A preliminary identification method and system for the quality of coal-bearing rare metal ore in the field

Abstract

This invention relates to the field of ore identification technology, specifically to a method and system for preliminary identification of the quality of rare metal coal-bearing ores in the field. The method includes the following steps: collecting ore samples and acquiring main peak, structural, and infrared images; extracting image features to generate mutation numbers; analyzing sampling coordinates to reconstruct the connectivity structure; drawing closed boundaries to determine coordinate attribution; and outputting an ore quality identification scheme. In this invention, by uniformly numbering and temporally binding the acquisition results of main peak reflection, microscopic layers, and infrared spectrum bands, a synchronous correspondence between multi-source images is established. By combining offset trajectories and structural features to identify continuously changing nodes and abrupt change regions, spatially concentrated patches are generated using numbering and coordinate mapping, and the connectivity structure is reconstructed. Closed boundaries are drawn based on trajectories to form the basis for regional classification. The direction of quality change is verified through image sequence coherence, and labeled identification results are output, achieving rapid positioning and accurate classification of field ore samples.

Description

Technical Field

[0001] This invention relates to the field of ore identification technology, and in particular to a method and system for preliminary identification of the quality of coal-based rare metal ores in the field. Background Technology

[0002] The field of ore identification technology involves the identification and analysis of ore samples obtained during geological exploration, focusing on their composition, structure, and quality. Its core aspects include observation of the ore's macroscopic appearance, microstructural analysis, determination of its physical and chemical composition, and rapid identification and classification of target elements or mineral content. This technology is widely used in field geological surveys, mineral resource exploration and development, and is one of the key foundations for achieving efficient development and scientific assessment of mineral resources. Traditional ore identification methods mainly rely on field geologists to initially determine the ore type through experience, manual comparison, and observation of color or texture. Subsequently, portable instruments are used for component testing, or samples are sent to laboratories for detailed testing such as chemical and spectral analysis. Traditional methods for preliminary quality identification of coal-bearing rare metal ores in the field refer to the means of making preliminary judgments on the quality grade of ores containing lithium, scandium, and rare earth elements (REEs) and occurring in coal seams during field sampling. This typically involves judging the density, color intensity, and surface weathering degree of the ore through visual observation and manual touch, combined with simple chemical titration or portable fluorescence instruments to make preliminary estimates of the presence and approximate content of rare elements in the ore. This process often employs a preliminary judgment model based on the coexistence of characteristic minerals, the stratigraphic environment, and the characteristics of the coal-bearing surrounding rocks to quickly classify the industrial utilization value grade of the ore on-site.

[0003] In existing technologies, ore identification often relies on a single perceptual dimension to acquire information. The types of sampled data are scattered and lack a unified time series binding mechanism, making it difficult to establish correlations between various types of image and signal data. This makes it impossible to form a systematic linkage between structure, composition, and spatial distribution. It is difficult to capture continuously changing critical paths in the map, and structural abrupt changes and quality change trends cannot be effectively located. The on-site acquisition process cannot achieve synchronous offset judgment between multiple image data, resulting in a lack of unified standards for quality grade classification. Field identification results are prone to problems such as ambiguous positioning and distorted label judgment. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a method for preliminary identification of the quality of coal-based rare metal ore in the field, comprising the following steps:

[0005] S1: Collect coal-bearing rare metal ore samples in the field, use a spectrometer to obtain the main peak reflection signal, collect microscopic images to establish a particle structure layer, record the changes in infrared absorption sections simultaneously, and generate a numbered list.

[0006] S2: Call the number list, extract the image main peak layer to compare the peak position direction, identify the broken connected area at the edge of the microscopic image particles, analyze the infrared spectral band shift trajectory, and generate abrupt image number;

[0007] S3: Based on the mutation image number, extract the sampling coordinates, analyze the distribution density within the image area, identify regions with consecutive and concentrated numbers, remove isolated markers, reconstruct the connection structure, and mark the turning point numbers.

[0008] S4: Based on the turning point number, draw a closed boundary, mark the sampling point number block within the boundary, determine whether the coordinates of the ore to be tested are within the boundary, if not, trace the path of the neighboring number change and delineate the reference boundary number.

[0009] S5: Call the reference boundary number to determine whether the sample number and the boundary map group are connected and have the same trend. If they are consistent, mark it as a change label; otherwise, classify it as an adjacent area label and output the field ore quality identification scheme.

[0010] As a further aspect of the present invention, the numbering list includes the main peak reflection number, particle layer number, and infrared spectral band number; the abrupt change image number includes the peak position shift image, particle edge image, and infrared deformation image; the turning point number includes the structural abrupt change point, the numbering concentration point, and the connection inflection point; the reference boundary number includes the labeled trajectory line, the closed boundary line, and the numbering block distribution; and the field ore quality identification scheme includes the change trend label, the connectivity label, and the area identification label.

[0011] As a further aspect of the present invention, the removal of isolated markers refers to deleting sampling markers that are spatially scattered and unrelated to the main connected region in the image numbering.

[0012] As a further aspect of the present invention, the change label refers to the characteristic identification mark assigned to the sample when the sample number and the boundary map group are consistent in terms of connectivity and trend.

[0013] As a further aspect of the present invention, the specific steps of S1 are as follows:

[0014] S101: Based on coal-bearing rare metal ore samples obtained from field exploration, a spectrometer is used to scan the sample surface, extract the main peak position and corresponding reflectance of the band in the reflection spectrum, and generate a set of main peak reflection signal parameters.

[0015] S102: Based on the main peak reflection signal parameter set, acquire a microscopic image of the corresponding region of the sample, extract the particle boundary, morphological features and texture direction in the image, and generate a particle structure layer data matrix.

[0016] S103: Call the infrared acquisition device to obtain the infrared absorption section of the particle structure layer data matrix region, extract the absorption peak position and bind it with the main peak reflection signal parameter set and the particle structure layer data matrix according to the region number, and obtain the number list.

[0017] As a further aspect of the present invention, the specific steps of S2 are as follows:

[0018] S201: Call the number list, extract the image numbers in sequence, monitor the wavelength change of the main peak position in adjacent images for the main peak layer recorded in the image corresponding to the number, and calculate whether the offset direction is consistent, and obtain the main peak offset direction sequence;

[0019] S202: Based on the main peak offset direction sequence, extract the particle structure layer data in the corresponding microscopic image, judge the contour curvature change of the edge region, identify the layer region with broken boundaries and connected structure, and generate a set of broken connected region numbers.

