Sediment evaluation system and sediment evaluation method

Abstract

To provide a sediment evaluation system and sediment evaluation method capable of evaluating moisture information of sediment to be conveyed on a belt conveyor.SOLUTION: A sediment evaluation system includes: acquisition means for acquiring an image where sediment to be conveyed on a belt conveyor appears; spectrum acquisition means for acquiring a spectrum image based on the image acquired by the acquisition means; and evaluation means for evaluating moisture information related to moisture contained in the sediment based on the spectrum image acquired by the spectrum acquisition means.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a soil evaluation system and a soil evaluation method for soil transported on a belt conveyor. [Background technology]

[0002] There has been a demand for automation of shield tunnel construction, which is one of the tunnel construction methods. For this reason, in order to automate shield tunnel construction, a technology capable of evaluating the soil quality of the removed soil and sand, such as that described in Patent Document 1, is required.

[0003] Patent Document 1 discloses a system in which vibrations are applied to the soil, images of the soil are taken before and after the vibration, the distribution of particle and soil clump diameters is measured from the images before and after vibration, and the soil quality is evaluated from the changes in the distribution of particle and soil clump diameters before and after vibration. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2018-66613 A Summary of the Invention [Problem to be solved by the invention]

[0005] On the other hand, there is a demand for tunnel shield construction to be carried out in accordance with moisture information, such as the water content, of the soil.

[0006] However, the system for evaluating soil quality of soil disclosed in Patent Document 1 does not assume the evaluation of moisture information. Therefore, the system for evaluating soil quality of soil disclosed in Patent Document 1 has a problem in that it cannot evaluate moisture information.

[0007] Conventionally, the moisture information has generally been measured using a method of drying in an oven or a method of using a microwave oven.

[0008] However, these methods required a day or about 30 minutes to measure the moisture information. Therefore, the moisture information could only be measured a few times a day, and the moisture information of the discharged soil and sand could not be measured continuously in real time. In addition, it was necessary to stop the conveyor belt that transported the soil and sand to measure the moisture information.

[0009] In addition, there are non-contact methods for measuring moisture information using RI (radioisotopes) or microwaves, but these methods have the problem that they are affected by the height of the soil flowing on the conveyor belt that transports it, making it impossible to measure when the height is not uniform.

[0010] The present invention has been developed to solve the above-mentioned problems, and its object is to provide a soil evaluation system and a soil evaluation method that are capable of continuously evaluating moisture information of soil of non-uniform height being transported on a conveyor belt, for example, in the construction of a shield tunnel. [Means for solving the problem]

[0011] The soil evaluation system of the first invention is characterized in that it comprises an acquisition means for acquiring an image containing soil, a spectral acquisition means for acquiring a spectral image based on the image acquired by the acquisition means, and an evaluation means for evaluating moisture information regarding the moisture contained in the soil based on the spectral image acquired by the spectral acquisition means.

[0012] The soil evaluation system of the second invention is characterized in that, in the first invention, the evaluation means sets a specific wavelength included in a predetermined wavelength range as a normalized wavelength based on the spectral image acquired by the spectrum acquisition means, selects a specific wavelength included in the predetermined wavelength range as an evaluation wavelength, and evaluates the moisture information based on the difference in spectral intensity between the normalized wavelength and the evaluation wavelength.

[0013] The soil evaluation system of the third invention is characterized in that, in the second invention, the evaluation means selects a specific wavelength included in the wavelength band of 450 nm to 1700 nm as a first normalized wavelength, and a specific wavelength included in the wavelength band of 900 nm to 1650 nm as a first evaluation wavelength.

[0014] The soil evaluation system of the fourth invention is characterized in that, in the first invention, the evaluation means uses a data set consisting of a previously acquired reference spectral image and an evaluation of moisture information as learning data, and evaluates the moisture information based on the spectral image acquired by the spectrum acquisition means using an evaluation model generated by machine learning, with the reference spectral image as input and the evaluation of the moisture information as output.

[0015] The soil evaluation system of the fifth invention is characterized in that, in the first invention, the spectrum acquisition means extracts wavelengths from the acquired spectral image whose difference from a previously acquired reference wavelength indicating soil is equal to or less than a threshold value, and acquires the spectral image indicating the extracted wavelengths.

[0016] The soil evaluation system of the sixth invention is characterized in that, in the first invention, the spectrum acquisition means uses a data set consisting of a previously acquired reference spectral image and a spectral image representing soil as learning data, takes the reference spectral image as an input and the spectral image representing the soil as an output, and acquires a spectral image representing the soil based on the acquired spectral images using an extraction model generated by machine learning.

[0017] The soil and sand evaluation system of the seventh invention is characterized in that, in the first invention, the acquisition means further acquires a light source image including a whiteboard, and the spectral acquisition means acquires a spectral image based on the relative reflectance of the image to the light source image acquired by the acquisition means.

[0018] The soil and sand evaluation system of the 8th invention is characterized in that, in the 7th invention, the acquisition means acquires imaging condition information indicating the imaging conditions when the light source image was captured, and corrects the image according to the acquired imaging condition information.

[0019] The soil and sand evaluation system of the 9th invention is characterized in that, in the 1st invention, the spectrum acquisition means acquires the spectral image based on wavelengths whose difference from a previously acquired reference wavelength indicating the transporting part transporting the soil is equal to or less than a threshold value.

[0020] The sediment evaluation system of the 10th invention is characterized in that, in the first invention, the acquisition means acquires sediment information indicating the amount of sediment, and the spectrum acquisition means acquires the spectral image based on the image acquired by the acquisition means and the sediment information.

[0021] The soil and sand evaluation system according to an eleventh aspect of the present invention is the first aspect of the present invention, wherein the acquisition means acquires the image using an imaging device capable of imaging linear pixels.

