Mineral Identification Systems and Programs

Abstract

To easily and highly accurately identify a mineral by thin piece observation.SOLUTION: A mineral identification system comprises: a polarization microscope body 2 which can observe a polarization characteristic of a thin piece 21 by irradiating the thin piece 21 with polarized light; a smart phone 3 which generates an open nicol image by imaging such a state that the thin piece 21 is observed in open nicol with the polarization microscope body 2 and generates a cross nicol image by imaging such a state that the thin piece 21 is observed in cross nicol; and a mineral identification server 4 which identifies a type of a mineral in the thin piece 21 on the basis of the open nicol image and the cross nicol image. The mineral identification server 4 uses a mineral identification learning model that has been machine-learned on the basis of past performance data so as to identify the type of the mineral in the thin piece 21 when the open nicol image and the cross nicol image are input.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a mineral identification system and a mineral identification program for identifying and specifying minerals by observing thin sections. [Background technology]

[0002] For example, when investigating geology and geological structures, it is sometimes necessary to identify and specify the minerals contained in specimens collected in drilling surveys. In this case, thin sections are sometimes observed using a polarizing microscope. In other words, a thin section of the specimen is placed under polarized light, and the types of minerals in the section are identified based on the differences in the optical properties of each mineral.

[0003] On the other hand, there is a known mineral analysis method that can extract the target mineral almost completely (for example, see Patent Document 1). This method involves photographing a sample containing the mineral with a color television camera connected to a reflecting microscope, performing contour-preserving smoothing processing for each of the three colors red, green, and blue, and synthesizing the processed images to obtain a color image, from which specific minerals can be identified using data on hue, brightness, and purity. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 06-331621 Summary of the Invention [Problem to be solved by the invention]

[0005] However, identifying differences in optical properties in thin section observation requires specialized knowledge and experience, and not only is it difficult for non-experts to distinguish, but interpretations and judgments can differ depending on the expert. In other words, identifying minerals through thin section observation is not something that anyone can easily do, and there is also a risk of variation in the accuracy of identification.

[0006] On the other hand, the mineral analysis method described in Patent Document 1 identifies specific minerals based on the hue, brightness, and purity of images taken with a reflecting microscope, and is not only less accurate than identifying minerals through thin section observation, but can only identify specific, targeted minerals that have been set in advance.

[0007] Therefore, an object of the present invention is to provide a mineral identification system and a mineral identification program that can easily and accurately identify and specify minerals by observing thin sections. [Means for solving the problem]

[0008] In order to solve the above problem, the invention of claim 1 comprises a polarizing microscope main body capable of irradiating a thin section with polarized light to observe the polarization characteristics of the thin section, an imaging unit that images the thin section observed with an open Nicol lens using the polarizing microscope main body to generate an open Nicol image and images the thin section observed with a crossed Nicol lens to generate a crossed Nicol image, and a mineral identification unit that identifies the type of mineral in the thin section based on the open Nicol image and the crossed Nicol image, wherein the mineral identification unit: Identifying the mineral type based also on the location where the thin section sample was mined; A learning model for mineral identification that has been machine-learned based on past performance data is used to identify the type of mineral in the thin section when the open Nicol image and the cross Nicol image are input. The learning model for mineral identification is machine-learned based on past performance data so as to identify the type of the mineral when the location where the thin section sample was mined is also input. The present invention relates to a mineral identification system.

[0010] Claim 2 The invention is as follows: 1 to In the described mineral identification system, the mineral identification unit identifies the type of mineral based also on the depth at which the thin section sample was mined, and the mineral identification learning model has been machine-learned based on past performance data so as to identify the type of mineral when the depth at which the thin section sample was mined is also input.

[0011] Claim 3The invention causes a computer to function as a mineral identification means for identifying the type of mineral in the thin section based on an open Nicol image obtained by photographing a state in which the thin section is observed with an open Nicol and a crossed Nicol image obtained by photographing a state in which the thin section is observed with a crossed Nicol using a polarizing microscope main body capable of irradiating polarized light onto the thin section and observing the polarization characteristics of the thin section, and the mineral identification means comprises: Identifying the mineral type based also on the location where the thin section sample was mined; A learning model for mineral identification that has been machine-learned based on past performance data is used to identify the type of mineral in the thin section when the open Nicol image and the cross Nicol image are input. The learning model for mineral identification is machine-learned based on past performance data so as to identify the type of the mineral when the location where the thin section sample was mined is also input. The present invention relates to a mineral identification program.

