Method for automatically searching and positioning target mineral by using microscope

Abstract

The invention belongs to the technical field of mineral resource exploration, and particularly relates to a method for automatically searching and positioning a target mineral by using a microscope. A set of target mineral automatic searching system is established based on a Python language, before mineral searching work is carried out, a target mineral is preliminarily and artificially searched, the range of the target mineral is accurately limited, and the target mineral automatic searching system is used for carrying out machine learning and limiting on transmission light, orthogonal light and reflected light lithofacies image features of the target mineral in a targeted manner; then the target mineral in the whole slice is automatically identified, searched and positioned, and if the observation multiple of the microscope is increased, the finer mineral can be rapidly and systematically identified, so that the time consumption of manual identification is saved, and the error probability and the neglect probability in the identification process can also be reduced. Meanwhile, coordinate information of the target mineral can be systematically recorded and then translated into positioning coordinates meeting the actual situation of experimental software through an algorithm, and the experimental time is greatly shortened.

Description

Technical Field

[0001] The present invention belongs to the technical field of mineral resource exploration, and particularly relates to a method for automatically searching and locating target minerals by using a microscope. Background Art

[0002] With the development of micro-area technology, in the current process of ore deposit research, it is often necessary to carry out some geochemical analysis and testing work on certain types of target minerals. For example, laser in-situ isotope dating work is carried out on minerals such as zircon and apatite; laser in-situ S isotope tracing work is carried out on minerals such as pyrite and chalcopyrite; in-situ electron probe major element analysis work is carried out on minerals such as pyrochlore, etc., or some Raman analysis work is carried out on inclusions. However, these minerals have a common characteristic in the vast majority of ore deposits: they are fine-grained, there are many interfering minerals, and it is difficult to find them.

[0003] The optical microscope is an essential instrument for carrying out preliminary ore deposit geological research work and is used to observe the morphology and position of minerals. At present, the performance of the optical microscope has been greatly improved, and it can observe the transmitted light, cross-polarized light, and reflected light of minerals with a size of 1 micron in a thin section under an objective lens of 50 times or even 100 times.

[0004] In recent years, the concept and method of using a computer language model to control the microscope to automatically identify minerals have been proposed in this field. However, due to certain differences in the occurrence of minerals in geological bodies in different regions, even for the same mineral, in different geological bodies, different mineral thin sections, or even under different light intensities, there will be differences in its reflected light, transmitted light, and cross-polarized light. As a result, the method of using a microscope to automatically identify minerals has a low accuracy rate in practical applications and lacks a mature software, so the promotion of this method is not high.

[0005] Although the manual microscope prospecting method can rely on petrographic observation experience to identify target minerals, it is very easy to overlook some minerals or need to repeatedly correct similar minerals. Especially for fine-grained minerals, the observation time required is exponentially increased, which will greatly increase the probability of overlooking, the probability of making mistakes, and the number of corrections.

[0006] At the same time, even after finding the target mineral under an ordinary optical microscope, during the subsequent laser, probe, or Raman experiment, due to the limitation of the experimental hardware facilities, it is only possible to search for the same position again at a magnification of 25 times or even greater. However, due to the excessive magnification, the field of view is too small, and it is only possible to rely on rough human judgment for positioning. Before the analysis, a large amount of time is often wasted for repositioning, resulting in a large loss of precious experimental time.

[0007] Therefore, a microscope automatic search system that can accurately, quickly, and comprehensively identify target minerals from thin-section samples is needed in this field. Summary of the Invention

[0008] Based on the above problems, the present application provides a method for automatically searching and locating target minerals using a microscope. This method can eliminate the time-consuming nature of manual identification, reduce the error probability and omission probability during the identification process; when the found target minerals are used in geochemical experiments, the coordinate information of the target mining area can also be translated into positioning coordinates that conform to the actual situation of the experimental software through an algorithm, greatly shortening the experimental time.