[0020] S203: Call the image number in the set of broken connected regions, extract the spectral data in the corresponding infrared absorption section, compare the trend of the absorption peak position with the image sequence, determine whether there is a continuous shift in the same direction as the main peak, and filter and mark the image number that meets the requirements to obtain the abrupt image number.

[0021] As a further aspect of the present invention, the specific steps of S3 are as follows:

[0022] S301: Based on the mutation image number, extract the field sampling coordinate point corresponding to each number, map all coordinate points to the map grid, and count the number density in each grid cell. Calculate the spacing between adjacent numbers on the map and arrange them in combination to obtain a set of spatially dense number blocks.

[0023] S302: Based on the spatially dense numbered block set, remove the numbered points on the map that have no connection relationship with any block, and construct a connection structure for the remaining coordinate points according to the spatial distribution relationship. Then, perform aggregation and division processing on all connection relationships to obtain a reconstructed continuous numbered structure map.

[0024] S303: Call the number sequence of the connected structural units in the reconstructed continuous numbering structure diagram, compare the directional change gradient of each number point in the two-dimensional coordinate system in sequence, identify the number position where the angle change is greater than the set angle change threshold, and record the number to obtain the turning position number.

[0025] As a further aspect of the present invention, the specific steps of S4 are as follows:

[0026] S401: According to the numbering of the turning point, connect adjacent coordinate points in the map in numerical order, monitor whether the geometric shape formed by the connecting path forms a closed loop, and if it forms a closed loop, mark the contour of the closed path formed by all numbered points to obtain the closed boundary graphic path set.

[0027] S402: Call the closed boundary graphic path set, extract the numbers of all sampling coordinate points within the boundary range, cluster and integrate the number distribution area, define the number belonging range within the closed boundary according to the aggregation relationship, and establish a sampling number block table within the boundary;

[0028] S403: Based on the sampling number block table within the boundary, compare whether the coordinates of the ore to be tested are located inside any closed path. If not, extract the numbered points adjacent to the coordinates, and retrieve the trend of structural direction change in the connected path. Select consecutive numbered points along the direction of change to construct the trajectory edge line and obtain the reference boundary number.

[0029] As a further aspect of the present invention, the specific steps of S5 are as follows:

[0030] S501: Call the reference boundary number, extract the synchronous image number of all samples to be tested, arrange the synchronous image number in the order of acquisition, compare the position with the consecutive number order in the reference boundary, filter the image number that is sequentially connected, and obtain the connected matching image number set.

[0031] S502: Based on the connected matching image number set, retrieve the corresponding main peak layer offset trend sequence, compare it point by point with the offset trend of the image number in the reference boundary number, determine whether the change direction is consistent, and assign a label value to the image number that simultaneously satisfies connectivity and direction consistency to obtain the change label number sequence.

[0032] S503: Based on the changed label number sequence, number the unlabeled synchronous image, retrieve the neighboring number structure region in the number space, find the number belonging label of the region, perform neighborhood belonging judgment and label assignment processing on all numbers without connectivity, integrate all number and label content, and output the field ore quality identification scheme.

[0033] A preliminary quality identification system for coal-bearing rare metal mineral deposits in the field, comprising:

[0034] The sampling and data initialization module is used to realize S1: collecting coal-series rare metal ore samples in the field, calling the spectrometer to obtain the main peak reflection signal, acquiring microscopic images to establish a particle structure layer, synchronously recording the changes in infrared absorption sections, uniformly numbering and generating a number list;

[0035] The anomaly identification and feature extraction module is used to implement S2: calling the number list, extracting the image main peak layer to compare the peak position direction, identifying the broken connected regions at the edge of particles in the microscopic image, analyzing the infrared spectral band shift trajectory, and generating abrupt image numbers;

[0036] The spatial structure reconstruction module is used to implement S3: based on the mutation image number, extract the sampling coordinates, analyze the distribution density within the map, identify the area map blocks with continuous and concentrated numbers, remove isolated punctuation marks, reconstruct the connection structure, and mark the turning position numbers;

[0037] The boundary drawing and position determination module is used to implement S4: draw a closed boundary according to the turning position number, mark the sampling point number block within the boundary, determine whether the coordinates of the ore to be measured are within the boundary, if not, trace the path of the neighboring number change and delineate the reference boundary number;

[0038] The label attribution and identification output module is used to implement S5: call the reference boundary number, determine whether the sample number and the boundary map group are connected and have the same trend. If they are consistent, mark it as a change label; otherwise, classify it as an adjacent area label and output the field ore quality identification scheme.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0040] In this invention, by uniformly numbering and temporally binding the acquisition results of the main peak reflection, microscopic layer, and infrared spectrum, a synchronous correspondence between multi-source images is established. By combining the offset trajectory and structural features, continuously changing nodes and abrupt change regions are identified. Spatial concentrated patches are generated using numbering and coordinate mapping, and the connected structure is reconstructed. Closed boundaries are drawn based on the trajectory to form the basis for regional classification. The direction of quality change is verified by the sequential coherence of the images, and labeled recognition results are output, thereby achieving rapid positioning and accurate classification of field mineral samples. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic diagram of the steps of the present invention;

[0043] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0044] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0045] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0046] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0047] Figure 6 This is a detailed schematic diagram of S5 of the present invention;

[0048] Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0050] 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.

[0051] 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.

[0052] 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.

[0053] 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.

[0054] Please see Figure 1 This invention provides a method for preliminary identification of the quality of coal-bearing rare metal ore in the field, comprising the following steps:

[0055] S1: At the field exploration site, obtain coal-bearing rare metal ore samples, call the spectral acquisition to capture the main peak reflection signal, collect microscopic images to establish a particle structure layer, and simultaneously call the infrared light acquisition device to record the absorption section change pattern. Bind the three types of acquisition results to a unified time sequence and set a numbering sequence to obtain a numbering list.

[0056] S2: Call the number list, extract the main peak layer in the consecutive image numbers and compare the peak position movement direction, extract the particle edge morphology in the microscopic image to identify the broken connected area, analyze the deformation and shift trajectory of the spectral band in the infrared absorption section, determine whether there is a continuous shift in the same direction, then mark the numbered images that have synchronous shifts, and identify the abrupt image numbers.