[0022] The soil evaluation method of the 12th invention is characterized in that it has a computer execute an acquisition step of acquiring an image including soil transported on a conveyor belt, a spectral acquisition step of acquiring a spectral image based on the image acquired by the acquisition step, and an evaluation step of evaluating moisture information related to the moisture contained in the soil based on the spectral image acquired by the spectral acquisition step. Effect of the Invention

[0023] According to the first to twelfth inventions, the soil evaluation system and the soil evaluation method evaluate the moisture information of soil based on a spectral image. This makes it possible to continuously evaluate the moisture information of soil of non-uniform height being transported on a belt conveyor, for example, in the construction of a shield tunnel. In addition, since it becomes possible to evaluate the moisture information of the surface of the soil, rather than the moisture information of a single point, the difference in values ​​depending on the measurement point becomes small, allowing for more accurate evaluation.

[0024] In particular, according to the second aspect of the present invention, the moisture information of the soil is evaluated based on the difference in spectral intensity between the normalized wavelength and the evaluation wavelength. This makes it possible to separate the features of the captured spectral image. This allows the moisture information of the soil to be evaluated with higher accuracy.

[0025] In particular, according to the third aspect of the present invention, a specific wavelength included in the wavelength band of 450 nm to 1700 nm is selected as the first normalized wavelength, and a specific wavelength included in the wavelength band of 900 nm to 1650 nm is selected as the first evaluation wavelength. This allows wavelengths in suitable wavelength bands to be used for the first normalized wavelength and the first evaluation wavelength, respectively, enabling more accurate evaluation.

[0026] In particular, according to the fourth aspect of the present invention, the moisture information of the soil is evaluated based on the spectral image using the evaluation model, which makes it possible to evaluate the moisture information of the soil with high accuracy by machine learning.

[0027] In particular, according to the fifth aspect of the present invention, wavelengths whose difference from a reference wavelength indicating the wavelength of the sediment obtained in advance are equal to or less than a threshold are extracted from the image, and a spectral image indicating the extracted wavelengths is obtained. This makes it possible to extract only the sediment contained in the spectral image and eliminate other noise, thereby enabling more accurate evaluation of the moisture information of the sediment.

[0028] In particular, according to the sixth aspect of the present invention, a spectral image showing sediment is obtained based on the spectral image using an extraction model, which makes it possible to extract only the sediment contained in the spectral image and eliminate other noise, thereby enabling more accurate evaluation of the moisture information of the sediment.

[0029] In particular, according to the seventh aspect of the present invention, a spectral image based on the relative reflectance of the image to the light source image is acquired. This allows the use of an image with noise removed, since the spectral image is acquired based on the light source image as a reference. Therefore, even if the lighting equipment that irradiates the soil and sand breaks down or deteriorates, it is possible to evaluate moisture information with high accuracy.

[0030] In particular, according to the eighth aspect of the present invention, imaging condition information is acquired, and the image is corrected according to the acquired imaging condition information. This makes it possible to correct conditions such as the amount of light during shooting, and to use an image with noise removed.

[0031] In particular, according to the ninth aspect of the present invention, a spectral image is acquired based on wavelengths whose difference from a reference wavelength representing a transport part is equal to or less than a threshold value. This makes it possible to extract a spectral image representing sediment by removing a spectrum representing a transport part such as a belt conveyor that transports sediment from the spectral image, thereby enabling evaluation of moisture information with higher accuracy.

[0032] In particular, according to the tenth aspect of the present invention, a spectral image is acquired based on an image and sediment information. As a result, for example, if the height of sediment indicated by the sediment information is equal to or less than a reference value, it is possible to determine that there is no sediment in the corresponding location and remove the corresponding location, thereby making it possible to extract a spectral image showing sediment with higher accuracy.

[0033] In particular, according to the eleventh aspect of the present invention, the acquiring means acquires an image using an imaging device capable of imaging pixels on a straight line. This makes it possible to easily acquire an image, for example, by imaging a straight line perpendicular to the transport direction of the transporting part that transports the soil and sand with a camera or sensor capable of line scanning. [Brief description of the drawings]

[0034] [Figure 1] FIG. 1 is a schematic diagram showing an example of the configuration of a soil and sand evaluation system in this embodiment. [Diagram 2] FIG. 2(a) is a schematic diagram showing an example of the configuration of the soil evaluation device in this embodiment, and FIG. 2(b) is a schematic diagram showing an example of the function of the soil evaluation device in this embodiment. [Diagram 3] FIG. 3 is a diagram showing an example of a flowchart of the operation of the soil and sand evaluation system in this embodiment. [Figure 4] Fig. 4(a) is a diagram showing an example of an image of soil and sand in this embodiment, and Fig. 4(b) is a diagram showing an example of a spectral image of soil and sand in this embodiment. [Diagram 5] FIG. 5 is a diagram showing an example of a graph showing the average soil height versus elapsed time in this embodiment. [Figure 6] FIG. 6 is a diagram showing an example of an extraction model in this embodiment. [Figure 7] FIG. 7 is a diagram showing an example of a graph showing the measured water content versus elapsed time in this embodiment. [Figure 8] FIG. 8 is a graph showing an example of the spectral intensity for each wavelength. [Figure 9] FIG. 9 is a diagram showing an example of a graph showing the water content versus NDSI in this embodiment. [Figure 10] FIG. 10 is a diagram showing an example of an evaluation model in this embodiment. [Figure 11] FIG. 11 is a diagram showing an example of a graph showing the measured water content versus the actual water content in this embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0035] Hereinafter, an example of a soil evaluation system according to an embodiment of the present invention will be described with reference to the drawings.

