Methods, apparatus and computer equipment for identifying target minerals

Abstract

This application relates to a method, apparatus, and computer equipment for identifying target minerals. It includes: firstly, performing preliminary identification of the concentrate in the main ore zone or main ore point, which effectively identifies concentrates containing the target mineral, reducing the burden of subsequent data processing and improving the efficiency and accuracy of preliminary detection; then, updating the pixels of the secondary ore zone to weaken false ore location information caused by ore edges and thickness, avoiding errors in subsequent identification; finally, accurately re-identifying the target ore point map based on the updated pixels of the main ore point, thus improving the overall performance of ore detection.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method, apparatus and computer device for identifying target minerals. Background Technology

[0002] Mineral identification is a crucial step in mineral resource exploration and development. With technological advancements, image recognition technology is increasingly being applied to mineral detection. Currently, images of the ore to be tested can be obtained through transmission electron microscopy (TEM), allowing for the separation of minerals such as copper and iron based on their mineralogical characteristics. However, the ore structure within the tested ore is often complex, and misidentification of the target mineral is common for different types of mineral zones and occurrences. Furthermore, the presence of other minerals with similar atomic numbers to the target mineral exacerbates these errors. Summary of the Invention

[0003] Based on this, the purpose of this application is to provide a method, apparatus, and computer device that can efficiently and accurately identify target minerals, so as to solve the technical problems mentioned in the background art.

[0004] Firstly, this application provides a method for identifying a target mineral. This includes:

[0005] Acquire a transmission image of the ore to be tested, and identify the ore enhancement features from the transmission image;

[0006] An initial mineral point map is determined to be associated with the ore enhancement features; the various mineral locations in the initial mineral point map include main mineral zones, main mineral points, or secondary mineral zones;

[0007] The main ore belt or the main ore point enables the initial identification of the concentrate of the ore to be tested;

[0008] The target mineral point map is obtained by updating the pixels of the secondary mineral zone, and minerals are re-identified in the target mineral point map through the main mineral point.

[0009] In one embodiment, before acquiring a transmission image of the ore to be tested, the method further includes: heating the ore to be tested using a thermal effect wave to determine the temperature change; and based on the temperature change and the dielectric properties of the target mineral, achieving initial identification of waste ore in the ore to be tested.

[0010] In one embodiment, identifying mineral enhancement features from the transmission image includes: binarizing the transmission image to obtain a target mask; obtaining the transmission image through a wave with ionizing radiation effect; determining the current pixel in the target mask and the set of neighboring pixels associated with the current pixel; updating the pixel value of the current pixel according to the set of neighboring pixels to obtain the mineral enhancement features.

[0011] In one embodiment, determining the initial mineral point map associated with the ore enhancement feature includes: processing the ore enhancement feature using multi-directional convolution kernels to obtain multiple texture response maps; performing gradient calculations on the multiple texture response maps to obtain their respective gradient maps; the number of pixels in different gradient maps is not the same; the number of pixels is associated with the main direction of the mineral location; and obtaining the initial mineral point map based on the multiple gradient maps and a preset pixel threshold.

[0012] In one embodiment, the theoretical label of the sub-mineral zone is 1; the method further includes: determining the target features and target labels corresponding to each mineral location in the sample ore; performing support vector machine classification using the target features and target labels to obtain a classification model; updating the pixels of the sub-mineral zone to obtain a target mineral point map, including: processing the actual features of the sub-mineral zone using the classification model to obtain actual labels; when the theoretical label is different from the actual label, updating the pixels at the sub-mineral zone to the target pixel value, until the target mineral point map is obtained.

[0013] In one embodiment, determining the target features and target labels corresponding to each ore location in the sample ore includes: determining sample response maps in different directions based on multi-directional convolution kernels and mask maps of the sample ore; determining the multi-scale energy associated with multiple sample response maps and obtaining initial features; weighting the initial features within the contour of each ore location to obtain target features; determining the weighting coefficients based on the number of pixels associated with multiple sample response maps; and determining the corresponding target label based on the main direction of each ore location.