[0057] S3: Based on the mutation image number, extract the field sampling coordinates pointed to by the number, arrange the distribution density of the coordinate points on the map, identify the area patches with consecutively arranged numbers and concentrated spacing, delete the spatially isolated number punctuation marks and reconstruct the connection structure, identify the continuous distribution area according to the spatial concentration trend, and mark the turning point number.

[0058] S4: Based on the turning point number, draw a closed boundary figure along the trajectory between the numbers on the map sheet, mark the sampling point number block within the closed figure, determine whether the coordinates of the ore to be measured are located in the boundary-enclosed area, if not, trace the structural change path presented by the adjacent numbers, and delineate the reference boundary number.

[0059] S5: Call the reference boundary number and compare whether the synchronous image number of the sample has a sequential connection relationship with the boundary number map group. If there is a connection and the change trend is consistent, it is marked as a change label category. If there is no corresponding relationship, it is classified into the adjacent area identification label and the field ore quality identification scheme is output.

[0060] The numbering list includes the main peak reflection number, particle layer number, and infrared spectral band number. The abrupt change image number includes peak position shift image, particle edge image, and infrared deformation image. The turning point number includes structural abrupt change point, number concentration point, and connection inflection point. The reference boundary number includes labeled trajectory line, closed boundary line, and numbered block distribution. The field ore quality identification scheme includes change trend label, connectivity label, and area identification label.

[0061] Please see Figure 2 The specific steps of S1 are as follows:

[0062] S101: Based on coal-bearing rare metal ore samples obtained from field exploration, a spectrometer is used to scan the sample surface, extract the main peak position and corresponding reflectance of the band in the reflection spectrum, and generate a set of main peak reflection signal parameters.

[0063] First, the collected ore samples are labeled with numbers on-site, and their corresponding geographical location, collection time, and rock stratum depth are recorded. If on-site conditions permit, portable drying and cleaning equipment can be used to preliminarily treat the sample surface to remove moisture, dust, and oil stains, ensuring the sample surface is clean and free of foreign matter. Then, the sample is placed in the standard tray of the spectrometer, the start and end positions of the collector scan are set, and the device is turned on to perform a point-by-point linear scan along the sample surface. The scan interval is set to 0.5 mm. Reflectance spectral data in the range of 400 nm to 2500 nm are acquired at each sampling point. The reflectance spectral data is the wavelength corresponding to the reflectance. A set of values ​​is generated, and then the main peak identification operation is performed on each set of spectral data. During the identification process, it is determined whether the reflectance of each wavelength point is higher than any five of its six adjacent wavelength points. At the same time, it is determined whether its reflectance value exceeds the threshold determined by the background reflectance level. For example, if the average reflectance of the background area is 0.06 and the standard deviation of fluctuation is 0.045, then the main peak threshold can be taken as the average value plus twice the standard deviation, i.e., 0.06 plus 0.09, which is 0.15. The main peak wavelength position and its corresponding reflectance value are recorded as a set of main peak information. After traversing all sampling points, the reflection signal information of each main peak is sorted out and summarized together with the sample number and geographical information to generate the main peak reflection signal parameter set.

[0064] S102: Based on the main peak reflection signal parameter set, acquire microscopic images of the corresponding region of the sample, extract particle boundaries, morphological features and texture direction from the images, and generate a particle structure layer data matrix.

[0065] Under field conditions, portable microscopic imaging equipment was used to acquire images of the corresponding numbered ore samples. The acquisition area should cover a range extending 2 mm outward from the main peak reflection position to ensure complete particle boundary representation. The image resolution was set to 0.5 micrometers per pixel. After image acquisition, grayscale conversion was performed, and median filtering was used to reduce image noise. Then, an image edge recognition tool was used to extract the image boundaries. During edge extraction, boundary segments were detected based on pixel gradient changes. Closed particle structures were formed by continuous boundary segments. The number of pixels and boundary length within the closed region of each particle were calculated and converted into particle area and perimeter, respectively. The principal axis direction was extracted by comparing the pixel lengths of the long and short sides, and then the particle area and perimeter were calculated based on the principal axis direction. The principal orientation angle of the particles is calculated, and the shape ratio of the particles is also calculated based on the area and perimeter data to evaluate the regularity of the particles. Particles with a shape factor higher than 0.85 are classified as regular shapes, those lower than 0.65 are classified as irregular shapes, and those in between are classified as sub-regular shapes. The image is then divided into multiple 64x64 pixel sub-regions, and the principal orientation parameters of the texture are extracted from each region in turn. The main extension direction of the texture is determined by analyzing the angle of grayscale direction change. The texture directions of all sub-regions in the image are then stitched and superimposed in the form of projected coordinates to generate the texture direction map of the entire image. Finally, the boundary coordinates, area, shape parameters, and principal orientation of each particle are numbered and archived, and summarized into a particle structure layer data matrix.

[0066] S103: Call the infrared acquisition device to obtain the infrared absorption section of the particle structure layer data matrix area, extract the absorption peak position and bind it with the main peak reflection signal parameter set and the particle structure layer data matrix according to the area number, and obtain the number list;

[0067] First, the sampling area was located one by one according to the particle number recorded in the image. Under field conditions, a simplified positioning method was used to adjust the sample orientation so that the target area was aligned with the infrared detector for data acquisition. The infrared spectrum acquisition time for each particle was set to 5 seconds, and the acquisition band range was set to 400 to 4000 wavenumbers. After the acquisition started, the system recorded the data curve of the absorbance as a function of wavenumber in the absorption spectrum. The data curve was traversed and processed to determine whether each wavenumber point was a local absorption peak. The judgment criteria were that the absorbance of the current wavenumber point was less than the absorbance of the three points to its left and right, and its absorbance was lower than the set absorption peak threshold. For example, the average background absorbance of the sample was 0.91, and the standard deviation of the fluctuation was [not specified]. If the value is 0.03, the absorption peak threshold can be taken as 0.91 minus twice the standard deviation, i.e., 0.85. If the absorptivity of a certain wavenumber point is 0.82 and meets the local minimum criterion, it is recorded as the absorption peak position. This information will be matched one by one with the main peak reflection parameters, particle image number, area, morphological features, texture direction, etc. of the particle. The matching conditions are that the image number is consistent and the distance between the particle center point and the center of the main peak region is less than 1 mm. If these conditions are met, they are bound into a complete data unit. Finally, all the matched data are generated into a unified number list according to the particle number. Each record contains information such as number, main peak wavelength, main peak reflectivity, particle area, shape features, texture direction, absorption peak position, and absorptivity.