[0036] Fig. 1 is a schematic diagram showing an example of the configuration of a sediment evaluation system 100. For example, as shown in Fig. 1, the sediment evaluation system 100 includes a sediment evaluation device 1, a server 3, and a user terminal 2 having an imaging device 5 and a lighting fixture 6, which are connected via a public communication network 4. Alternatively, the sediment evaluation system 100 may include only the sediment evaluation device 1 and the imaging device 5, which are connected without going through the public communication network 4.

[0037] The server 3 is a storage medium that stores various data such as images transmitted from the sediment evaluation device 1 and the user terminal 2. Furthermore, the server 3 transmits the stored various data to the sediment evaluation device 1 and the user terminal 2 as necessary. The server 3 may, for example, have at least some of the functions of the sediment evaluation device 1, and may, for example, perform at least some of the processing in place of the sediment evaluation device 1.

[0038] The public communication network 4 is, for example, an Internet network to which the sediment evaluation device 1 is connected via a communication circuit. The public communication network 4 may be configured as a so-called optical fiber communication network. The public communication network 4 may be realized by a known communication technology such as a wired communication network or a wireless communication network. The user terminal 2 is owned by, for example, a user of a service using the sediment evaluation system 100, and is connected to the sediment evaluation device 1 via the public communication network 4. The user terminal 2 may represent, for example, an electronic device that generates a database. The user terminal 2 may be, for example, an electronic device such as a personal computer or a tablet terminal. The user terminal 2 may have at least a part of the functions of the sediment evaluation device 1. The user terminal 2 may have a display or a speaker (not shown) that can present the evaluation result of the moisture information of the sediment 7 to the user.

[0039] The lighting fixture 6 is any lighting fixture that illuminates the dirt 7 when the imaging device 5 captures an image including the dirt 7 of the user. The lighting fixture 6 may emit light of any wavelength. The lighting fixture 6 may be a halogen light or a fiber optic light source. The lighting fixture 6 may also use two light sources, a halogen light and a fiber optic light source.

[0040] The imaging device 5 is a known camera that captures the user's soil 7 to generate an image and a spectral image. For example, an RGB camera, a multispectral camera, a target spectrum camera, a hyperspectral camera, etc. may be used as the imaging device 5. The imaging device 5 may also be a camera that captures a monochrome image showing the intensity of light at a specific wavelength. The imaging device 5 may also be a light receiving sensor that measures the intensity of light at a specific wavelength. The imaging device 5 may also be, for example, a video camera that captures a video, and may be built into the soil evaluation device 1. When the imaging device 5 is a spectral video camera, for example, a spectral image may be extracted from a part of the captured video. The imaging device 5 outputs the captured image to the user terminal 2. The imaging device 5 may also output the captured image to the soil evaluation device 1 via the public communication network 4 without going through the user terminal 2. The imaging device 5 may also be a camera or sensor that can line scan a straight line perpendicular to the transport direction of the transport unit that transports the soil. The imaging device 5 may also be equipped with a moving unit (not shown) for moving on the transport unit 8 according to the progress of the tunnel construction. This makes it possible to capture an image of the soil 7 immediately after excavation.

[0041] The soil 7 is, for example, soil, gravel, stones, etc., removed by shield construction of a tunnel. The soil 7 is piled up in, for example, a transport section 8 and transported.

[0042] The transport unit 8 is any equipment for transporting soil and sand. The transport unit 8 is, for example, a belt conveyor that transports the soil and sand 7 in a transport direction x.

[0043] The sediment evaluation device 1 performs processing based on the image and the spectral image output from the user terminal 2. The sediment evaluation device 1 may be an electronic device such as a personal computer (PC), or may be an electronic device such as a smartphone, a tablet terminal, a wearable terminal, an IoT (Internet of Things) device, or a single board computer such as Raspberry Pi (registered trademark), and may have an imaging device 5 built in.

[0044] The soil evaluation device 1 acquires, for example, an image including soil 7 output from an imaging device 5, acquires a spectral image based on the acquired image, and evaluates moisture information of the soil 7 based on the acquired spectral image.

[0045] Next, an example of the soil evaluation device 1 in this embodiment will be described with reference to Fig. 2. Fig. 2(a) is a schematic diagram showing an example of the configuration of the soil evaluation device 1 in this embodiment, and Fig. 2(b) is a schematic diagram showing an example of the function of the soil evaluation device 1 in this embodiment.

[0046] 2(a), the soil evaluation device 1 includes a housing 10, a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a storage unit 104, and I / Fs 105 to 107. The CPU 101, the ROM 102, the RAM 103, the storage unit 104, and the I / Fs 105 to 107 are connected via an internal bus 110.

[0047] The CPU 101 controls the entire sediment evaluation device 1. The ROM 102 stores the operation code of the CPU 101. The RAM 103 is a working area used when the CPU 101 is operating. The storage unit 104 stores various information such as images, spectral images, extraction models, and evaluation models. The storage unit 104 may be, for example, a data storage device such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), an SD card, or a miniSD card. For example, the sediment evaluation device 1 may have a GPU (Graphics Processing Unit) (not shown).

[0048] The I / F 105 is an interface for transmitting and receiving various types of information via the public communication network 4. The I / F 106 is an interface for transmitting and receiving information to and from the input unit 108. For example, a keyboard is used as the input unit 108, and a user or the like who uses the sediment evaluation device 1 inputs various types of information or control commands for the sediment evaluation device 1 via the input unit 108. The I / F 107 is an interface for transmitting and receiving various types of information to and from the display unit 109. The display unit 109 outputs various types of information such as the evaluation results stored in the storage unit 104, or the processing status of the sediment evaluation device 1. A display is used as the display unit 109, and may be, for example, a touch panel type.

[0049] The storage unit 104 stores, for example, images and spectral images acquired from the imaging device 5, as well as information related to the spectral images, algorithms used to evaluate moisture information, extraction models, evaluation models, and the like.