[0009] To this end, the technical solution of the present application is: a method for automatically searching and locating target minerals using a microscope, including the following steps:

[0010] S1. Sample preparation: Prepare the sample to be analyzed into a standard rock mineral thin section;

[0011] S2. Target mineral identification and adjustment of optical information parameters: Manually search for 1-2 groups of target minerals and interfering minerals on the rock thin section, adjust the optical information parameters of transmitted light, reflected light, and orthogonal light, confirm the image differences between the two, record the optical information parameters at this time, and obtain the overall scanned image of the mineral thin section and the transmitted light, reflected light, and orthogonal light mineral images of the target minerals and interfering minerals under this optical information parameter through the microscope.

[0012] S3. Limiting the image range of the target mineral and feature extraction: Use computer language to process the transmitted light, reflected light, and orthogonal light images of the target minerals and interfering minerals, confirm the image differences between the two, then accurately limit the image range of the target mineral using the image information threshold, and extract and learn the reflected light, transmitted light, and orthogonal light image features of the target mineral within the limited range based on computer language;

[0013] S4. Extracting multi-dimensional information from the scanned image and matching the target mineral: Extract multi-dimensional image information from the overall scanned image of the mineral thin section and the target mineral image, and after image information thresholding processing and algorithm comparison, match and confirm the target mineral;

[0014] S5. Circling the coordinates of the target mineral: According to the particle size of the target mineral required, divide the analyzable beam spot area based on whether there is a single clean area within the target mineral, screen, label, and circle the mineral coordinates of the target mineral;

[0015] S6. Locating the target mineral: Transform the coordinates of the target mineral in the overall scanned image into labelable coordinate information through an image affine transformation matrix to complete the positioning of the target mineral.

[0016] Further, for the 1 - 2 groups of target minerals and interfering minerals found on the mineral thin section in S2, adjust the optical information parameters of transmitted light, reflected light, and cross - polarized light, and confirm the image differences between the two. Specifically: Select transmitted light that can distinguish target minerals and interfering minerals using different objective lens magnifications, light sources, and observation methods of an optical microscope; control the cross - polarized angle of the microscope, select cross - polarized light that can distinguish target minerals and interfering minerals, and determine the interference color; by adjusting the reflected light parameter settings, select reflected light that can distinguish target minerals and interfering minerals.

[0017] Further, the computer language in S3 is Python language.

[0018] Further, the multi - dimensional image information extraction in S4 includes extracting picture information in HSV, YCrCb, HLS, XYZ, LAB, and YUV dimensions.

[0019] Further, for the specific matching and confirmation of target minerals after image information thresholding processing and algorithm comparison in S4: According to the results of multi - dimensional image information extraction, perform pixel traversal on the overall scanned image of the mineral thin section. Through the thresholding processing and algorithm comparison of the image information of the thin section scanned image, compare and verify according to the processing results of the matching minerals in one or more targeted scanned images, exclude interfering minerals, and then perform matching, calibration, and confirmation on the determined target minerals.

[0020] Further, when the circled target minerals are used for geochemical experiments, the coordinate information that can be marked in S6 is the coordinate information corresponding to the experimental conditions of the coordinate system of the experimental software that can be imported.

[0021] Compared with the prior art, the present invention has the following beneficial effects: This method has strong learning ability and high pertinence, and is more generally applicable to the work of finding target minerals in the research of various geological bodies. Moreover, the test is convenient. Only need to manually find 1 - 2 target minerals, then it can automatically identify, find, and locate the matching minerals in the thin section, and can automatically screen its subsequent work to avoid too much final data volume of identification. It can screen out samples that can be used for the next test work from a large number of samples for researchers in the early stage of geological research work, mark the positions in the images of the samples, and convert them into coordinate information in actual experiments, greatly improving the work efficiency of researchers and saving valuable experimental time. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a result diagram of different calculation methods for converting a color image into a grayscale image;

[0023] Figure 2 It is a result diagram of different methods for enhancing grayscale image information comparison;

[0024] Figure 3 For the range limitation and feature extraction of the reflected light, transmitted light, and polarized light images of the target mineral;

[0025] Figure 4 For the enhancement and extraction of the image information of the scanned image;

[0026] Figure 5 For the identification of the target mineral and the acquisition of coordinate information;

[0027] Figure 6 For the positioning of the target mineral in the experiment;