[0068] Please see Figure 3 The specific steps of S2 are as follows:

[0069] S201: Call the number list, extract the image numbers in sequence, monitor the wavelength change of the main peak position in adjacent images for the main peak layer recorded in the image corresponding to the number, calculate whether the offset direction is consistent, and obtain the main peak offset direction sequence;

[0070] During extraction, the position of the main peak layer, the wavelength value of the main peak, and the acquisition order corresponding to each image number are recorded as the current processing object. Then, the preceding and following image numbers adjacent to the image number are retrieved. By calling the main peak information stored in the number list, the wavelength value of the main peak at the corresponding spatial position in the adjacent image is found. Using the current image's main peak wavelength value as a comparison benchmark, the wavelength values ​​of the main peaks in the adjacent images are compared numerically with the current value one by one. If the wavelength of the main peak in the adjacent image is greater than that in the current image, the offset direction of this comparison is recorded as a positive offset; if the wavelength of the main peak in the adjacent image is less than that in the current image, it is recorded as a negative offset; if the two are equal, it is recorded as a zero offset. The above operation is repeated continuously during the recording process to form the main peak offset of each image. The system records the offset direction of each image. For example, if the main peak wavelength of image A is 620 nm and the corresponding main peak wavelength of the adjacent image B is 630 nm, then the offset direction is recorded as positive. If the main peak wavelength of the adjacent image C is 610 nm, then the offset direction is recorded as negative. Then, when processing multi-peak images, the same comparison process is performed on each main peak in ascending order of wavelength, and the offset direction of each main peak in the adjacent images is recorded separately. The offsets between different main peaks are not cross-substituted and are kept independently recorded. Then, all offset directions are arranged in the order of image numbers to form a direction sequence. The order of the offset directions between consecutive numbers in this sequence is recorded. Finally, the main peak offset direction sequence corresponding to each image number is obtained through continuous comparison, recording and sorting.

[0071] S202: Based on the main peak offset direction sequence, extract the particle structure layer data in the corresponding microscopic image, judge the contour curvature change of the edge region, identify the layer region with broken boundaries and connected structure, and generate a set of broken connected region numbers.

[0072] Image numbers showing abrupt shifts or large shifts (e.g., wavelength changes exceeding 10 nanometers) in the sequence are extracted. The corresponding microscopic images are loaded, and the particle structure layer data matrix for each image is called. The edge regions of the particles are located, and the pixel coordinates of the edges are read point by point. The local curvature value is determined by the magnitude of the angular change between three consecutive pixels. If the angular change exceeds a set curvature change threshold of 30 degrees, the point is marked as a curvature abrupt change point. After organizing all curvature abrupt change points in the image, their distribution on the same particle boundary is examined. If the abrupt change points are continuously and densely packed on the boundary and their length exceeds 10% of the total boundary length, a significant breakage feature is identified at that location. Simultaneously, the connected regions of the particle in the image are read. Information is used to determine whether the two ends of a broken region are still connected to the same structural region based on pixel connectivity conditions. If multiple paths connect the two ends to the same region, such as two boundary segments returning to the same connected domain via more than three connected paths, then this is identified as a connectivity break region. The image number and the corresponding particle number are combined and recorded. When multiple images show similar curvature anomalies and connectivity structures in their corresponding regions, these numbers are written one by one. For example, if particle number 8 of image number A shows 14 consecutive curvature abrupt changes in edge detection, accounting for about 12% of the total pixel length of the boundary, and both ends of the break are connected to the pixel paths corresponding to the same connected domain, then number A-8 is recorded in the number set. This screening method is continuously executed to finally generate a set of broken connectivity region numbers.

[0073] S203: Call the image number in the set of broken connected regions, extract the spectral data in the corresponding infrared absorption section, compare the trend of the absorption peak position with the image sequence, determine whether there is a continuous shift in the same direction as the main peak, and filter and mark the image number that meets the requirements to obtain the abrupt image number.

[0074] The infrared absorption segment data corresponding to each image number is read one by one. After reading the image band, the infrared absorption segment data corresponding to the broken area in the number set is obtained. The position of the absorption peak in the infrared absorption segment is used as the comparison content. The absorption peak wave count is recorded item by item from the first image in the order of the image sequence number. The absorption peak wave count of the first image is used as the reference value. Then, the absorption peak wave count at the corresponding position in the next image is directly compared with the reference value. If the latter is greater than the former, it is recorded as a positive shift; if the latter is less than the former, it is recorded as a negative shift; if they are equal, it is recorded as zero shift. The above comparison action is repeated to form an absorption peak shift direction sequence. For example, the absorption peak of image number A is 14. 50. If B is numbered 1460 and C is numbered 1472, then the three images have consecutive positive shifts. Next, the shifts of the absorption peaks are compared one by one with the shifts of the main peaks obtained in the previous steps. When both shifts are positive, negative, or zero, they are recorded as images with consistent orientation. When three or more images have consistent orientations, the shift trends of these images are checked to find the position where the shifts change in the opposite direction. For example, if the first two images have positive shifts and the third image has a negative shift, then the third image number is recorded as a mutation image. Finally, the mutation image numbers are obtained by filtering and marking all images that meet the criteria of consistent orientation and have undergone directional mutations in the sequence.

[0075] Please see Figure 4 The specific steps of S3 are as follows:

[0076] S301: Based on the mutation image number, extract the field sampling coordinate point corresponding to each number, map all coordinate points to the map grid, and count the number density in each grid cell. Calculate the spacing between adjacent numbers on the map and arrange them in combination to obtain a set of spatially dense number blocks.