[0013] Claim 4 The invention is as follows: 3 In the mineral identification program described in the above, the mineral identification means identifies the type of mineral based also on the depth at which the thin section sample was mined, and the mineral identification learning model has been machine-learned based on past performance data so as to identify the type of mineral when the depth at which the thin section sample was mined is also input. Effect of the Invention

[0014] Claim 1 and claims 3 According to the invention described in the above, the type of mineral in the thin section is automatically identified based on an open Nicol image and a crossed Nicol image obtained by observing the thin section with a polarizing microscope body. Therefore, anyone can identify the mineral by observing the thin section without any variation in accuracy, even if they do not have specialized knowledge or experience. In addition, since it is only necessary to photograph the state in which the thin section is observed with the polarizing microscope body, it is possible to easily and quickly identify the mineral. Moreover, since the type of mineral is identified based on the open Nicol image and the crossed Nicol image using a machine-learned learning model for mineral identification, it is possible to accurately and appropriately identify the type of mineral.

[0015] Claim 1 and claims 3According to the invention described in, in addition to the open Nicol image and the crossed Nicol image, the type of mineral is identified based on the location where the thin section sample was mined, so that the type of mineral can be identified accurately and appropriately. That is, since the polarization characteristics may change depending on the location where the thin section sample was mined, the type of mineral can be identified accurately and appropriately by taking into account the location where the thin section sample was mined.

[0016] Claim 2 and claims 4 According to the invention described in, in addition to the open Nicol image and the crossed Nicol image, the type of the mineral is identified based on the depth at which the thin section sample was mined, so that the type of the mineral can be identified accurately and appropriately. That is, since the polarization characteristics may change depending on the depth at which the thin section sample was mined, the type of the mineral can be identified accurately and appropriately by taking into account the depth at which the thin section sample was mined. [Brief description of the drawings]

[0017] [Figure 1] 1 is a schematic diagram showing a mineral identification system according to an embodiment of the present invention; [Diagram 2] 2A and 2B are diagrams showing a first open Nicol image (a) and a first crossed Nicol image (b) taken with the smartphone of the mineral identification system of FIG. 1. [Diagram 3] 2A and 2B show a second open Nicol image (a) and a second crossed Nicol image (b) taken with the smartphone of the mineral identification system of FIG. 1. [Figure 4] FIG. 2 is a schematic block diagram of a mineral identification server of the mineral identification system of FIG. 1. [Diagram 5] FIG. 5 is a functional block diagram showing a schematic configuration of a learning model for mineral identification in the mineral identification server of FIG. 4. [Figure 6] 2A and 2B are diagrams showing a first open Nicol image (a) and a first crossed Nicol image (b) showing the type of mineral identified by the mineral identification system of FIG. 1. [Figure 7]2A and 2B are diagrams showing a second open Nicol image (a) and a second crossed Nicol image (b) showing the type of mineral identified by the mineral identification system of FIG. 1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, the present invention will be described based on the illustrated embodiment.

[0019] Fig. 1 is a schematic diagram showing a mineral identification system 1 according to an embodiment of the present invention. This mineral identification system 1 is a system for identifying and specifying minerals by observing thin sections, and in this embodiment, it comprises a polarizing microscope main body 2, a smartphone (imaging unit) 3, and a mineral identification server (mineral identification unit) 4. Here, the observation subject is a thin section 21 obtained by processing a specimen collected by a boring survey or the like into a thin piece.

[0020] The polarizing microscope body 2 is a microscope capable of observing the polarization characteristics, optical properties, and birefringence characteristics of the flake 21 by irradiating the flake 21 with polarized light, and has a configuration equivalent to that of a ready-made, commercially available polarizing microscope. That is, a stage 23 for positioning the flake 21 is provided below the mirror column 22, and a plurality of objective lenses 24 are provided above the mirror column 22 so that their positions can be freely changed.

[0021] A polarizer (lower Nicol) 26 is provided directly above the light source 25 arranged below the stage 23, and an analyzer (upper Nicol) 27 is provided above the objective lens 24. Observation with only the polarizer 26 inserted in the optical path is called open Nicol (single Nicol), and observation with both the polarizer 26 and the analyzer 27 inserted in the optical path is called crossed Nicol (crossed Nicol). A straight cylindrical eyepiece 28 is provided above the analyzer 27.