[0028] Figure 7 This is the flowchart of the method for quickly searching and positioning the target mineral of the present invention. Specific Embodiments

[0029] The present invention will be specifically described below in conjunction with embodiments for the understanding of those skilled in the art. It is necessary to particularly point out here that the embodiments are only used to further illustrate the present invention and should not be construed as limiting the protection scope of the present invention. Those skilled in the art, based on the above invention content, make non-essential improvements and adjustments to the present invention, which should still fall within the protection scope of the present invention. At the same time, for the raw materials not described in detail, they are all commercially available products; for the process steps or preparation methods not mentioned in detail, they are all process steps or preparation methods known to those skilled in the art.

[0030] Example 1 In the study of the Bayan Obo deposit, zircon, a target mineral, was searched for geochemical analysis, and its method flowchart is as Figure 7 shown.

[0031] S1. Sample preparation: Prepare the sample to be analyzed into a standard rock mineral thin section.

[0032] Prepare a standard rock mineral thin section, polish the thin section. The smaller the target mineral, the higher the magnification of the microscope used to search for the mineral, and the higher the polishing requirement. The polishing can be gradually carried out using 3-micron and 1-micron diamond polishing paste. For fine mineral thin sections, fine polishing can be carried out using 0.25-micron diamond polishing paste, and ultrasonic cleaning can be used to remove the residues between the minerals in the thin section. For searching for small target minerals, the microscope can be placed on a shock-absorbing table to minimize the fine fluctuations and horizontal stability of the sample on the stage caused by placing the sample or adjusting the objective lens.

[0033] S2. Target Mineral Identification and Optical Information Parameter Adjustment: Manually search for 1-2 groups of target minerals and interfering minerals on the rock thin section. Under transmitted light, reflected light, and cross-polarized light, expand the optical information difference between the two by adjusting the light intensity and cross-polarized angle, record the optical information parameters at this time, and obtain the overall scanned image of the mineral thin section and the transmitted light, reflected light, and cross-polarized light mineral images of the target minerals and interfering minerals under these optical information parameters through a microscope.

[0034] Before starting the first work, it is necessary to manually search for a small amount of target minerals and interfering minerals with an optical microscope. According to the differences between the target minerals and interfering minerals found, use different objective lens magnifications, light sources, and observation methods. The microscope control software can control the light intensity device and adjust different rock thin section samples and different objective lens observation magnifications specifically, and select the appropriate light intensity that can distinguish the target minerals and interfering minerals. At the same time, control the cross-polarized angle of the microscope and select the appropriate interference color that can distinguish the target minerals and interfering minerals.

[0035] In the standard rock mineral thin section prepared from the sample to be analyzed obtained from the Bayan Obo deposit in this embodiment, by adjusting the transmitted light, reflected light, and cross-polarized light, it is found that zircon is the target mineral and apatite is the interfering mineral. Both are transparent-light yellow under transmitted light, but there are differences under cross-polarized light and reflected light. Under 90° cross-polarized light, zircon shows vivid interference colors of 3-4 grades, and apatite shows grayish-brown interference colors; under reflected light, zircon mainly shows bright gray reflected light at 70% light intensity, while apatite shows dark white under the same reflected light intensity. Therefore, this target mineral is mainly identified by reflected light, so this embodiment focuses on setting the reflected light parameters.

[0036] At the same time, use the microscope to automatically take pictures and piece together the entire thin section under the same reflected light and transmitted light condition parameters to obtain the overall scanned image of the thin section. Even due to the anisotropy of some minerals, under reflected light, different rotation angles will show different colors. The scanned images at different rotation angles can be compared with the original scanned image, and the coordinate calibration of the thin section scanned image can be performed through the image affine transformation matrix, and then the positions of each mineral in the scanned image can be aligned.

[0037] S3. Target Mineral Image Range Limitation and Feature Extraction: Use computer language to process the transmitted light, reflected light, and cross-polarized light images of the target minerals and interfering minerals, confirm the image differences between the two, and then accurately limit the target mineral image range using the image information threshold, and extract and learn the reflected light, transmitted light, and cross-polarized light image features of the target minerals within the limited range based on computer language.