[0077] The field sampling coordinates corresponding to each image number are extracted sequentially from the original records. This information includes parameters such as the latitude and longitude coordinates recorded during sampling, the unique sampling point number, and the map sheet number to which it belongs. All extracted coordinate data are uniformly converted to a unified coordinate system such as WGS84 or UTM. After conversion, all points are projected onto a standard map sheet grid. The grid is divided according to a preset size, typically using a square grid with a side length of 50 meters. All coordinate points are traversed, and the grid number of each point is determined. The number of projected points in each grid is counted, and the grid number and the number of aberration image numbers it contains are recorded during the counting process. All grid numbers and counts are then summarized to form a complete map sheet grid. The numbering density distribution table is used, and cells with 3 or more numbers in the grid are defined as densely numbered cells. Then, the positions of all numbered points in the map coordinate system are combined and arranged. The horizontal straight-line distance on the map is calculated for each pair of points. If the straight-line distance between two points is less than 150 meters, they are considered to be close in space and recorded as a pair of densely connected numbered pairs. All point pairs that meet the conditions are traversed and combined and classified to form multiple interconnected and closely spaced numbered subsets. These numbered subsets are demarcated according to geographical distribution to form multiple spatially densely numbered blocks. Each block contains several adjacent or overlapping numbered points, and the block number and the number of numbers within the block are recorded. Finally, the set of spatially densely numbered blocks is output.

[0078] S302: Based on the set of spatially dense numbered blocks, remove the numbered points on the map that have no connection relationship with any block, and construct a connection structure for the remaining coordinate points according to the spatial distribution relationship. Then, perform aggregation and division processing on all connection relationships to obtain a reconstructed continuous numbering structure map.

[0079] The entire original mutation image is traversed. For each number, it is determined whether it belongs to any identified spatial numbering block. If a number does not belong to any block, it is marked as a disconnected number, removed, and its coordinates are taken out of the subsequent processing. Then, the remaining numbered points are spatially sorted according to their position in the map. Each numbered point is connected to its adjacent numbered points in order of minimum coordinate distance to form a line segment structure. A connection relationship table between adjacent numbers is established, with the number pair as the smallest unit. A complete spatial connection path structure is gradually constructed. For cases where multiple connectable paths exist... Prioritizing the shortest path, if multiple points form a closed structure around a common number, all connections are retained. Based on this, line segments in the connection graph are merged and categorized. Duplicate numbered points in the path structure are merged, and the direction of the path segments is uniformly adjusted to maintain consistency. Subsequently, all connection paths are aggregated and grouped according to the extensibility and connectivity of the connecting line segments. Several short paths are merged into a complete path structure graph according to the continuity of the numbering and the stability of the connection structure. Finally, the reconstructed continuous numbering structure graph composed of the starting point number, ending point number, intermediate numbered point sequence and their spatial geometric positions connected by each path is output.

[0080] S303: Call the number sequence of the connected structural units in the reconstructed continuous numbering structure diagram, compare the directional change gradient of each number point in the two-dimensional coordinate system in sequence, identify the number position where the angle change is greater than the set angle change threshold, record the number, and obtain the turning position number.

[0081] The system sequentially obtains the two-dimensional coordinates of each number. It then extracts the coordinates of every three consecutive numbered points as an analysis unit, using each group of three points as a basis. It reads the coordinates of the current three points and constructs two direction vectors for the preceding and following segments. By comparing the angle changes between these two direction vectors, it determines the path's turning points. If the angle change between the two segments exceeds a set angle abrupt change threshold, the middle number in that group is recorded as a turning point. The standard angle abrupt change threshold is 45 degrees. The angle is determined using the absolute value of the angle change between the two vectors. During the analysis, the number window is slid sequentially, moving one number point at a time, continuously traversing the entire number sequence. When encountering multiple consecutive angle abrupt change points, record the abrupt change point number and simultaneously record the number of the point before and after it as a reference. If the two directions before and after a certain number point form a broken line structure in the coordinate plane, and the direction change exceeds the threshold, it is determined to be a turning point number. For example, the number points A, B, and C correspond to the plane coordinates (100, 100), (120, 120), and (130, 100) respectively. The direction from A to B is due northeast, while the direction from B to C is biased towards due east, forming a nearly 90-degree bend in the middle, which meets the abrupt change judgment criteria. Therefore, the number B is recorded as the turning point. Finally, integrate all the number points that meet the abrupt change conditions and output a complete list of turning point numbers.

[0082] Please see Figure 5 The specific steps of S4 are as follows:

[0083] S401: Based on the turning point number, connect adjacent coordinate points in the map sheet in numerical order, monitor whether the geometric shape formed by the connecting path forms a closed loop, if a loop is formed, mark the contour of the closed path formed by all numbered points, and obtain the closed boundary graphic path set.

[0084] First, read the sequence of numbers marking all turning points, and extract the corresponding two-dimensional coordinate points in the order they are recorded in the structure diagram. Connect these coordinate points in the map coordinate system according to their numbering order, generating connecting line segments one by one. Construct a straight line segment with every two adjacent numbered points as endpoints. Multiple consecutive line segments form a path structure. After the path is connected, determine whether the first and last points of the path coincide in the coordinate system. The criterion is whether the Euclidean distance between the two points is less than the closure threshold. The closure threshold is set according to the map scale, and a commonly used value is within 5 meters. If the distance between the two points is less than this threshold, the path is considered to have formed a closed structure with the beginning and end connected. At the same time, check whether there are self-intersecting or intersecting paths in the middle segment. If there are no self-intersecting or intersecting paths, the path is considered to have formed a closed structure with the beginning and end connected. If the paths intersect, the path satisfies the closed contour requirement. The path is marked as a closed path, and its starting point number, ending point number, all intermediate points passed through, and a list of coordinate points are recorded. In the generated closed path structure, all points are further checked to see if they overlap with the previous path segment to avoid misidentification of duplicate structures. After confirming that the structure is closed, all numbered points that make up the path are extracted, and the graphic path formed by combining these numbered points is stored in the closed path graphic set. For example, if a path is from numbered point A→B→C→D→A, and the distance between point A and the last connecting point is 3.2 meters, it is judged as a closed structure, the path number is ABCDA, and its structural information is included in the closed boundary graphic path set.