[0050] The display unit 109 displays various information, such as evaluation results and images.

[0051] Fig. 2(b) is a schematic diagram showing an example of the functions of the sediment evaluation device 1. The sediment evaluation device 1 includes an acquisition unit 11, an image processing unit 12, an evaluation unit 13, a storage unit 14, an output unit 15, a conversion unit 16, an extraction unit 17, and a control unit 18. The acquisition unit 11, the image processing unit 12, the evaluation unit 13, the storage unit 14, the output unit 15, the conversion unit 16, the extraction unit 17, and the control unit 18 shown in Fig. 2(b) are realized by the CPU 101 using the RAM 103 as a working area to execute a program stored in the storage unit 104 or the like, and may be controlled by artificial intelligence, for example.

[0052] The acquisition unit 11 acquires images and spectral images. The acquisition unit 11 may acquire images captured by the imaging device 5 via the user terminal 2, or may acquire images and spectral images from, for example, the imaging device 5 built into the sediment evaluation device 1. The acquisition unit 11 may also acquire images and spectral images from the server 3 or the like via the public communication network 4. The frequency and period at which the acquisition unit 11 acquires various types of information are arbitrary. The acquisition unit 11 may also acquire sediment information indicating the volume of the sediment 7.

[0053] The image processing unit 12 performs processes such as correction on the image acquired by the acquisition unit 11 .

[0054] The conversion unit 16 obtains a spectral image based on the image processed by the image processing unit 12. In this case, for example, the conversion unit 16 converts the image into a spectral image indicating the spectrum contained in the image.

[0055] The extraction unit 17 extracts a spectral image showing the soil 7 from the spectral image acquired by the conversion unit 16 .

[0056] The evaluation unit 13 evaluates the moisture information based on the spectral image extracted by the extraction unit 17 .

[0057] The control unit 18 controls various equipment based on the moisture information evaluated by the evaluation unit 13.

[0058] The memory unit 14 retrieves various pieces of information stored in the storage unit 104 as necessary. The memory unit 14 stores in the storage unit 104 various pieces of information acquired or output by the acquisition unit 11, the image processing unit 12, the evaluation unit 13, the conversion unit 16, the extraction unit 17, and the control unit 18.

[0059] The output unit 15 outputs various information and transmits the evaluation result to the display unit 109 via the I / F 107.

[0060] Next, a description will be given of an example of the operation of the sediment evaluation system 100 in this embodiment. Fig. 3 is a flowchart showing an example of the operation of the sediment evaluation system 100 in this embodiment.

[0061] First, in step S110, the acquisition unit 11 acquires an image including the soil 7. In this case, for example, an image captured by the imaging device 5 may be acquired via the public communication network 4. When the imaging device 5 captures an image, the soil 7 may be irradiated with light using the lighting device 6, but this is not limited thereto, and the image may be captured without using the lighting device 6. For example, an image including the soil 7 stored in the server 3 may be acquired via the public communication network 4. The acquisition unit 11 may acquire an image captured by the imaging device 5 via the I / F 105. The acquisition unit 11 stores the image in the storage unit 104, for example, via the storage unit 14. The image including the soil 7 is, for example, an image including the soil 7 as shown in FIG. 4(a). In addition, in step S110, the imaging device 5 may capture an image of the soil 7 being transported by the transport unit 8. In addition, in step S110, the imaging device 5 may capture a video of the soil 7 being transported by the transport unit 8. Here, the moving image of the soil 7 may be a plurality of images of the soil 7 that are continuous in time.

[0062] Furthermore, in step S110, the acquisition unit 11 may acquire sediment information. The sediment information is information indicating the volume of sediment. The sediment information may be information indicating the height or width of the sediment, for example. The sediment information may be a graph indicating the average sediment height versus elapsed time, for example, as shown in FIG. 5. In step S110, the acquisition unit 11 may acquire the sediment information using any sensor, such as a laser distance sensor.

[0063] Furthermore, in step S110, the acquisition unit 11 may acquire a light source image including a white board. The light source image is an image including a white board, and may be, for example, an image including a white board irradiated with light equivalent to the light irradiated on the soil 7. The white board is a reference board that requires a certain reflectance. The white board may be provided on the transport unit 8, for example. Furthermore, the acquisition unit 11 may acquire a light source image at any period. Furthermore, the acquisition unit 11 may acquire an image including the soil and the white board.

[0064] Next, in step S120, the image processing unit 12 performs processing such as correction on the image acquired by the acquisition unit 11 in step S110. The image processing unit 12 smoothes the image, for example. In this case, the image processing unit 12 first smoothes the image using, for example, a median filter, a Gaussian filter, a moving average, a Savitzky-Golay method, or the like. The image processing unit 12 may also smooth the image, for example, and sharpen the image using a Laplacian filter (second order differential), an unsharp method, or the like based on the smoothed image and the image before smoothing. In this case, the image processing unit 12 calculates, for example, the difference between the smoothed image and the image before smoothing, and sharpens the image so as to increase the difference. Smoothing is a process of blurring an image, for example, a process of reducing the change in pixel values ​​of an image. Sharpening is a process of increasing the change in pixel values ​​of an image, for example.

[0065] Next, in step S130, the conversion unit 16 acquires a spectral image based on the image processed by the image processing unit 12. In step S130, the conversion unit 16 acquires a spectral image, for example, as shown in Fig. 4(b). The conversion unit 16 acquires a spectral image based on the relative reflectance of an image with respect to a light source image including, for example, a white board on which light is irradiated to the soil 7. This makes it possible to calculate and compare moisture information and the relative reflectance of other components regardless of fluctuations in the illumination light source.