[0022] The smartphone 3 is a multi-function mobile phone equipped with a touch panel (display function, input function), a communication function, a photographing function, a camera, etc., and photographs the state in which the slice 21 is observed in an open Nicol mode with the polarizing microscope main body 2 to generate an open Nicol image, and photographs the state in which the slice 21 is observed in a crossed Nicol mode to generate a crossed Nicol image. That is, the smartphone 3 is set to a photographing mode and the camera is placed on the eyepiece 28 of the polarizing microscope main body 2. At this time, the smartphone 3 may be temporarily fixed to the polarizing microscope main body 2 with an adapter or the like. Then, when the camera is turned on in a state in which the slice 21 is observed in an open Nicol mode, an open Nicol image is photographed and generated, and when the camera is turned on in a state in which the slice 21 is observed in a crossed Nicol mode, a crossed Nicol image is photographed and generated. For example, for the first slice 21, an open Nicol image (a) and a crossed Nicol image (b) as shown in FIG. 2 are generated, and for the second slice 21, an open Nicol image (a) and a crossed Nicol image (b) as shown in FIG. 3 are generated.

[0023] In addition, mineral identification application software (hereinafter referred to as the "mineral identification app") is installed on the smartphone 3, and by starting this mineral identification app, a mineral identification service can be received from the mineral identification server 4. That is, when a mineral identification request including an open Nicol image and a cross Nicol image is sent to the mineral identification server 4 on the mineral identification app, an identification result identified as described below can be received from the mineral identification server 4. At this time, the location and depth where the sample for the thin section 21 was excavated can be included in the mineral identification request as necessary. In this way, by simply installing the mineral identification app, any smartphone 3, that is, anyone, can receive the mineral identification service.

[0024] The mineral identification server 4 is a server for providing a mineral identification service that identifies the type of mineral in the thin section 21, and as shown in Figure 4, mainly comprises a communication unit 41, a mineral identification task (mineral identification unit) 42, a learning task 43, a learning model for mineral identification 44, a performance database for mineral identification 45, and a central processing unit 46 that controls these.

[0025] The communication unit 41 is an interface for communicating with the outside, such as the smartphone 3.

[0026] The mineral identification task 42 is a task program that identifies the type of mineral in the thin section 21 based on a mineral identification request received from the smartphone 3. That is, the task program identifies the type of mineral in the thin section 21 based on the open Nicol image and cross Nicol image in the mineral identification request, and further, if included in the mineral identification request, the location and depth at which the thin section 21, i.e., the specimen, was excavated. Hereinafter, a case where the location and depth at which the sample for the thin section 21 was excavated is included in the mineral identification request will be described.

[0027] Specifically, an open Nicol image and a cross Nicol image, in which the mineral name, mining location, and depth are known, are stored in advance as performance data. Then, taking into consideration the mining location and depth of the mineral identification request, the open Nicol image and the cross Nicol image of the mineral identification request are image-analyzed and compared with the performance data to identify the type of mineral in the thin section 21 (in the open Nicol image and the cross Nicol image). For example, the RGB values ​​of the open Nicol image and the cross Nicol image of the mineral identification request are compared with the RGB values ​​of the open Nicol image and the cross Nicol image of the performance data to identify the type of mineral.

[0028] When the open Nicol image, the cross Nicol image, and the mining location and depth of the sample for the thin section 21 are input, the mineral identification task 42 uses a mineral identification learning model 44 that has been machine-learned based on past performance data to identify the type of mineral in the thin section 21. This mineral identification learning model 44 is created by a learning task 43.

[0029] That is, the learning task 43 creates a learning model for mineral identification 44 by a known machine learning algorithm such as a neural network using past performance data recorded and accumulated in a performance database for mineral identification 45. This performance database for mineral identification 45 is a database in which performance data including, as input information, open Nicol images and crossed Nicol images of thin sections 21 observed in the past, the mining location and depth of the sample, and the type and name of the mineral identified and confirmed by an expert or the like actually observing the thin section 21, are recorded and accumulated. The past performance data includes data created based on the actual open Nicol images and crossed Nicol images, etc., and the type of the mineral actually identified and confirmed, as well as data created for advance training, etc.

[0030] As shown in Fig. 5, this learning task 43 uses machine learning and deep learning using a neural network to create a neural network based on performance data recorded in a mineral identification performance database 45, with an input layer including an open Nicol image, a crossed Nicol image, and the mining location and mining depth of the sample of the thin section 21, an output layer including the type and name of the mineral, and an intermediate layer including image analysis processing from the input layer to the output layer. The learning task 43 then uses the performance data of a mineral identification learning model 44 as learning data to learn various parameters in the intermediate layer. That is, the learning task 43 learns various parameters in the intermediate layer so that the type of mineral is extracted and identified without omission and with high accuracy from the open Nicol image, the crossed Nicol image, and the mining location and mining depth of the sample.