[0085] S402: Call the closed boundary graphic path set, extract the numbers of all sampling coordinate points within the boundary range, cluster and integrate the number distribution area, define the number belonging range within the closed boundary according to the aggregation relationship, and establish a sampling number block table within the boundary;

[0086] Each closed path is processed one by one, and its boundary 2D coordinate point set is extracted. This set is then used to construct a polygonal closed structure. The 2D coordinates of all sampling points are sequentially projected onto this structure in the map coordinate system. It is determined whether each sampling point falls inside any closed boundary. The criterion is whether the number of ray intersections between the point and the polygon is odd. If it is odd, the point is inside the closed structure. All sampling points that meet this condition are recorded, forming an initial set of sampling numbers inside the boundary. The points in the initial set are then divided according to their coordinate density. An area with at least 4 points within a 20-meter radius can be defined as a [area name missing]. In the clustering unit, during the clustering process, it is determined whether the connecting edges between numbered points and other numbered points are continuous. If there are multiple continuous connecting segments, these points are grouped into the same cluster group. The closed structure formed by the boundary of each group of numbered points is taken as the geometric boundary of the cluster group, and the list of numbers contained within it is recorded to form the numbering range. Then, all cluster groups and their boundary ranges are integrated to construct a sampling number block table within the boundary. The table records the number of each block, the list of sampling point numbers contained within it, and its boundary coordinates. For example, the boundary formed by the numbered path ABCDA contains numbered points 11, 13, 14, 15, and 16. After aggregation, the block number is Z1, and this information is stored in the block table.

[0087] S403: Based on the sampling number block table within the boundary, compare whether the coordinates of the ore to be tested are located inside any closed path. If not, extract the numbered points adjacent to the coordinates, and retrieve the trend of structural direction change in the connected path. Select consecutive numbered points along the direction of change to construct the trajectory edge line and obtain the reference boundary number.

[0088] Obtain the spatial coordinates corresponding to the ore to be tested, and determine whether these coordinates fall within the boundary area defined by any closed path. The criterion is whether the projection of the coordinates onto the map coordinate system satisfies the criteria for points inside the boundary, i.e., whether the number of ray intersections within any closed polygon structure is odd. If not within any closed area, the coordinates are determined to be outside the boundary. Next, extract the nearest sampling number point to these coordinates, determine its number information in the structural path, find the connecting path structure to which this number point belongs, obtain its adjacent connected number points, and record the direction change of this connecting path segment. The direction change is calculated by the angle difference between adjacent numbered coordinate points. For example, the direction from numbered point A to B is northeast, while the direction from B to... The direction C is southeast, indicating a continuous offset. Using this offset direction as the basis for path extension, 3 to 5 consecutive numbered points are selected in the subsequent directions of the path according to their numbers to form a trajectory edge line structure. This trajectory line segment reflects the extension direction and structural orientation of the path. The set of numbered points contained in this line segment is recorded as the reference boundary structure number, which is the reference path line number corresponding to the coordinate to be measured outside the boundary. For example, if the point to be measured P is not inside the closed structure Z1, its nearest point is number 13. The connecting path of number 13 extends to numbers 14, 15, and 16 in the same direction. Numbers 13 to 16 are selected to form the trajectory edge line, and numbers 13, 14, 15, and 16 are marked as the reference boundary numbers of this coordinate.

[0089] Please see Figure 6 The specific steps of S5 are as follows:

[0090] S501: Call the reference boundary number, extract the synchronous image number of all samples to be tested, arrange the synchronous image number in the order of acquisition, compare the position with the consecutive number order in the reference boundary, filter the image number that is sequentially connected, and obtain the connected matching image number set.

[0091] Extract the corresponding number sequence from the established trajectory edge lines. Each number corresponds to a synchronously acquired image number. Then, call the set of synchronous image numbers for all samples to be tested, and sort all numbers in ascending order according to the image acquisition time or the timestamp recorded by the image acquisition device to construct an image time series. Next, using the image number order in the reference boundary number as the matching benchmark, compare each number in the image number sequence to be tested. The comparison method is a point-by-point sliding window comparison. For example, if the reference numbers are 101, 102, 103, and 104, group the four adjacent image numbers in the synchronous image sequence and determine whether they are a continuous subset from 101 to 104. If they exist... If a combination of images is completely continuous or the numbering interval is less than 2, then the image group is considered to be connected to the reference path. The image numbers in the combination are extracted and stored in the connected candidate set. Further, the numbering interval of each numbering sequence segment in the candidate set is judged by difference. If the difference between two adjacent image numbers in the sequence exceeds 5, they are considered not connected and removed from the group. Finally, all image numbers that meet the requirements of continuous numbering order, continuous spacing, and consistency with the image numbers of the reference boundary path are selected. For example, if the reference numbers are 150 to 155, and the image numbers to be tested contain 152, 153, 154, and 155 in the same order, then this sequence is extracted as a set of data in the connected matching image number set.

[0092] S502: Based on the connected matching image number set, retrieve the corresponding main peak layer offset trend sequence, compare it point by point with the offset trend of the image number in the reference boundary number, determine whether the change direction is consistent, and assign a label value to the image number that simultaneously satisfies connectivity and direction consistency, to obtain the change label number sequence.

[0093] For each image number, its main peak layer information is retrieved sequentially, and the trend of the main peak wavelength change with the image sequence is extracted. The direction of the main peak wavelength change between any two adjacent images in the image number set is determined. If the main peak wavelength of the later image number is higher than that of the earlier image number, the direction is marked as rising; otherwise, it is marked as falling; if they are equal, it is marked as stable, forming a complete main peak offset trend sequence. Then, the main peak offset trend sequence of the image numbers in the reference boundary number is used as the comparison standard, and comparison is performed point by point according to the image number order. If the offset directions of the corresponding positions are consistent, that is, both directions are rising, falling, or stable, it is recorded as a match. Any inconsistency is recorded as a mismatch. For the same image number, if it is consistent with the reference image number in terms of connectivity and offset direction, then the image number is assigned a specific tag value, such as "1" to indicate consistent change. For image numbers with inconsistent offset directions, no value is assigned or the value is "0". For example, if the numbers 210, 211, and 212 are continuous in the connected structure and the main peak change is 600→612→620, which is consistent with the reference path change 600→612→620, then all three image numbers are assigned the value "1", forming a change tag number sequence.

[0094] S503: Based on the changing label number sequence, number the unlabeled synchronous images, retrieve the neighboring number structure regions in the number space, find the number belonging label of the region, perform neighborhood belonging judgment and label assignment processing on all unconnected numbers, integrate all number and label content, and output the field ore quality identification scheme.