[0066] Furthermore, in step S110, the conversion unit 16 may acquire a spectral image based on the relative reflectance with respect to the white board when a white board is included in the image acquired by the acquisition unit 11. In this case, in step S110, the conversion unit 16 may perform correction based on the relative reflectance with respect to the white board when a spectrum indicating a preset wavelength of the white board is included in the image acquired by the acquisition unit 11.

[0067] Furthermore, in step S130, the conversion unit 16 may convert each image acquired by the acquisition unit 11 in step S110 into a spectral image. For example, the image including the soil 7 shown in Fig. 4(a) acquired in step S110 may be converted into a spectral image as shown in Fig. 4(b).

[0068] In step S130, the converter 16 may correct the spectral intensity indicated by the spectral image. In this case, the converter 16 may perform dark current noise correction on the spectral intensity I(λ) using, for example, the formula shown in formula (1). Dark current noise is noise caused by dark current that occurs due to unpredictable thermal noise in a solid-state imaging element such as a CCD (Charge Coupled Devices) image sensor or a CMOS (Complementary Metal Oxide Semiconductor) image sensor. In addition, λ is the wavelength, I(λ) is the spectral intensity, Ic(λ) is the spectral intensity after dark current correction, D(λ) is the dark current noise, E(λ) is the light source intensity (spectral distribution), S(λ) is the sensor sensitivity (spectral sensitivity), and R(λ) is the reflectance (spectral reflection intensity).

number

[0069] In step S130, the conversion unit 16 may correct the image based on the imaging conditions, which are conditions when the light source image is captured, and the imaging conditions when the soil 7 is captured. The imaging conditions are, for example, the gain, exposure time, brightness, etc. when capturing the image. For example, the conversion unit 16 corrects the gain and exposure time of the image based on the difference between the light intensity of the lighting device 6 when the light source image is captured and the light intensity of the lighting device 6 when the image acquired in step S110 is captured. The conversion unit 16 may perform the reference plate correction, for example, using equation (2). In addition, Istd(λ) indicates the standard reference plate data, Icstd(λ) indicates the standard reference plate data after dark current correction, and the relative reflectance of the reference plate.

number

[0070] Next, in step S140, the extraction unit 17 extracts the sediment 7 from the spectral image acquired in step S130. For example, the extraction unit 17 extracts wavelengths whose difference from a reference wavelength indicating the wavelength of the sediment 7 acquired in advance is equal to or less than a threshold value from the spectral image, and acquires a spectral image indicating the extracted wavelengths. For example, the extraction unit 17 extracts a spectrum that is close to the reference wavelength indicating the wavelength of the sediment 7 acquired in advance using a Spectral Angle Mapper (SAM) method or the like. The reference wavelength is, for example, any wavelength selected from a wavelength range of 450 nm to 1650 nm. The reference wavelength may be, for example, a wavelength or spectrum set relatively to the wavelength or spectrum of a reference object such as a whiteboard. This makes it possible to extract only the sediment 7 included in the image and eliminate noise. The extraction unit 17 may also remove noise from the spectral image acquired in step S130 based on a value of a normalized difference spectral index (NDSI) to be described later. In such a case, for example, the relative reflectance when the wavelength λ1 is 1250 nm is I 1250 , the relative reflectance when the wavelength λ2 is 1450 nm is I 1450 When the NDSI×100 value is set as 1, the area where the NDSI×100 value is outside the range of 5 to 60 may be removed as noise. Furthermore, the extraction unit 17 may extract a spectrum indicating the sediment 7 based on the ratio of pixels indicating the sediment 7 contained in one line of pixels of the spectral image acquired in step S130. In this case, when the ratio of pixels indicating the sediment 7 contained in one line of pixels is 50% or less, the extraction unit 17 may determine that there is no sediment 7 on the line or that the water content is 0%. Noise includes, for example, things other than the sediment 7 to be evaluated, shadow parts, halation, etc. Furthermore, noise includes the transport unit 8 and other background.

[0071] Furthermore, in step S140, the extraction unit 17 may extract the sediment 7 based on the spectral image acquired in step S130 and the sediment information acquired in step S110. In this case, for example, when the height of the sediment 7 indicated by the sediment information is equal to or less than a reference value, the extraction unit 17 may determine that there is no sediment 7 in the corresponding location and remove the corresponding location on the spectral image.

[0072] Furthermore, in step S140, the extraction unit 17 may acquire a new spectral image from the spectral image acquired in step S130, based on wavelengths whose difference from a previously acquired reference wavelength indicating the transport unit 8 is equal to or smaller than a threshold value. In step S140, the extraction unit 17 may acquire a new spectral image, for example, by excluding a spectrum indicating the transport unit 8 from the spectral image acquired in step S130. In such a case, for example, a spectrum that is close to the reference wavelength indicating the transport unit 8 may be excluded. Furthermore, the extraction unit 17 may remove the transport unit 8 from the spectral image acquired in step S130, based on the numerical value of NDSI or BDSI (bilateral differential spectral index) described later. In such a case, for example, the relative reflectance when the wavelength λ1 is 1050 nm is calculated as I 1050 , the relative reflectance when the wavelength λ2 is 1350 nm is I 1350 When the value of NDSI×100 is less than 0, the portion may be removed as the conveying portion 8. In addition, the extraction unit 17 may calculate the relative reflectance when the wavelength λ1 is 1050 nm as I 1050 , the relative reflectance when the wavelength λ2 is 1350 nm is I 1350 , the relative reflectance at wavelength λ3 is I 1600 When the value of BDSI×100 is −7 or more, the portion may be removed as the transport portion 8.

[0073] In addition, in step S140, the extraction unit 17 may use a data set consisting of a previously acquired reference spectral image and a sediment spectral image showing the spectrum of sediment 7 as learning data, and extract a sediment spectral image based on the spectral image acquired in step S130 using an extraction model generated by machine learning, with the reference spectral image as input and the spectral image showing sediment 7 as output.