[0014] In one embodiment, determining the corresponding target label based on the main direction of each ore location includes: when the ore location is a secondary ore zone, sampling a target quantity along the main direction of the secondary ore zone to obtain a first set; the target quantity is associated with the length of the secondary ore zone; performing a logical NOT process on the initial ore point map to obtain an auxiliary ore point map; sampling the auxiliary ore point map along the direction perpendicular to the main direction to obtain a second set; and determining the target label of the secondary ore zone based on the first set and the second set.

[0015] Secondly, this application also provides a device for identifying target minerals. It includes:

[0016] The concentrate initial identification module is used to acquire a transmission image of the ore to be tested and identify ore enhancement features from the transmission image; determine an initial mineral point map associated with the ore enhancement features; the initial mineral point map includes multiple mineral locations such as main mineral belts, main mineral points, or secondary mineral belts; and achieves the initial identification of the concentrate of the ore to be tested through the main mineral belts or the main mineral points.

[0017] The mineral re-identification module is used to update the pixels of the secondary mineral zone to obtain the target mineral point map, and to re-identify the minerals in the target mineral point map through the main mineral point.

[0018] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the above-described method for identifying the target mineral.

[0019] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-described method for identifying the target mineral.

[0020] The aforementioned method, apparatus, computer equipment, and readable storage medium for identifying target minerals achieve layered identification of target minerals by recognizing ore enhancement features from transmission images and combining this with the classification of main ore zones, main ore points, and secondary ore zones in the initial ore point map. First, initial identification of concentrates in the main ore zone or main ore point effectively identifies concentrates containing target minerals, reducing the burden of subsequent data processing and improving the efficiency and accuracy of initial detection. Next, the pixels of the secondary ore zone are updated to weaken false ore location information caused by ore edges and thickness, avoiding errors in subsequent identification. Finally, based on the main ore point, the updated target ore point map is accurately re-identified, improving the overall performance of ore detection and providing strong support for the efficient development of mineral resources. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a method for identifying target minerals in one embodiment;

[0022] Figure 2 This is a transmission image of the ore to be detected in one embodiment;

[0023] Figure 3 This is a comparison image of the initial mineral point map before and after generation in one embodiment;

[0024] Figure 4 This is an infrared temperature image of the ore to be tested in one embodiment;

[0025] Figure 5 This is an example data graph of multi-directional convolution kernels in one embodiment;

[0026] Figure 6 This is a comparison image before and after the initial mineral point map was generated in another embodiment;

[0027] Figure 7 An exemplary data diagram showing the initial characteristics of a ore site in one embodiment;

[0028] Figure 8This is an exemplary data diagram of candidate features of a ore site in one embodiment;

[0029] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] This application pertains to an environment where a terminal communicates with a computer device. The terminal is a different image acquisition device, and the computer device is used to acquire transmission images from the terminal and implement a method for identifying target minerals, including the following steps:

[0032] Step 102: Obtain a transmission image of the ore to be detected, and identify the ore enhancement features from the transmission image.

[0033] Transmission images are obtained through waves with ionizing radiation effects, including but not limited to X-rays. X-ray transmission detection is an important detection method for fusion mineral processing equipment, and transmission images can be obtained through X-ray emission sources and detectors.

[0034] Specifically, the ore to be tested enters an X-ray transmission imaging device, such as an X-ray transmission detector, via a conveyor belt, and the resulting transmission image is shown in Figure 2. Based on the mineralogical characteristics of the process, the minerals often exhibit distinct vein-like and fracture-like features during their formation. This results in the acquired XRT transmission image showing ore veins of varying lengths and thicknesses, representing ore locations. The computer equipment performs feature recognition on the transmission image to determine the ore enhancement characteristics of these ore locations, such as ore veins. Ore locations represent the areas in the ore to be tested where minerals are present; these areas can be represented by elongated ore veins or scattered point-like ore dots.