[0031] By such a mineral identification task 42, for example, identification results such as those shown in Figures 6 and 7 are output. That is, for the open Nicol image in Figure 2(a) and the crossed Nicol image in Figure 2(b) above, the open Nicol image in Figure 6(a) and the crossed Nicol image in Figure 6(b) on which the names and symbols of the identified minerals are written are output as identification results. Similarly, for the open Nicol image in Figure 3(a) and the crossed Nicol image in Figure 3(b) above, the open Nicol image in Figure 7(a) and the crossed Nicol image in Figure 7(b) on which the names and symbols of the identified minerals are written are output as identification results.

[0032] Here, the mineral symbol "SAP" in Figure 6 indicates saponite, and it is identified that saponite fills the gaps between rock fragments of basaltic pyroclastic rock. Also, the mineral symbol "GLC" in Figure 7 indicates glauconite, and it is identified that glauconite is produced as pellet-like particles in sandstone. In this way, when a mineral is present in the open Nicol image and the crossed Nicol image, the identification result is output with the mineral's name and symbol clearly indicated for the area where the mineral is present. This makes it easy to visually confirm where the mineral is present, its shape, size, and type.

[0033] Such an identification result is then transmitted to the smartphone 3 that transmitted the mineral identification request, and the identification result is displayed on the touch panel of the smartphone 3. Here, the identification result may be transmitted only to the smartphone 3, or to the smartphone 3 and other devices (for example, a computer at a center that investigates and manages minerals), or may be transmitted only to other devices. In other words, the identification result is transmitted to a predetermined destination that has been registered in advance.

[0034] Next, the operation of the mineral identification system 1 configured as above and the mineral identification method using the mineral identification system 1 will be described.

[0035] First, the thin section 21 to be identified and determined is placed on the stage 23 of the polarizing microscope body 2, and the smartphone 3 with the mineral identification app running is placed on the eyepiece 28 of the polarizing microscope body 2. Next, the polarizing microscope body 2 is set to open Nicols to photograph the thin section 21 with the smartphone 3, and similarly, the polarizing microscope body 2 is set to cross Nicols to photograph the thin section 21 with the smartphone 3. Next, a mineral identification request including the open Nicols image and the cross Nicols image as shown in Figures 2 and 3 is sent to the mineral identification server 4. At this time, if the location and depth where the sample for the thin section 21 was excavated are known, the location and depth are entered on the smartphone 3, included in the mineral identification request, and sent to the mineral identification server 4.

[0036] Upon receiving this mineral identification request, the mineral identification server 4 starts a mineral identification task 42, and as described above, the types of minerals in the thin section 21 (in the open Nicol image and the crossed Nicol image) are identified. Then, as shown in Figs. 6 and 7, the identification results indicating the names and symbols of the minerals identified in the open Nicol image and the crossed Nicol image are transmitted to the smartphone 3 or the like.

[0037] As described above, according to the mineral identification system 1 and the mineral identification method, the type of the mineral in the thin section 21 is automatically identified based on the open Nicol image and the crossed Nicol image obtained by observing the thin section 21 with the polarizing microscope main body 2. Therefore, even if one does not have specialized knowledge or experience, it becomes possible to identify the mineral by observing the thin section without variation in accuracy. In addition, since it is only necessary to take a picture of the state in which the thin section 21 is observed with the polarizing microscope main body 2 with the smartphone 3, it becomes possible to easily and quickly identify the mineral. Moreover, since the type of the mineral is identified based on the open Nicol image and the crossed Nicol image using the machine-learned learning model for mineral identification 44, it becomes possible to accurately and appropriately identify the type of the mineral.

[0038] Furthermore, since the type of mineral is identified based on the location where the sample for the thin section 21 was mined in addition to the open Nicol image and crossed Nicol image, it becomes possible to accurately and properly identify the type of mineral. That is, since the polarization characteristics may change depending on the location where the thin section 21, i.e., the specimen, was mined, it becomes possible to accurately and properly identify the type of mineral by taking into account the location where the sample for the thin section 21 was mined.

[0039] Furthermore, since the type of mineral is identified based on the depth at which the sample for the thin section 21 was mined in addition to the open Nicol image and the crossed Nicol image, it becomes possible to accurately and properly identify the type of mineral. That is, since the polarization characteristics may change depending on the depth at which the sample for the thin section 21 was mined, by taking into account the depth at which the sample for the thin section 21 was mined, it becomes possible to accurately and properly identify the type of mineral.