[0074] As a method for generating the extraction model, for example, machine learning based on a neural network model may be used to generate the extraction model. The extraction model may be generated using machine learning based on a neural network model such as a CNN (Convolution Neural Network), or any other model may be used. In addition, as a method for generating the extraction model, for example, linear discrimination, support vector machine, k-nearest neighbor method, random forest, deep learning, etc. may be used to generate the extraction model.

[0075] In this case, the extraction model stores the correlation having the correlation between the reference spectral image, which is the input data, and the sediment spectral image, which is the output data, showing the spectrum of the sediment 7, as shown in FIG. 6. The correlation indicates the degree of correlation between the input data and the output data, and it can be determined that the higher the correlation, the stronger the connection between the data. The correlation may be indicated by three or more values ​​or three or more stages, such as a percentage, or may be indicated by two values ​​or two stages. The reference spectral image is a spectral image including the sediment 7 to be used for the learning data acquired in advance. The sediment spectral image may be a spectral image acquired by the same method as in step S140. The sediment spectral image may be a spectral image obtained by extracting the spectrum of the sediment 7 that is close to a reference wavelength indicating the wavelength of the sediment 7 acquired in advance from a spectral image including the sediment 7, using the SAM method or the like.

[0076] For example, the association is constructed by the degree of association between a plurality of input data, pairs, and a plurality of output data. The association is appropriately updated in the process of machine learning, and indicates a classifier using a function optimized, for example, based on a plurality of input data and a plurality of output data. Note that the association may have, for example, a plurality of association degrees indicating the degree of association between each data. For example, when the database is constructed by a neural network, the association degree can correspond to a weight variable. For example, as shown in FIG. 6, the association may indicate the degree of association between a plurality of input data and a plurality of output data. In this case, by using the association, the degree of relationship with the plurality of output data of "sediment spectrum image A" to "sediment spectrum image C" in FIG. 6 can be linked and stored. Therefore, for example, a plurality of input data can be linked to one output data via the association. This makes it possible to realize a multifaceted selection of output data for the input data.

[0077] The correlation has, for example, a plurality of correlations linking each input data with each output data. The correlation is expressed in three or more levels, for example, a percentage, a 10-level scale, or a 5-level scale, and is expressed, for example, by the characteristics of the line (for example, thickness, etc.). For example, the "reference spectral image A" included in the input data shows a correlation AA of "73%" with the "sediment spectral image A" included in the output data, and a correlation AB of "12%" with the "sediment spectral image B" included in the output data. In other words, the "correlation" indicates the degree of connection between each data, and for example, the higher the correlation, the stronger the connection between each data.

[0078] Three or more levels of correlation as shown in Fig. 6 are acquired in advance. In other words, past data sets are accumulated to show which of the input data and output data was adopted and evaluated when determining the actual solution, and these are analyzed to create the correlation shown in Fig. 6.

[0079] For example, let us say that in the past, "Sediment Spectral Image B" was judged to be the most suitable input data for "Reference Spectral Image B" and evaluated as being the most suitable. By collecting and analyzing such data sets, the degree of correlation between the input data and the output data becomes stronger.

[0080] This analysis may be performed by artificial intelligence. In such a case, for example, if there are many cases where "sediment spectroscopic image B" is estimated for input data of "reference spectroscopic image B," the degree of association between this "reference spectroscopic image B" and "sediment spectroscopic image B" is set higher.

[0081] This correlation may be formed by a node of a neural network in artificial intelligence. That is, the weighting coefficient for the output of the node of the neural network corresponds to the correlation. In addition, it is not limited to a neural network, and may be formed by any decision-making factor that constitutes artificial intelligence.

[0082] In addition, the extraction model may be machine-learned by providing at least one or more hidden layers between the input data and the output data. The above-mentioned correlation is set in either the input data or the hidden layer data, or in both, and this becomes the weighting of each data, and the output is selected based on this. Then, when this correlation exceeds a certain threshold, the output may be selected.

[0083] Such correlation is what is called learned data in artificial intelligence. After creating such learned data, a new sediment spectrum image is actually estimated from the new spectrum image. In this case, a new sediment spectrum image is acquired for the spectrum image acquired in step S130. When making the estimation, for example, the correlation shown in FIG. 6 acquired in advance is referenced. For example, if the newly acquired spectrum image is the same as or similar to the "reference spectrum image A", it is associated with the "sediment spectrum image A" through the correlation, with a correlation AA of "73%" and with the "sediment spectrum image B" of "12%". In this case, the "sediment spectrum image A" with the highest correlation is selected as the optimal solution. However, it is not essential to select the one with the highest correlation as the optimal solution, and the "sediment spectrum image B" with a low correlation but a correlation itself that is recognized may be selected as the optimal solution. In addition, it is of course possible to select an output solution that is not connected by an arrow other than the above, and any other priority order may be selected as long as it is based on the correlation.

[0084] By referring to such correlations, it is possible to quantitatively select output data suitable for input data, not only when the spectral image is identical to or similar to the input data, but also when the spectral image is dissimilar.

[0085] Furthermore, in step S140, the extraction unit 17 may extract a sediment spectrum indicating the sediment 7 by extracting a range 71 of preset coordinates from the spectral image acquired in step S130. The range 71 may be, for example, a range having a width a and a length b as shown in FIG. 4(b). The width a and the length b may be any values. The range 71 may be determined based on the width of the transport unit 8, for example.

[0086] Next, in step S150, the evaluation unit 13 evaluates the moisture information of the soil 7 based on the soil spectrum image extracted in step S140. The moisture information is information related to the moisture contained in the soil 7, and is, for example, information indicating the moisture content of the soil 7. The moisture information may also be information indicating the proportion of the mass or volume of the moisture contained in the soil 7. The moisture information may also be a graph showing the moisture content versus elapsed time, as shown in FIG. 7. The moisture content is the proportion of the mass of moisture contained per certain mass of the soil 7.