[0035] Step 104: Determine the initial mineral point map associated with the ore enhancement features.

[0036] The initial mineral location map includes various mineralization zones, primary mineralization points, or secondary mineralization zones. Ore enhancement features characterize pixel features that enhance the texture or structure of the ore.

[0037] Specifically, the computer device processes the ore enhancement features using convolution kernels in different directions to obtain multiple texture response maps, and then performs pixel processing and fusion on these multiple texture response maps to obtain the initial ore point image (ore_line_image). For example... Figure 3 The image shown is a comparison of the initial mineral deposit map before and after its generation. Figure 3a is a transmission image that does not include mineral enhancement features. Figure 3 b is a transmission image that includes the ore enhancement features. Figure 3 c represents the initial mineral point image after pixel processing and fusion. It can be seen that... Figure 3 The texture features of various ore locations in b are compared to Figure 3 a has been enhanced. Figure 3 In the image c, white pixels theoretically represent real ore sites, while black pixels represent non-ore sites. The main ore zone is the most obvious and wider ore zone within the white area. Scattered ore points and narrower secondary ore zones circled in yellow within the white area may be ore sites, or they may be false ore points or false ore zones caused by the edges or thickness of non-ore sites.

[0038] Step 106: Initial identification of concentrate of the ore to be tested is achieved through the main ore belt or main ore point.

[0039] Specifically, the computer equipment extracts the main ore zones with prominent features from the initial ore point map and presets a width threshold (ore_line_width) and a ratio threshold (ore_area_ratio). When the width of the main ore zone exceeds the width threshold, it indicates that a relatively coarse ore zone has been extracted, and the ore to be tested is a high-grade ore with a high content of the target mineral. Main ore zones whose width does not exceed the width threshold are considered secondary ore zones. Similarly, the total number of main ore points is calculated as a percentage of the ore area. If it exceeds the ratio threshold, it indicates that the ore to be tested is a high-grade ore with a high content of the target mineral. This achieves the identification that the ore to be tested is a pure concentrate containing the target mineral; otherwise, it determines that only secondary ore zones or secondary ore points have been extracted, and further re-identification of the ore to be tested is required.

[0040] Step 108: Update the pixels of the secondary mineral zone to obtain the target mineral point map, and re-identify the minerals in the target mineral point map through the main mineral point.

[0041] In one embodiment, updating the pixels of the secondary ore zone to obtain the target ore point map includes: processing the actual features of the secondary ore zone through a classification model to obtain actual labels; when the theoretical labels are different from the actual labels, updating the pixels at the secondary ore zone to the target pixel values, until the target ore point map is obtained.

[0042] Specifically, while a portion of pure concentrate has been identified through the well-defined main ore zone, situations arise where the ore lacks a coarse ore zone, only having indistinct, elongated secondary ore zones, or where the width of the main ore zone does not exceed a certain threshold. Directly extracting these secondary ore zones through initial concentrate identification can result in significant deficiencies, potentially leading to some low- and medium-grade ores being misclassified as waste rock, thus resulting in excessively high tailings grades. Furthermore, there are instances where secondary ore zones extracted from ore edges and thicker sections are identified as false ore zones, severely interfering with ore sorting and potentially misclassifying waste rock as medium-grade ore, thereby lowering the grade of ore containing the target mineral components.

[0043] Therefore, the computer equipment performs mapping and classification calculations on a large number of sample ores, training a classification model and adjusting thresholds for the target features and labels corresponding to each ore location to obtain a classification model. Then, the main ore points and secondary ore zones with actual features from the initial ore point map are input into the classification model to obtain corresponding classification results, for example: 0 for false ore points / zones, and 1 for true ore points / zones. Although the theoretical label of a secondary ore zone pixel before the update is 1, when it is identified as a false ore zone (actual label is 0), its pixel is updated to the target pixel value of 0, which is equivalent to filtering out interference caused by ore thickness fluctuations or edge transformations, ultimately obtaining the target ore point map. Finally, the actual number of main ore points in the target ore point map is traversed. When the ratio of the actual number of main ore points to the ore area exceeds a certain threshold, it indicates that the ore to be detected contains the target mineral; otherwise, it is waste rock.