[0038] Among them, taking the reflected light as an example, since the reflected light is mainly a gray image, the image can be grayscaled based on OpenCV in the Python language and pixel processing languages (maximum value method, average value method, weighted average method, component method), and finally a new grayscale image is created.

[0039] Among them, the maximum value method for grayscale processing means taking the maximum value among the R, G, and B channel values of each pixel as the grayscale value of that pixel, which is suitable for highlighting the brighter areas in the scanned image; the average value method for grayscale processing of the image means that the grayscale value is the average value of the RGB three components, and this method is suitable for processing images with faster requirements; the weighted average method takes into account the differences in the sensitivity of the human eye to different colors. When calculating the grayscale value, different weights are assigned to the R, G, and B three channels to be more in line with the visual characteristics of the human eye; the component method is to select the value of a certain channel among the R, G, and B three channels of the pixel as the grayscale value. This method can highlight the requirements of a certain color in more detail and is suitable for target minerals with fixed colors; the result of the mineral thin section after grayscaling in this embodiment is as Figure 1 (Result diagram of different calculation methods for converting a color image into a grayscale image).

[0040] After creating the grayscale image, feature enhancement is performed on the grayscale image, mainly including grayscale linear transformation and grayscale non-linear transformation. There are two types of grayscale linear transformation of the image: 1. Image grayscale + fixed value, uniformly increasing the grayscale by one level. 2. Contrast enhancement transformation DB = DA * 1.5, × fixed value, uniformly increasing by a multiple. There are three types of grayscale non-linear transformation of the image: 1. Power transformation, which is a basic non-linear grayscale transformation method. The given formula DB = DA × DA / 255 belongs to a special form of power transformation (here the exponent is 2 and there is a normalization operation). Where DA represents the grayscale value of the original image pixel, and DB represents the grayscale value of the transformed image pixel. This transformation can enhance the contrast of the higher grayscale regions in the image while compressing the contrast of the lower grayscale regions by squaring and normalizing the original grayscale values. 2. Logarithmic transformation, the basic formula of logarithmic transformation is DB = c × log(1 + DA), where DA is the grayscale value of the original image, DB is the transformed grayscale value, and c is a constant used to adjust the amplitude of the transformation. The main function of logarithmic transformation is to stretch the narrow low grayscale range in the image into a wider grayscale range while compressing the wide high grayscale range into a narrower grayscale range. It can enhance the details in the dark part of the image, making the information that was originally difficult to distinguish in the low grayscale region clearer. At the same time, relatively compress the contrast of the high grayscale region so that the bright part will not be too bright and lose details. 3. Gamma transformation, the formula of gamma transformation is DB = c × DAγ, where DA is the grayscale value of the original image, DB is the transformed grayscale value, c is usually taken as 1, and γ is the gamma coefficient. When γ > 1, the grayscale value of the image will become smaller as a whole, the image will become darker, and at the same time, the contrast of the bright part region will be compressed; when γ < 1, the grayscale value of the image will become larger as a whole, the image will become brighter, and at the same time, the contrast of the bright part region will be stretched. This method can flexibly control the brightness and contrast of the image by adjusting the γ value, and specifically enhance or weaken the information in different grayscale regions of the image; the result after enhancing the contrast of the grayscale image information is as Figure 2 (shown in the result diagrams of different methods for enhancing the contrast of grayscale image information).

[0041] As shown in the above method, based on the Python language, extract the reflected light image information parameters of the target mineral and interfering minerals in the image, calculate the difference in their image information, and add exclusion item information. Use a variety of image processing methods to increase the difference in the reflected light image features between the interfering minerals and the target minerals. Finally, accurately limit the range of the target mineral using a fine grayscale range threshold to distinguish the range of the target mineral and the interfering minerals. Subsequently, based on the Python language, extract and learn the interference color, transmitted light, and reflected light image features of the target mineral. The processed image is as Figure 3 (shown in the range limitation and feature extraction of the reflected light, transmitted light, and orthogonal light images of the target mineral).