[0095] Image IDs not marked as "1" in the sequence are extracted as unassigned samples. For each ID, its location region is searched in the ID space structure. Within this region, neighboring image IDs are searched to determine if they belong to a group of IDs with a certain label. If more than two neighboring IDs have been assigned the label "1", the unassigned ID is assigned to that label group and given the same label. If the labels of neighboring IDs are inconsistent, the two closest label IDs are selected, and their structural connection relationship is determined according to the ID number order. If a connection relationship exists, a label with the same trend is selected and inherited according to the connection path direction. The above neighborhood analysis and label inheritance operation is repeated for each unassigned ID until all unassigned image IDs have been assigned. During the assignment process, the label of each ID point comes from the majority vote or directional trend inheritance of its spatial neighborhood label set. Finally, all image IDs and their assigned labels are integrated to form a complete identification label table. This table records the image ID, whether it belongs to a connected path, the main peak direction matching status, the final label value, etc. Based on this table, a field ore quality identification scheme can be output.

[0096] Please see Figure 7A preliminary identification system for the quality of coal-bearing rare metal ore in the field, comprising:

[0097] The sampling and data initialization module is used to realize S1: collecting coal-series rare metal ore samples in the field, calling the spectrometer to obtain the main peak reflection signal, acquiring microscopic images to establish a particle structure layer, synchronously recording the changes in infrared absorption sections, uniformly numbering and generating a number list;

[0098] The anomaly identification and feature extraction module is used to implement S2: calling the number list, extracting the main peak layer of the image to compare the peak position direction, identifying the broken connected regions at the edge of particles in the microscopic image, analyzing the infrared spectral band shift trajectory, and generating abrupt image numbers;

[0099] The spatial structure reconstruction module is used to implement S3: based on the mutation image number, extract the sampling coordinates, analyze the distribution density within the map sheet, identify the area map patches with continuous and concentrated numbers, remove isolated markers, reconstruct the connection structure, and mark the turning position numbers;

[0100] The boundary drawing and position determination module is used to implement S4: draw closed boundaries according to the turning position number, mark the sampling point number blocks within the boundary, determine whether the coordinates of the ore to be measured are within the boundary, if not, trace the path of neighboring number changes and delineate the reference boundary number;

[0101] The label attribution and identification output module is used to implement S5: call the reference boundary number, determine whether the sample number and the boundary map group are connected and have the same trend. If they are consistent, mark it as a change label; otherwise, classify it as an adjacent area label and output the field ore quality identification scheme.

[0102] 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 preliminary identification of the quality of coal-bearing rare metal ore in the field, characterized in that, Includes the following steps: S1: Collect coal-bearing rare metal ore samples in the field, use a spectrometer to obtain the main peak reflection signal, collect microscopic images to establish a particle structure layer, record the changes in infrared absorption sections simultaneously, and generate a numbered list. The specific steps of S1 are as follows: S101: Based on coal-bearing rare metal ore samples obtained from field exploration, a spectrometer is used to scan the sample surface, extract the main peak position and corresponding reflectance of the band in the reflection spectrum, and generate a set of main peak reflection signal parameters. S102: Based on the main peak reflection signal parameter set, acquire a microscopic image of the corresponding region of the sample, extract the particle boundary, morphological features and texture direction in the image, and generate a particle structure layer data matrix. S103: Call the infrared acquisition device to obtain the infrared absorption section of the particle structure layer data matrix region, extract the absorption peak position and bind it with the main peak reflection signal parameter set and the particle structure layer data matrix according to the region number, and obtain the number list; S2: Call the number list, extract the image main peak layer to compare the peak position direction, identify the broken connected area at the edge of the microscopic image particles, analyze the infrared spectral band shift trajectory, and generate abrupt image number; The specific steps of S2 are as follows: S201: Call the number list, extract the image numbers in sequence, monitor the wavelength change of the main peak position in adjacent images for the main peak layer recorded in the image corresponding to the number, and calculate whether the offset direction is consistent, and obtain the main peak offset direction sequence; S202: Based on the main peak offset direction sequence, extract the particle structure layer data in the corresponding microscopic image, judge the contour curvature change of the edge region, identify the layer region with broken boundaries and connected structure, and generate a set of broken connected region numbers. S203: Call the image number in the set of broken connected regions, extract the spectral data in the corresponding infrared absorption section, compare the trend of the absorption peak position with the image sequence, determine whether there is a continuous shift in the same direction as the main peak, and filter and mark the image number that meets the requirements to obtain the abrupt image number. Read the infrared absorption segment data corresponding to the image number one by one. After obtaining the image band, extract the infrared absorption segment information of the region according to the location of the broken area in the number set. Use the absorption peak wave number in the infrared absorption segment as the comparison content. Record the absorption peak wave number one by one from the first image according to the image number order. Using the absorption peak wavenumber in the first image as a benchmark, the wavenumbers of absorption peaks at the same location in the image are compared numerically. If the wavenumber of the image is greater than the benchmark wavenumber, it is recorded as a positive offset. If the offset is less than the reference wavenumber, it is recorded as a negative offset; if it is equal to the reference wavenumber, it is recorded as a zero offset. All images are compared sequentially to form an absorption peak wavenumber offset direction sequence. The offset direction sequence is compared item by item with the main peak offset direction sequence. When the offset directions of the two sets of sequences are the same, the image is recorded as a directionally consistent image. In the directionally consistent images, three or more consecutive directionally consistent sequences are identified, and the changes in the offset direction are analyzed. If the offset direction is reversed in the sequence, the image number of the reversed image is recorded as a mutation image. All image numbers that show a direction reversal in the offset direction consistent sequence are selected and marked to form mutation image numbers. S3: Based on the mutation image number, extract the sampling coordinates, analyze the distribution density within the image area, identify regions with consecutive and concentrated numbers, remove isolated markers, reconstruct the connection structure, and mark the turning point numbers. The specific steps for S3 are as follows: S301: Based on the mutation image number, extract the field sampling coordinate point corresponding to each number, map all coordinate points to the map grid, and count the number density in each grid cell. Calculate the spacing between adjacent numbers on the map and arrange them in combination to obtain a set of spatially dense number blocks. S302: Based on the spatially dense numbered block set, remove numbered points that are not connected to any block on the map, and construct a connection structure for the remaining coordinate points according to their spatial distribution. Aggregate and divide all connection relationships to obtain a reconstructed continuous numbering structure map. Traverse all original mutation image numbers, and determine whether each number belongs to any identified spatial numbered block. If a number does not belong to any block, mark it as a non-connected number, remove it, and remove its coordinates from subsequent processing. Then, spatially sort the remaining numbered points according to their position on the map, connect each numbered point to its adjacent numbered points in order of minimum coordinate distance to form a line segment structure, and establish a connection relationship table between adjacent numbers. The connection relationship is constructed step by step with the numbered pair as the smallest unit. When there are multiple connectable paths, the path with the shortest connection distance is given priority. If multiple points form a closed structure around a common number, all connection relationships are retained. On this basis, the line segments in the connection diagram are merged and classified. Duplicate numbered points in the path structure are merged and the direction of the path segments is uniformly adjusted to maintain directional consistency. Then, all connection paths are aggregated and grouped according to the extensibility and connectivity of the connecting line segments. According to the continuity of the number and the stability of the connection structure, several short paths are merged into a complete path structure diagram. Finally, the reconstructed continuous numbering structure diagram composed of the starting point number, ending point number, intermediate numbered point sequence and their spatial geometric positions connected by each path is output. S303: Call the number sequence of the connected structural units in the reconstructed continuous numbering structure diagram, compare the directional change gradient of each number point in the two-dimensional coordinate system in sequence, identify the number position where the angle change is greater than the set angle change threshold, and record the number to obtain the turning position number. S4: Based on the turning point number, draw a closed boundary, mark the sampling point number block within the boundary, determine whether the coordinates of the ore to be tested are within the boundary, if not, trace the path of the neighboring number change and delineate the reference boundary number. S5: Call the reference boundary number to determine whether the sample number and the boundary map group are connected and have the same trend. If they are consistent, mark it as a change label; otherwise, classify it as an adjacent area label and output the field ore quality identification scheme.