[0044] In the aforementioned method for identifying target minerals, layered identification of target minerals is achieved by identifying ore enhancement features from transmission images and combining this with the classification of main ore zones, main ore points, and secondary ore zones in the initial ore point map. First, preliminary identification of concentrates in the main ore zones or main ore points effectively identifies concentrates containing target minerals, reducing the burden on subsequent data processing and improving the efficiency and accuracy of initial detection. Next, the pixels of the secondary ore zones are updated to weaken false ore location information caused by ore edges and thickness, avoiding errors in subsequent identification. Finally, based on the main ore points, the updated target ore point map is accurately re-identified, improving the overall performance of ore detection and providing strong support for the efficient development of mineral resources.

[0045] In one embodiment, before acquiring a transmission image of the ore to be tested, the method further includes: heating the ore to be tested using a thermal effect wave to determine the temperature change; and based on the temperature change and the dielectric properties of the target mineral, achieving initial identification of waste ore in the ore to be tested.

[0046] Among them, thermal effect waves, such as the 915MHz microwave frequency commonly used in industry, have low energy and selective heating characteristics. Different metals have different dielectric properties such as dielectric constant and loss factor, resulting in different heating rates and different microwave absorption capabilities. Therefore, thermal effect waves can achieve differentiated heating of different mineral components in ores.

[0047] Specifically, using an infrared camera with temperature monitoring, the average temperature of the ore to be tested before and after heating can be accurately captured. Based on the characteristic that pure waste ore has small temperature fluctuations before and after heating, initial identification of waste ore can be performed. For example, when the target mineral to be recovered is copper ore, the temperature difference between metals with atomic numbers far from copper and copper before and after heating is large. Therefore, this characteristic can be used to exclude a small portion of other waste ores that do not contain copper. First, the ore to be tested is crushed to, for example, less than 50mm, because the penetration ability of thermal waves is limited; larger stones cannot penetrate and will interfere with the results. Next, the ore to be tested is conveyed through multiple channels. Temperature monitoring devices at the channel inlets automatically record the corresponding feed temperature. As the ore to be tested passes through the heating device at a uniform speed, an infrared camera with temperature monitoring at the channel outlet records the discharge temperature and acquires infrared images simultaneously. The results are as follows: Figure 4 As shown. For example, the ore to be tested in column 3 has a very low temperature, almost the same as the temperature of the special heat-resistant rubber belt, which is pure waste ore that does not contain the target mineral at all.

[0048] In this embodiment, the ore to be tested is heated by thermal wave and the temperature change is monitored. Combined with the dielectric properties of the target mineral, waste ore is initially identified. This allows for the early screening of waste ore that does not contain the target mineral, reducing the pressure on subsequent data processing and improving the efficiency of subsequent detection algorithms. Furthermore, the initial identification of waste ore is integrated with subsequent initial identification of concentrate and mineral re-identification, ensuring the accuracy and comprehensiveness of target mineral detection.

[0049] In one embodiment, identifying mineral enhancement features from a transmission image includes: binarizing the transmission image to obtain a target mask; determining the current pixel in the target mask and the set of neighboring pixels associated with the current pixel; and updating the pixel value of the current pixel based on the set of neighboring pixels to obtain the mineral enhancement features.

[0050] In this process, the transmitted image is obtained through waves that exhibit ionizing radiation effects.