[0042] S4. Multi-dimensional Information Extraction from Scanning Images and Matching with Target Minerals: Extract multi-dimensional image information from the overall scanning image of the mineral thin section and the target mineral image. After thresholding the image information and algorithm comparison, match the target minerals.

[0043] After calibrating the position information of the minerals in the scanning image and extracting and learning the characteristics of the target minerals, it is necessary to enhance and extract the image information of the overall scanning image of the thin section and the target mineral image. Multiple image information extraction schemes can be used to extract the multi-dimensional image information of the scanning image, including extracting image information in dimensions such as HSV, YCrCb, HLS, XYZ, LAB, and YUV. Among them, HSV (Hue, Saturation, Value) mainly extracts information on hue, saturation, and value. The HSV color space is more in line with human perception of colors and is commonly used in color segmentation and recognition tasks, mainly for identifying objects of specific colors in images. In YCrCb, Y represents luminance, and Cr and Cb represent chrominance. It can separate the luminance information and chrominance information, facilitating compression and processing. HLS (Hue, Lightness, Saturation) refers to hue, lightness, and saturation, which is similar to HSV but represents lightness differently and is also applied in some color processing tasks. XYZ is a color space based on the human visual system and is commonly used in color science and color management. LAB consists of a luminance channel (L) and two chrominance channels (a and b), with perceptual uniformity, and is commonly used in image enhancement and color correction. In YUV, Y represents luminance, and U and V represent chrominance, which is commonly used in video transmission and storage because the data volume can be reduced by reducing the sampling rate of chrominance information. The images after multi-dimensional information extraction are as Figure 4 (shown in the enhanced and extracted image information of the scanning image).

[0044] According to the characteristics of the target minerals extracted and learned in the early stage and the parameters of the target mineral images processed by multi-dimensional information, under the same image processing conditions, traverse the pixels of the overall scanning image of the thin section. Through thresholding the image information of the thin section scanning image and algorithm comparison, compare and verify according to the processing results of the matching minerals in one or more targeted scanning images, exclude the interfering minerals, then locate and proofread the determined target minerals, retain the coordinate information of the target minerals that pass the proofreading, and then create a new layer.

[0045] S5. Encircling the Coordinates of Target Minerals: According to the required particle size of the target minerals, divide the analyzable beam spot area based on whether there is a single clean area inside the target minerals, and screen, label, and encircle the mineral coordinates of the target minerals.

[0046] Since the main purpose of searching for target minerals is to conduct geochemical experiments, and different geochemical experiments require different particle sizes of minerals. For example, LA-ICP-Ms requires some minerals to have a particle size of more than 32 microns, SIMS requires some minerals to reach more than 10 microns, EMPA requires some hydrous minerals to have a particle size of more than 3 microns, and Raman has not many requirements for mineral particle size. Therefore, for different experimental requirements, the conditions for finding target minerals are also different. Taking the LA-ICP-Ms experiment as an example, minerals with a particle size of more than 32 microns need to be found. If the particle size of the found minerals is too small, the work cannot be carried out. Therefore, using the mineral search system, the analyzable beam spot area of whether there is a single clean area in the found target minerals is divided to screen the target minerals suitable for subsequent experiments and divide the levels. Finally, a new layer is drawn, and a square is marked on the original scan map to circle the mineral coordinates. The map after the mineral coordinates are circled is as Figure 5 (Target Mineral Identification and Coordinate Information Acquisition) shown.

[0047] S6. Target Mineral Location: Transform the target mineral coordinates in the overall scanned image into coordinate information that can be marked through the image affine transformation matrix to complete the target mineral location.

[0048] After circling the target mineral coordinates, organize the overall coordinate information of the target minerals, batch export all the coordinate information of the target minerals, store the coordinate information, and then determine the positions of three target minerals during the experiment to determine the coordinate information. Then, through the image affine transformation matrix, stretch, rotate, translate, etc. the coordinate information in the scanned image into the coordinate information corresponding to the experimental conditions and batch import it into the coordinate system of the experimental software. All the target minerals can be marked at one time, and then the automatic dotting experiment can be directly carried out on the target minerals, which can reduce the preset time of the experimental operation as much as possible, as Figure 6 (Target Mineral Location in the Experiment) shown.