[0087] In step S150, for example, the evaluation unit 13 selects a specific wavelength included in a predetermined wavelength range as a normalized wavelength based on the sediment spectrum image, selects a specific wavelength included in the predetermined wavelength range as an evaluation wavelength, and evaluates moisture information of the sediment 7 based on the difference in spectral intensity between the normalized wavelength and the evaluation wavelength. The normalized wavelength is a wavelength that is to be normalized in order to evaluate the moisture information while reducing the effects of different conditions, such as uneven lighting and shadows. The evaluation wavelength is a wavelength used for evaluation to normalize the normalized wavelength.

[0088] FIG. 8 shows a number of spectrum graphs shown by the spectrum image. FIG. 8 is a graph with the vertical axis representing light intensity and the horizontal axis representing wavelength [nm]. The solid and dashed lines of the multiple spectrum graphs correspond to the spectra of the image of the soil and sand 7 in different states. The evaluation unit 13 selects, for example, a specific wavelength included in a wavelength band of 450 nm to 1700 nm as the first normalized wavelength, and a specific wavelength included in a wavelength band of 900 to 1650 nm as the first evaluation wavelength. The evaluation unit 13 may select, for example, 1250 nm as the first normalized wavelength, and 1450 nm as the first evaluation wavelength.

[0089] The evaluation unit 13 selects wavelengths included in such a spectrograph as specific wavelengths and as normalized wavelengths and evaluation wavelengths. The evaluation unit 13 may select a wavelength at which the difference value of the spectral intensity between each spectrograph is the largest as the specific wavelength. The evaluation unit 13 may also specify a singular point at which a convex peak is formed on each spectrograph as the specific wavelength. The evaluation unit 13 may also select a specific wavelength by analysis using a trained model using regression analysis, brute force analysis, machine learning, or the like.

[0090] Here, for example, the normalized wavelength and the evaluation wavelength may be specified at one point or multiple points. In addition, a wavelength range may be set with these normalized wavelength and evaluation wavelength as the center. The wavelength range may be configured as a predetermined wavelength range set in advance, such as a wavelength width in which the difference between the normalized wavelength and the evaluation wavelength is ±10 nm. Therefore, if the normalized wavelength is 780 nm and the wavelength range is ±10 nm, the range in which the spectrum data is actually evaluated is 770 to 790 nm. In such a case, the normalized wavelength and the evaluation wavelength may be determined, for example, by setting the central wavelength of each wavelength range as the specific wavelength.

[0091] Next, the evaluation unit 13 may evaluate the moisture information based on the difference in spectral intensity between the selected normalized wavelength and the evaluation wavelength.

[0092] The evaluation unit 13 calculates moisture information from a spectral change due to, for example, the difference in spectral intensity between the normalized wavelength and the evaluation wavelength. In this case, by normalizing with the sum of the spectral intensity of the normalized wavelength and the evaluation wavelength, even if conditions are different, such as uneven lighting or shadows, the influence of these factors can be reduced and the moisture amounts can be compared. The moisture information is calculated using, for example, a known spectrum measurement method, a spectrum analysis technique (for example, NDSI), or the like, according to equation (3). "Iλ" is, for example, "λ nm ", the normalized wavelength is "λ2" and the evaluation wavelength is "λ1".

number

[0093] Furthermore, the evaluation unit 13 may use the BDSI to determine the reflection index from equation (4). In this case, λ1, λ2, and λ3 may be, for example, 1250 nm, 1450 nm, and 1600 nm, respectively.

number

[0094] The evaluation unit 13 evaluates the moisture information based on the NDSI or BDSI. In this case, the moisture information may be evaluated based on the NDSI with reference to a graph showing the moisture ratio versus the NDSI, such as that shown in Fig. 9. The graph showing the moisture ratio versus the NDSI, such as that shown in Fig. 9, may be created in advance, for example, based on a correlation between an actual measurement value measured in advance and the NDSI.

[0095] The evaluation unit 13 may assign moisture information from the distribution of pixels for each part based on, for example, the normalized spectral reflectance index of the spectral image. In this case, the evaluation unit 13 may evaluate the moisture information based on the sediment information acquired in step S110 and the spectral image extracted in step S140. For example, the evaluation unit 13 may evaluate moisture information indicating the moisture content for each range 71. The evaluation unit 13 may also calculate the amount of sediment contained in the range 71 from the sediment information indicating the height of the sediment 7 acquired in step S110 and the area of ​​the range 71, and evaluate moisture information such as a percentage indicating the water content from the calculated amount of sediment and the moisture amount.

[0096] The evaluation unit 13 generates an evaluation result by using format data such as an output format stored in the storage unit 104. The evaluation unit 13 stores the evaluation result in the storage unit 104 via the memory unit 14, for example.

[0097] In addition, in step S150, the evaluation unit 13 may use a data set consisting of a reference spectral image and moisture information acquired in advance as learning data, and may use an evaluation model generated by machine learning with the reference spectral image as input and the moisture information as output to evaluate the moisture information contained in the soil 7 based on the spectral image. The reference spectral image is a spectral image used as learning data for the evaluation model, and is, for example, a spectral image showing the soil 7 acquired in advance, but is not limited thereto, and any image including a spectrum showing the soil 7 may be used. The moisture information may also be moisture information acquired by a method similar to that of step S150. That is, the moisture information may be moisture information acquired based on the spectral image by selecting a specific wavelength included in a predetermined wavelength range as a normalized wavelength, selecting a specific wavelength included in the predetermined wavelength range as an evaluation wavelength, and determining the difference in spectral intensity between the normalized wavelength and the evaluation wavelength.