[0051] Specifically, the computer device first binarizes the transmission image image_XRT to remove background interference and obtains a mask image mask1 for all targets. A threshold thresh1 is set, and each pixel in the transmission image image_XRT corresponding to mask1 is traversed. If the pixel value is greater than thresh1, it is set to 0, indicating that it may be a pixel formed by a mineral region. This process yields the binarized target mask image mask2.

[0052] Since the current pixel pixel2(x,y) corresponding to the target mask image mask2 could be a mineral deposit, mineral belt, or other mineral location, or it could be an edge or a thicker area outside the mineral location, the computer device obtains the current pixel pixel2(x,y) and the pixel values ​​of its associated 8-neighbor pixel set neighbor{9}. The minimum value min_local in the neighbor pixel set neighbor{9} is calculated and updated to the current pixel value pixel2(x,y) = min_local. This process of highlighting the visual features of mineral belts or deposits is achieved, resulting in an enhanced image distinct_ore containing mineral enhancement features. (See reference...) Figure 3 a and Figure 3 As shown in b.

[0053] In this embodiment, by binarizing the transmission image and updating the neighboring pixels, the enhanced features of the ore can be effectively extracted, highlighting the difference between the mineral and the background, improving the contrast and clarity of the image, and thus providing a more accurate image basis for subsequent ore identification and analysis.

[0054] In one embodiment, determining the initial mineral point map associated with the ore enhancement features includes: processing the ore enhancement features through multi-directional convolution kernels to obtain multiple texture response maps; performing gradient calculations on the multiple texture response maps to obtain their respective gradient maps; and obtaining the initial mineral point map based on the multiple gradient maps and a preset pixel threshold.

[0055] The multiple directions include 0°, 45°, 90°, and 135° directions; the number of pixels in different gradient maps is not the same; the number of pixels is related to the main direction of the ore location. For example, when the ore location is a mineral belt type, the more pixels there are in a certain direction, the greater the probability that the mineral belt will extend in that direction, and that direction is the main direction.

[0056] Specifically, computer equipment via, for example Figure 5The four convolutional kernels shown are used to calculate texture response maps. Specifically, a 3×3 horizontal convolutional kernel is used to obtain the horizontal texture response map image_laws0; a 4×4 convolutional kernel is used to obtain the vertical texture response map image_laws90; a 5×5 rotational convolutional kernel is used to obtain the 45° texture response map image_laws45; and a 5×5 rotational convolutional kernel is used to obtain the 135° texture response map image_laws135. The computer then uses the Laplacian gradient to calculate the corresponding gradient maps gradient0, gradient45, gradient90, and gradient135 for each texture response map. Since a preset pixel threshold gradient_th is used, pixels in each gradient map greater than this threshold are considered to represent different types of ore deposits, thus obtaining the four gradient maps corresponding to ore point images ore_line_image0, ore_line_image90, ore_line_image45, and ore_line_image135. Finally, the computer equipment performs comprehensive processing on the multi-directional mineral point maps to obtain the initial mineral point map.

[0057] In this embodiment, ore enhancement features are processed using multi-directional convolution kernels to generate multiple texture response maps, and gradient maps are further calculated. Combined with pixel threshold filtering, the texture and structural information of the ore is effectively extracted, and the mineral points and mineral zones are accurately located, providing a more reliable basis for mineral identification.

[0058] In one embodiment, the method further includes: determining the target features and target labels corresponding to each ore location in the sample ore; and performing support vector machine classification using the target features and target labels to obtain a classification model.