[0040] Furthermore, by using an existing polarizing microscope body 2 and smartphone 3, identification results can be obtained simply by transmitting open Nicol images, crossed Nicol images, and the like to the mineral identification server 4. In other words, since existing equipment is used at the site of thin section observation and no new equipment is required, it is possible to easily and inexpensively build a mineral identification system 1 and obtain identification results from a variety of locations.

[0041] Although the embodiment of the present invention has been described above in detail, the specific configuration is not limited to this embodiment, and even if there are design changes within the scope of the present invention, they are included in the present invention. For example, in the above embodiment, the polarizing microscope body 2, the smartphone 3 as the photographing unit, and the mineral identification server 4 as the mineral identification unit constitute the mineral identification system 1, the photographing unit may be integrally incorporated into the polarizing microscope body 2, or the photographing unit and the mineral identification unit may be integrally incorporated into the polarizing microscope body 2. The photographing unit may also be composed of a multi-function mobile terminal other than the smartphone 3, a camera, etc. Furthermore, the identification result in which the name and symbol of the mineral is clearly indicated in the area where the mineral exists in the open Nicol image and the cross Nicol image is output, but the identification result in which the area where the mineral exists and the boundary line are colored in a color that can be easily and clearly identified may be output. In addition, the smartphone 3 and the mineral identification server 4 exchange information and data via a mineral identification app, but it is of course possible to exchange information and data manually without using the mineral identification app or the like.

[0042] On the other hand, the above-mentioned mineral identification server 4, i.e., the mineral identification unit, may be configured by installing the following mineral identification program in a general-purpose computer. That is, the computer is made to function as a mineral identification means (mineral identification task 42) that identifies the type of mineral in the thin section based on an open Nicol image taken of a state in which the thin section is observed with an open Nicol and a crossed Nicol image taken of a state in which the thin section is observed with a crossed Nicol, and the location and depth at which the sample for the thin section was excavated, using a polarizing microscope main body capable of observing the polarization characteristics of the thin section by irradiating the thin section with polarized light, and the mineral identification means uses a mineral identification learning model 44 that has been machine-learned based on past performance data to identify the type of mineral in the thin section when the open Nicol image, the crossed Nicol image, and the location and depth at which the sample for the thin section was excavated are input. [Explanation of symbols]

[0043] 1 Mineral Identification System 2 Polarizing microscope body 21 Thin section 3 Smartphone (photography unit) 4. Mineral Identification Server (Mineral Identification Department) 42 Mineral Identification Task (Mineral Identification Section) 44 Learning Model for Mineral Identification

Claims

1. A polarizing microscope body capable of irradiating a thin section with polarized light and observing the polarization characteristics of the thin section; an imaging unit that images the slice observed with an open Nicol microscope using the polarizing microscope body to generate an open Nicol image, and images the slice observed with a crossed Nicol microscope to generate a crossed Nicol image; a mineral identification unit that identifies a type of mineral in the thin section based on the open Nicol image and the crossed Nicol image, wherein the mineral identification unit identifies a type of mineral based on a place where the sample for the thin section was mined, and uses a mineral identification learning model that has been machine-learned based on past performance data so as to identify a type of mineral in the thin section when the open Nicol image and the crossed Nicol image are input, A mineral identification system characterized in that the learning model for mineral identification has been machine-learned based on past performance data so as to identify the type of mineral when the location where the thin section sample was mined is also input.

2. The mineral identification unit identifies the type of the mineral based on the depth at which the thin section sample was mined; The mineral identification learning model is machine-learned based on past performance data so as to identify the type of the mineral when the depth at which the thin section sample was mined is also input.

2. The mineral identification system of claim 1.

3. Computer, a polarizing microscope main body capable of irradiating a thin section with polarized light and observing the polarization characteristics of the thin section functions as a mineral identification means for identifying the type of mineral in the thin section based on an open Nicol image obtained by photographing the state in which the thin section is observed with an open Nicol and a crossed Nicol image obtained by photographing the state in which the thin section is observed with a crossed Nicol; the mineral identification means identifies the type of the mineral based on the location where the thin section sample was mined, and uses a mineral identification learning model that has been machine-learned based on past performance data so as to identify the type of the mineral in the thin section when the open Nicol image and the cross Nicol image are input; The mineral identification learning model is machine-learned based on past performance data so as to identify the type of the mineral when the location where the thin section sample was mined is also input. A mineral identification program.

4. the mineral identification means also identifying the type of mineral based on the depth from which the thin section sample was extracted; The mineral identification learning model is machine-learned based on past performance data so as to identify the type of the mineral when the depth at which the thin section sample was mined is also input.

4. The mineral identification program according to claim 3.

Images