has undergone the previous angle correction to ensure spatial continuity.

[0040] S502: Based on the image pixel content of each spatial location in the spatial stitching image matrix, synchronously process the image data of the same area under different observation directions, sequentially extract the data frame content of the corresponding position in the image matrix under each direction, call the reflectance value of each direction to perform channel-level corresponding processing, and obtain the multi-directional reflectance stitching matrix. Based on the image pixel content at each spatial location in the spatially stitched image matrix, image data of the same region under different viewing directions are processed synchronously. For the same pixel location... The data frame content at corresponding positions in the image matrix from each direction is extracted sequentially. The reflectance values ​​from each direction are then processed at the channel level to calculate the fusion weights. The weight calculation consists of two parts: geometric weights. (in (for the zenith angle), and mass weight. (in (This refers to the local coefficient of variation calculated in S3). Calculate the overall weight. The weighted average formula was used to calculate the reflectivity after fusion. By employing this dual weighting mechanism based on physical geometry and data quality, redundant observation data from multiple angles are merged into a single high signal-to-noise ratio composite value, resulting in a multi-directional reflectivity composite matrix.

[0041] S503: For the image data frames in the multi-directional reflectivity stitching matrix, extract the pixel positions corresponding to each spatial index, and use spatial parameter mapping rules to map each pixel position in the image to the index number of the coal pile surface area. Fill in the image data content according to the corresponding relationship to obtain the coal surface reflectivity enhancement image result. For image data frames in a multi-directional reflectivity stitching matrix, extract the pixel positions corresponding to each spatial index and the fused reflectivity data. By using spatial parameter mapping rules (such as projection transformation matrices), the location of each pixel in the image is mapped to the index number of the coal pile surface area. Based on the correspondence, the fused reflectivity texture map is applied to the corresponding physical model mesh patch, or a geocoded orthoreflectivity map is generated. During this process, image edges are smoothed to eliminate stitching gaps. Finally, an enhanced coal surface reflectivity image is obtained, exhibiting coal surface characteristics of uniform illumination, clear details, and no obvious angular artifacts. This image can be directly used for subsequent coal quality analysis or volumetric measurement, achieving a complete transformation from data to visual information.

[0042] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for enhancing the surface reflectance of coal based on multi-angle spectral consistency constraints, characterized in that, Includes the following steps: S1: By controlling multiple unmanned aerial vehicles to scan the surface of the coal pile along different flight paths, spectral reflectance data of the covered area is collected. At the same time, the shooting orientation and flight coordinates are recorded. The images are registered and track matched to construct a coal area reflectance image group under multi-angle correspondence. S2: Based on the coal area reflectance image group under the multi-angle correspondence, extract coal surface area points with positional consistency, group image pixels in the same spatial position according to the observation angle, establish the pairing relationship between direction and reflectance value, and construct the coal surface pixel and angle reflectance data structure. S3: Based on the aggregation structure of the coal surface pixels and angular reflectance data, the reflectance values ​​at different angles are sequence-expanded and the difference is calculated to identify the band regions with large variation amplitudes and mark the corresponding observation directions, thereby constructing the identification results of the directional difference region of coal surface reflectance. S4: Based on the abnormal observation direction in the identification results of the directional difference region of coal surface reflectance, extract the relevant reflectance sequence in the original image, perform offset correction and trend sorting on the difference bands, and generate an angle-corrected reflectance numerical set.

2. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that: The coal area reflectance image set includes multi-view image data, reflectance distribution matrix, and spatiotemporal identification information. The aggregation structure of coal surface pixels and angular reflectance data includes angle numbering information, spatially corresponding pixels, and directional reflectance set. The identification results of directional difference regions of coal surface reflectance include high difference band intervals, abnormal observation direction markers, and directional change characteristics. The set of coal area reflectance values ​​after angle correction includes adjusted spectral sequences, difference band correction values, and unified directional reflectance data.

3. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that, The steps for obtaining S1 are as follows: S101: Based on the spectral image frame sequence acquired by multiple unmanned aerial vehicles during flight along different routes and attitude angles, the continuous spectral reflectance data within the coverage area is extracted frame by frame. Combined with the flight path coordinate sequence and shooting azimuth angle data acquired synchronously by the aircraft, the image frames are matched with the corresponding time series numbers during the processing. Through the matching processing of image frames with their position information, the spatiotemporal mapping dataset corresponding to the spectral image frames is obtained. S102: Based on the image time sequence number and path coordinate sequence in the spatiotemporal mapping dataset corresponding to the spectral image frame, the pairing process is executed according to the order in which the image frames appear, and the path coordinate changes between adjacent image frames are processed in correspondence with the shooting azimuth angle data. The displacement correlation information group of multi-view image frames is obtained by comparing the continuity difference of the image frames with the attitude angle data. S103: Based on the coordinate relationship content in the displacement association information group of the multi-view image frames, and according to the path value and attitude angle value relationship between adjacent image frames, the image frame content is readjusted under unified spatial parameters so that the reflection data of each image frame is processed in the same coordinate system, and finally the coal area reflectivity image group under the multi-angle correspondence is obtained.

4. The method for enhancing the surface reflectance of coal based on multi-angle spectral consistency constraints according to claim 3, characterized in that: The coordinate relationship content in the displacement association information group of the multi-view image frame is used to filter image frame pairs that satisfy both the path coordinate change range and the shooting azimuth angle change range within a preset threshold range to establish an effective matching relationship. The process of readjusting the image frame content under unified spatial parameters includes: normalizing the spatial position of the image frame based on the effective matching relationship, and performing image reprojection operation according to the pose angle change between image frames to achieve spatial alignment. The coal area reflectance image group under the multi-angle correspondence is obtained by reconstructing image frames under a unified coordinate system, and includes multiple image frames with different shooting azimuth angles that satisfy the effective matching relationship.

5. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that, The steps for obtaining S2 are as follows: S201: Based on the multi-angle correspondence, the position coordinates of the same pixel in the coal area reflectance image group under different angles are cross-filtered, the spatial reference coordinate values ​​of each image frame are called, and the spatial position index values ​​of all pixels are compared to obtain the spatial consistency index set of position points. S202: Based on the coordinate numbers of each location point in the spatial consistency index set, perform classification operations on pixels with the same coordinate numbers according to the acquisition order of the spectral image frames, call different angle number values ​​to assign corresponding values ​​to the reflectance data corresponding to each angle, and obtain the reflectance data group under the angle order. S203: For each reflectance value and direction number data in the reflectance data group under the angular order, based on the corresponding spatial coordinate number of each group, the direction number and the corresponding reflectance value of each group are structurally merged, and all spatial position index numbers are uniformly mapped to obtain the coal surface pixel angle reflectance data structure.

6. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that, The steps for obtaining S3 are as follows: S301: Based on the reflectance value sequence corresponding to each position index in the coal surface pixel angle reflectance data structure, for each set of reflectance value sequences under the observation direction, extract its numerical order at all band positions, call the sequence order data corresponding to each direction for position comparison processing, and obtain the band-corresponding reflectance sequence set; S302: Based on the same band position in the reflectance sequence set corresponding to the band, extract the reflectance values ​​under each observation direction, call the direction index order to compare the values ​​under adjacent observation directions pairwise, and number the band position sequence after each comparison according to the direction order to obtain the direction sequence band comparison result set. S303: For the numerical variation of each band in the directional sequence band comparison result set under different observation directions, monitor the degree of numerical difference in the directional sequence under each group of bands, extract and process according to the directional index corresponding to the range of difference, classify all bands with prominent range of variation and their corresponding directional index contents, and obtain the identification results of the directional difference region of coal surface reflectivity.

7. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that, The steps for obtaining S4 are as follows: S401: Based on the observation direction index contained in the identification result of the directional difference region of coal surface reflectance, extract the spectral reflectance sequence under the corresponding direction in the original reflectance image frame, call the spectral values ​​of each band position in the difference region, combine the spatial index data to locate the content of the corresponding image region, and obtain the abnormal direction reflectance band sequence. S402: For the spectral value of each band position in the abnormal directional reflectance band sequence, call the spectral value sequence of the same position in other directional sequences, compare their numerical relationships, and perform position adjustment operation on the spectral value sequence according to the order offset to obtain the band position comparison adjustment result; S403: Based on the order of the adjusted spectral values ​​in the band position comparison adjustment results, match them one-to-one with the spatial indexes in the original sequence, classify and organize the reflectance data frame content according to each spatial position number, re-classify the values ​​under all abnormal observation directions into the corresponding spatial mapping index, and obtain the angle-corrected reflectance value set.

8. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 7, characterized in that: When the spectral values ​​at each band position in the abnormal directional reflectance band sequence are compared sequentially with the spectral values ​​at the same position in other directional sequences, the position adjustment operation is only performed on spectral values ​​that meet the condition of setting a limit on the order offset range, and the order difference does not exceed the preset adjustment tolerance. In the process of matching the adjusted spectral values ​​in the band position comparison adjustment results with the spatial indices in the original sequence, the image region content with continuous spatial numbers is selected first for sorting based on the premise of spatial index continuity. When the spectral values ​​in the angle-corrected reflectance numerical set are assigned to the corresponding spatial mapping index, they are normalized according to the distribution trend of spectral values ​​of different directional sequences at the same spatial location, so as to eliminate the interference of observation direction differences on reflectance data.

9. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 1, characterized in that, The method further includes: S5: Based on the set of coal area reflectance values ​​after angle correction, the reflectance values ​​at each location are stitched together into an image matrix, all directional data are fused and mapped to spatial coordinates to construct an enhanced coal surface reflectance image result. The enhanced reflectance image of the coal surface includes a fused reflectance image, a corrected image matrix, and a spatial coordinate mapping result.

10. The method for enhancing the reflectivity of coal surface based on multi-angle spectral consistency constraints according to claim 9, characterized in that, The steps for obtaining S5 are as follows: S501: Based on the reflectance values ​​corresponding to each spatial index in the set of angle-corrected reflectance values, extract the spectral values ​​of each location point according to their coordinate order, and call the spatial coordinate sequence to stitch the reflectance of each point. Arrange all reflectance values ​​in a two-dimensional structure to obtain a spatial stitched image matrix. S502: Based on the image pixel content of each spatial position in the spatial stitching image matrix, synchronously process the image data of the same area under different observation directions, sequentially extract the data frame content of the corresponding position in the image matrix under each direction, call the reflectance value of each direction to perform channel-level corresponding processing, and obtain a multi-directional reflectance stitching matrix. S503: For the image data frames in the multi-directional reflectivity stitching matrix, extract the pixel positions corresponding to each spatial index, and map each pixel position in the image to the index number of the coal pile surface area through spatial parameter mapping rules. Fill in the image data content according to the correspondence to obtain the coal surface reflectivity enhancement image result.

Images