[0059] Specifically, computer equipment uses multi-directional convolution kernels to perform multi-scale energy calculations and weighting to determine the target features corresponding to different ore locations. For example... Figure 6 The image shown is a comparison of the initial mineral occurrence map of the sample ore before and after generation. Figure 6 'a' represents the initial mineral deposit map before its generation. Figure 6b has clearly extracted the secondary mineral zone within the red circle, with a theoretical label of 1; however, it has also extracted the false mineral zones and mineral points within the pink circle, caused by the edges and thickness of the rocks, with a theoretical label of 0. Since the initial mineral point map contains a large number of mineral points and zones, manual label identification is not an ideal solution. Based on the formation of mineral points and zones, it is easy to see that a true mineral zone has the following characteristic: a large difference in pixel values ​​between its two neighboring pixels and those perpendicular to the direction of its extension. A false mineral zone, on the other hand, has the following characteristic: at least one pixel value is relatively small compared to those perpendicular to the direction of its extension. Therefore, based on the characteristics of true and false mineral zones, pixel points can be collected separately along the main direction and the vertical direction of the ore location, and the target label for the secondary mineral zone can be determined based on the collection results.

[0060] Finally, after mapping and classifying a large number of sample ores, the computer equipment can obtain the target features and labels corresponding to different ore locations, and then perform support vector machine classification (SVM classification) to obtain the final classification model. Therefore, by determining the features and labels of sample ores and using support vector machine classification to build a model, accurate ore classification can be achieved, improving the efficiency and accuracy of subsequent ore identification.

[0061] In one embodiment, determining the target features and target labels corresponding to each ore location in the sample ore includes: determining sample response maps in different directions based on multi-directional convolution kernels and mask maps of the sample ore; determining the multi-scale energy associated with multiple sample response maps and obtaining initial features; weighting the initial features within the contour of each ore location to obtain target features; and determining the corresponding target label based on the main direction of each ore location.

[0062] The weighting coefficients are determined based on the number of pixels associated with multiple sample response maps.

[0063] Specifically, the computer equipment treats all the minerals to be detected as sample minerals for model training. Therefore, the target mask image corresponding to the mineral to be detected is regarded as the mask image corresponding to the sample minerals during model training; the texture response image obtained by passing the mineral to be detected through multi-directional convolution kernels is regarded as the sample response images image_laws0, image_laws45, and image_laws90 corresponding to the sample minerals during model training. The computer equipment calculates multi-scale energy through multiple sample response images, that is, it traverses each image and calculates the mean absolute deviation (MAD) of 3×3, 5×5, 7×7, 9×9, 11×11, and 15×15 windows for each pixel. For example, MAD0_ represents the local window energy feature corresponding to the pixel in image_laws0, and MAD45_ represents the local window energy feature corresponding to the pixel in image_laws45.

[0064] Next, the computer device concatenates the multi-directional feature vectors of each pixel to obtain the initial feature feature(x,y), and then performs weighted processing on the initial feature feature(x,y) to obtain the candidate feature feature_ratio(x,y). Example data of the initial and candidate features are shown below. Figure 7 , Figure 8 As shown, after performing gradient calculations on multiple sample response maps, a gradient map reflecting pixel features is obtained. Therefore, the weighting coefficient is determined by the ratio between the number of pixels in the gradient map of each direction. That is, the more pixels in a certain direction, the greater the probability that the mineral belt extends in that direction. That direction is the main direction, and the weighting coefficient is the highest. Furthermore, the corresponding target label can be determined based on the main direction of each mineral location.

[0065] Finally, the computer equipment uses the initial mineral point map as a mask to determine the contour of the mineral location. For each contour, it calculates the corresponding candidate features for all pixel sets in the region represented by each contour. Then, it calculates the weighted and averaged standardized result of the contour and uses it as the target feature of the mineral location.

[0066] In this embodiment, sample response maps in different directions are generated using multi-directional convolution kernels and mask maps, and initial features are extracted using multi-scale energy. The initial features within different ore contours are weighted to obtain more representative target features. Simultaneously, target labels are determined based on the main direction of the ore location, enabling more accurate identification and differentiation of ore types and locations, thus improving the accuracy of mineral identification and classification.