[0049] It can be seen from the technical solutions provided by the present method that the present method has strong learning ability and high pertinence, and is more generally applicable to the work of finding target minerals in the research process of various geological bodies. And the test is convenient. Only 1-2 target minerals need to be found manually, and then the matching minerals in the thin section can be automatically identified, searched and located in a targeted manner, and the subsequent work can be automatically screened to avoid excessive final data volume of identification. It can screen out the samples that can be used for the next test work from a large number of samples for the researchers in the early stage of geological research work, and mark the positions in the images of the samples, and convert them into the coordinate information in the actual experiment, which greatly improves the work efficiency of the researchers and saves valuable experimental time.

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the operation process of this method. For other persons, it is also possible to apply this method in other fields without creative efforts, such as finding and locating defective areas on the material surface, etc.

[0051] In this article, specific examples are used to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation on the present invention.

Claims

1. A method for automatically searching and locating target minerals using a microscope, characterized in that, It includes the following steps: S1. Sample preparation: Prepare the sample to be analyzed into a standard rock mineral thin section; S2. Target mineral identification and adjustment of optical information parameters: Manually search for 1-2 groups of target minerals and interfering minerals on the rock thin section, adjust the optical information parameters of transmitted light, reflected light, and cross-polarized light, confirm the image differences between the two, record the optical information parameters at this time, and obtain the overall scanned image of the mineral thin section and the transmitted light, reflected light, and cross-polarized light mineral images of the target minerals and interfering minerals under these optical information parameters through a microscope; S3. Target mineral image range limitation and feature extraction: Use computer language to process the transmitted light, reflected light, and cross-polarized light images of the target minerals and interfering minerals, confirm the image differences between the two, then accurately limit the target mineral image range using the image information threshold, and extract and learn the reflected light, transmitted light, and cross-polarized light image features of the target minerals within the limited range based on computer language; S4. Multi-dimensional information extraction from the scanned image and matching of target minerals: Extract multi-dimensional image information from the overall scanned image of the mineral thin section and the target mineral image, and through image information thresholding processing and algorithm comparison, match and confirm the target minerals; S5. Target mineral coordinate delineation: According to the required particle size of the target minerals, divide the analyzable beam spot area based on whether there is a single clean area within the target minerals, screen, label, and delineate the mineral coordinates of the target minerals; S6. Target mineral positioning: Transform the target mineral coordinates in the overall scanned image into labelable coordinate information through an image affine transformation matrix to complete the target mineral positioning.

2. The method according to claim 1, wherein The step of searching for 1-2 groups of target minerals and interfering minerals on the mineral thin section, adjusting the optical information parameters of transmitted light, reflected light, and cross-polarized light, and confirming the image differences in S2 is specifically as follows: Select the transmitted light that can distinguish the target minerals and interfering minerals using different objective lens magnifications, light sources, and observation methods of an optical microscope; Control the cross-polarized angle of the microscope and select the cross-polarized light that can distinguish the target minerals and interfering minerals to determine the interference color; Select the reflected light that can distinguish the target minerals and interfering minerals by adjusting the reflected light parameter settings.

3. The method according to claim 1, wherein The computer language mentioned in S3 is the Python language.

4. The method according to claim 1, wherein The multi-dimensional image information extraction mentioned in S4 includes extracting image information in HSV, YCrCb, HLS, XYZ, LAB, and YUV dimensions.

5. The method according to claim 1, wherein The step of matching and confirming the target minerals through image information thresholding processing and algorithm comparison in S4 is specifically as follows: According to the results of multi-dimensional image information extraction, traverse the pixels of the overall scanned image of the mineral thin section, and through the thresholding processing and algorithm comparison of the image information of the thin section scanned image, compare and verify according to the processing results of the matching minerals in one or more targeted scanned images, exclude the interfering minerals, and then match, calibrate, and confirm the determined target minerals.

6. The method according to claim 1, wherein When the delineated target minerals are used for geochemical experiments, the labelable coordinate information mentioned in S6 is the coordinate information corresponding to the experimental conditions that can be imported into the coordinate system of the experimental software.

Images