2. The method for preliminary identification of the quality of coal-bearing rare metal ore in the field according to claim 1, characterized in that, The numbering list includes the main peak reflection number, particle layer number, and infrared spectral band number; the abrupt change image number includes peak position shift image, particle edge image, and infrared deformation image; the turning point number includes structural abrupt change point, number concentration point, and connection inflection point; the reference boundary number includes labeled trajectory line, closed boundary line, and numbered block distribution; and the field ore quality identification scheme includes change trend label, connectivity label, and region identification label.

3. The method for preliminary identification of the quality of coal-bearing rare metal ore in the field according to claim 1, characterized in that, The removal of isolated markers refers to deleting spatially scattered sampling markers that are not associated with the main connected region in the image numbering.

4. The method for preliminary identification of the quality of coal-bearing rare metal ore in the field according to claim 1, characterized in that, The change label refers to the characteristic identification mark assigned to a sample when the sample number and the boundary map group are consistent in terms of connectivity and trend.

5. The method for preliminary identification of the quality of coal-bearing rare metal ore in the field according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: According to the numbering of the turning point, connect adjacent coordinate points in the map in numerical order, monitor whether the geometric shape formed by the connecting path forms a closed loop, and if it forms a closed loop, mark the contour of the closed path formed by all numbered points to obtain the closed boundary graphic path set. S402: Call the closed boundary graphic path set, extract the numbers of all sampling coordinate points within the boundary range, cluster and integrate the number distribution area, define the number belonging range within the closed boundary according to the aggregation relationship, and establish a sampling number block table within the boundary; S403: Based on the sampling number block table within the boundary, compare whether the coordinates of the ore to be tested are located inside any closed path. If not, extract the numbered points adjacent to the coordinates, and retrieve the trend of structural direction change in the connected path. Select consecutive numbered points along the direction of change to construct the trajectory edge line and obtain the reference boundary number.

6. The method for preliminary identification of the quality of coal-bearing rare metal ore in the field according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Call the reference boundary number, extract the synchronous image number of all samples to be tested, arrange the synchronous image number in the order of acquisition, compare the position with the consecutive number order in the reference boundary, filter the image number that is sequentially connected, and obtain the connected matching image number set. S502: Based on the connected matching image number set, retrieve the corresponding main peak layer offset trend sequence, compare it point by point with the offset trend of the image number in the reference boundary number, determine whether the change direction is consistent, and assign a label value to the image number that simultaneously satisfies connectivity and direction consistency to obtain the change label number sequence. S503: Based on the changed label number sequence, number the unlabeled synchronous image, retrieve the neighboring number structure region in the number space, find the number belonging label of the region, perform neighborhood belonging judgment and label assignment processing on all numbers without connectivity, integrate all number and label content, and output the field ore quality identification scheme.

7. A preliminary identification system for the quality of coal-bearing rare metal ore in the field, characterized in that, The system is used to implement the preliminary identification method for the quality of coal-based rare metal minerals in the field as described in any one of claims 1-6. The system includes: The sampling and data initialization module is used to realize S1: collecting coal-series rare metal ore samples in the field, calling the spectrometer to obtain the main peak reflection signal, acquiring microscopic images to establish a particle structure layer, synchronously recording the changes in infrared absorption sections, uniformly numbering and generating a number list; The anomaly identification and feature extraction module is used to implement S2: calling the number list, extracting the image main peak layer to compare the peak position direction, identifying the broken connected regions at the edge of particles in the microscopic image, analyzing the infrared spectral band shift trajectory, and generating abrupt image numbers; The spatial structure reconstruction module is used to implement S3: based on the mutation image number, extract the sampling coordinates, analyze the distribution density within the map, identify the area map blocks with continuous and concentrated numbers, remove isolated markers, reconstruct the connection structure, and mark the turning position numbers; The boundary drawing and position determination module is used to implement S4: draw a closed boundary according to the turning position number, mark the sampling point number block within the boundary, determine whether the coordinates of the ore to be measured are within the boundary, if not, trace the path of the neighboring number change and delineate the reference boundary number; The label attribution and identification output module is used to implement S5: call the reference boundary number, determine whether the sample number and the boundary map group are connected and have the same trend. If they are consistent, mark it as a change label; otherwise, classify it as an adjacent area label and output the field ore quality identification scheme.

Images