[0067] In one embodiment, determining the corresponding target label based on the main direction of each ore location includes: when the ore location is a secondary ore zone, sampling the target number along the main direction of the secondary ore zone to obtain a first set; performing logical NOT processing on the initial ore point map to obtain an auxiliary ore point map; sampling the auxiliary ore point map along the direction perpendicular to the main direction to obtain a second set; and determining the target label of the secondary ore zone based on the first set and the second set.

[0068] The target quantity is related to the length of the secondary mineral belt; the vertical direction of the main direction is calculated using the slope relationship formula.

[0069] Specifically, the computer equipment determines the length of each sub-mineral belt, classifying it into three levels: less than length1, greater than length2, or between length1 and length2. Then, it uniformly selects a target number of sampling points (count) from each sub-mineral belt along its main direction, forming the first set. The target number of sampling points for sub-mineral belts less than length1 is count=count1, for those greater than length2 it is count=count2, and for those between length1 and length2 it is count=count1_2. The computer equipment performs a logical NOT operation on the initial mineral point map to obtain an auxiliary mineral point map (sample_mask). Using this auxiliary map as a mask, and taking the current sampling point (sample_point) in the first set as an example, it uniformly selects a target number of sampling points (count) on both sides of the current sampling point (sample_point) in the main direction (vertical direction), forming the second set.

[0070] Finally, the computer device determines the actual pixel value of the current sampling point in the first set and the comparison pixel values ​​of multiple auxiliary sampling points in the corresponding second set. If the comparison pixel value of any auxiliary sampling point is greater than the actual pixel value, and the difference between the two pixel values ​​is greater than a first preset value, then this auxiliary sampling point is determined as a target point. When the proportion of the first number of target points in the second set is greater than a second preset value, then the current sampling point in the first set is determined to be a positive sampling point. The above process is then repeated for each sampling point in the first set until the second number of positive sampling points is determined. When the proportion of the second number of positive sampling points in the first set is greater than a second preset value, then the target label for this sub-mineral belt is determined to be 1; otherwise, it is 0.

[0071] In one embodiment, the final target label is obtained in response to the user's identification, cleaning, and correction operations on the label.

[0072] In this embodiment, by sampling along the main direction and vertical direction of the secondary ore zone respectively, and combining this with the logical NOT processing of the initial ore point map, the target labels of the secondary ore zone are accurately determined. The number of samples is correlated with the length of the ore zone, which improves sampling efficiency and accuracy, and enhances the reliability of ore point identification.

[0073] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0074] Based on the same inventive concept, this application also provides a target mineral identification device for implementing the target mineral identification method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more target mineral identification device embodiments provided below can be found in the limitations of the target mineral identification method described above, and will not be repeated here.

[0075] In one embodiment, a target mineral identification device is provided, comprising: a concentrate initial identification module and a mineral re-identification module, wherein:

[0076] The concentrate initial identification module is used to acquire the transmission image of the ore to be tested and identify the ore enhancement features from the transmission image; determine the initial mineral point map associated with the ore enhancement features; the initial mineral point map includes various mineral locations such as main mineral belt, main mineral point or secondary mineral belt; and realize the initial identification of the concentrate of the ore to be tested through the main mineral belt or main mineral point.

[0077] The mineral re-identification module is used to update the pixels of the secondary mineral zone to obtain the target mineral point map, and to re-identify the minerals in the target mineral point map through the main mineral point.

[0078] In one embodiment, the target mineral identification device further includes a waste ore preliminary identification module, which is used to heat the ore to be tested by thermal effect wave to determine the temperature change; and to realize the waste ore preliminary identification of the ore to be tested based on the temperature change and the dielectric properties of the target mineral.

[0079] The various modules involved in the identification of the aforementioned target minerals can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can invoke and execute the corresponding operations of each module.

[0080] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores transmission images and mineral point maps. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for identifying target minerals.

[0081] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0082] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0083] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0084] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0085] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method of identifying a target mineral, characterized by, The method comprises: acquiring a transmission image of the ore to be detected, and identifying ore enhancement features from the transmission image; including: performing binaryzation processing on the transmission image to obtain a target mask image; the transmission image is obtained by using waves with ionizing radiation effect; determining a current pixel in the target mask image and a neighborhood pixel set associated with the current pixel; updating the pixel value of the current pixel according to the neighborhood pixel set to obtain ore enhancement features; determining an initial ore point image associated with the ore enhancement features; a plurality of ore positions in the initial ore point image include a main ore belt, a main ore point or a secondary ore belt; realizing preliminary identification of a concentrate of the ore to be detected through the main ore belt or the main ore point; performing pixel updating on the secondary ore belt to obtain a target ore point image, including: processing actual features of the secondary ore belt through a classification model to obtain actual labels; when the theoretical label is different from the actual label, updating the pixel at the secondary ore belt to a target pixel value until the target ore point image is obtained; and performing mineral re-identification on the target ore point image through the main ore point.

2. The method of claim 1, wherein, Before acquiring the transmission image of the ore to be detected, the method further comprises: performing heating treatment on the ore to be detected through thermal effect waves to determine temperature change conditions; based on the temperature change conditions and dielectric properties of the target mineral, realizing preliminary identification of waste ore of the ore to be detected.

3. The method of claim 1, wherein, The method of determining the initial ore point image associated with the ore enhancement features comprises: processing the ore enhancement features through multi-directional convolution kernels to obtain a plurality of texture response maps; performing gradient calculation on the plurality of texture response maps respectively to obtain respective corresponding gradient maps; the number of pixel points in different gradient maps is not the same; the number of pixel points is associated with the main direction of the ore position; obtaining an initial ore point image according to a plurality of gradient maps and a preset pixel threshold.

4. The method of claim 1, wherein, The theoretical label of the secondary ore belt is 1; the method further comprises: determining target features and target labels corresponding to each ore position in the sample ore; performing support vector machine classification through the target features and target labels to obtain a classification model; performing pixel updating on the secondary ore belt to obtain a target ore point image, including: processing actual features of the secondary ore belt through the classification model to obtain actual labels; when the theoretical label is different from the actual label, updating the pixel at the secondary ore belt to a target pixel value until the target ore point image is obtained.

5. The method of claim 4, wherein, The method of determining target features and target labels corresponding to each ore position in the sample ore comprises: determining sample response maps in different directions according to multi-directional convolution kernels and a mask image of the sample ore; determining multi-scale energy associated with a plurality of sample response maps to obtain initial features; performing weighted processing on the initial features in the contour of each ore position to obtain target features; a weighting coefficient is determined according to the number of pixel points associated with a plurality of sample response maps; determining a corresponding target label according to the main direction of each ore position.

6. The method of claim 5, wherein, The method of determining a corresponding target label according to the main direction of each ore position comprises: when the ore position is a secondary ore belt, sampling a target number along the main direction of the secondary ore belt to obtain a first set; the target number is associated with the length of the secondary ore belt; performing logical NOT processing on the initial ore point graph to obtain an auxiliary ore point graph; sampling the auxiliary ore point graph in a direction perpendicular to the main direction by the target number to obtain a second set; determining a target label of the secondary ore belt according to the first set and the second set.

7. An apparatus for identifying a target mineral, characterized by The device for implementing the method in any one of claims 1 to 6 comprises: a concentrate primary identification module, configured to acquire a transmission image of the ore to be detected, identify ore enhanced features from the transmission image, determine an initial ore point graph associated with the ore enhanced features, and perform concentrate primary identification of the ore to be detected through the main ore belt or the main ore point, wherein the multiple ore positions in the initial ore point graph comprise a main ore belt, a main ore point, or a secondary ore belt; a mineral re-identification module, configured to perform pixel updating on the secondary ore belt to obtain a target ore point graph, and perform mineral re-identification on the target ore point graph through the main ore point. 8.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-7. The processor executes the computer program to implement the steps of the method in any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 6.

Images