[0098] As a method for generating the evaluation model, for example, machine learning based on a neural network may be used to generate the evaluation model. The evaluation model may be generated using machine learning based on a neural network such as a CNN (Convolution Neural Network), or any other model may be used.

[0099] In this case, the evaluation model stores, for example as shown in FIG. 10, an association having a degree of association between the reference spectroscopic image, which is input data, and the evaluation of moisture information, which is output data.

[0100] In addition, the evaluation model may be machine-learned by providing at least one or more hidden layers between the input data and the output data. The above-mentioned correlation is set in either the input data or the hidden layer data, or in both, and this becomes the weighting of each data, and the output is selected based on this. Then, when this correlation exceeds a certain threshold, the output may be selected.

[0101] After creating such learned data, the evaluation of moisture information will be actually estimated from a new spectral image. In this case, a new evaluation of moisture information for the spectral image extracted in step S140 is obtained.

[0102] By referring to such correlations, it is possible to quantitatively select output data suitable for input data, not only when the spectral image is identical to or similar to the input data, but also when the spectral image is dissimilar.

[0103] Next, in step S160, the control unit 18 controls various equipment based on the moisture information evaluated in step S150.

[0104] Next, the evaluation result and the control result are output by the output unit 15. The output unit 15 outputs the evaluation result and the control result to the display unit 109 or the like. In addition, the output unit 15 may output the evaluation result and the control result to the user terminal 2 via the public communication network 4.

[0105] This ends the operation of the soil evaluation device 1 in this embodiment. This makes it possible to continuously evaluate the moisture information of soil 7, which is not uniform in height and is transported on a conveyor belt or other conveying unit 8, for example, in the construction of a shield tunnel. In addition, since it is possible to automatically evaluate the moisture information, it is possible to reduce labor costs. In addition, since it is possible to evaluate the moisture information without stopping the conveyor belt or other conveying unit 8, work efficiency is improved. In addition, since it is possible to evaluate the moisture information without affecting the height of the soil 7, a shaping plate or the like for aligning the height of the soil to a constant level is not required, so it is possible to reduce costs for maintenance of the shaping plate, etc. In addition, the soil evaluation system may evaluate the moisture information of any soil 7, not limited to the soil 7 discharged in the shield construction of a tunnel.

[0106] Next, the results of an experiment using the soil evaluation system 100 according to the present invention will be described with reference to Fig. 11. Fig. 11 is a graph showing the measured moisture content against the actual moisture content in this embodiment. As shown in Fig. 11, almost no difference between the actual moisture content and the measured moisture content can be confirmed. This shows that the soil evaluation system 100 according to the present invention can evaluate the moisture information of the discharged soil 7 with high accuracy.

[0107] Although the embodiment of the present invention has been described, this embodiment is presented as an example and is not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, and are included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0108] 1: Sediment evaluation device 2: User terminal 3: Server 4:Public communication network 5: Imaging device 6: Lighting equipment 7: Soil 8: Transport section 10: Housing 11: Acquisition part 12: Image processing section 13: Evaluation section 14: Storage section 15: Output section 16: Conversion section 17:Extraction part 18: Control section 71 : Range 100: Sediment evaluation system 101: CPU 102: ROM 103: RAM 104: Preservation Department 105: Interface 106: Interface 107: Interface 108: Input section 109: Display section 110: Internal bus

Claims

1. An acquisition means for acquiring an image including soil and sand; a spectrum acquisition means for acquiring a spectrum image based on the image acquired by the acquisition means; and an evaluation means for evaluating moisture information related to the moisture contained in the soil and sand based on the spectral image acquired by the spectrum acquisition means. A soil and sand evaluation system characterized by:

2. The evaluation means, based on the spectral image acquired by the spectrum acquisition means, sets a specific wavelength included in a predetermined wavelength range as a normalized wavelength, selects a specific wavelength included in the predetermined wavelength range as an evaluation wavelength, and evaluates the moisture information based on a difference in spectral intensity between the normalized wavelength and the evaluation wavelength. The soil evaluation system according to claim 1 .

3. The evaluation means selects a specific wavelength included in a wavelength band of 450 nm to 1700 nm as a first normalized wavelength, and a specific wavelength included in a wavelength band of 900 nm to 1650 nm as a first evaluation wavelength. The soil evaluation system according to claim 2 .

4. the evaluation means uses a data set consisting of a reference spectral image and moisture information acquired in advance as learning data, and evaluates the moisture information based on the spectral image acquired by the spectrum acquisition means using an evaluation model generated by machine learning with the reference spectral image as an input and the moisture information as an output. The soil evaluation system according to claim 1 .

5. The spectrum acquisition means extracts wavelengths whose difference from a reference wavelength indicating soil and sand, which is previously acquired, is equal to or smaller than a threshold value from the acquired spectrum image, and acquires the spectrum image indicating the extracted wavelengths. The soil evaluation system according to claim 1 .

6. The spectrum acquisition means uses a data set consisting of a reference spectrum image and a spectrum image representing sediment, which have been acquired in advance, as learning data, and acquires the spectrum image representing the sediment based on the acquired spectrum image by using an extraction model generated by machine learning with the reference spectrum image as an input and the spectrum image representing the sediment as an output. The soil evaluation system according to claim 1 .

7. The acquisition means further acquires a light source image including a white board; The spectrum acquisition means acquires a spectral image based on a relative reflectance of the image with respect to the light source image acquired by the acquisition means. The soil evaluation system according to claim 1 .

8. The acquiring means acquires imaging condition information indicating imaging conditions when the light source image was captured, and corrects the image in accordance with the acquired imaging condition information. The soil evaluation system according to claim 7 .

9. The spectrum acquisition means removes, from the acquired spectral image, a spectrum indicating a wavelength whose difference from a reference wavelength indicating a transport unit transporting the soil and sand is equal to or less than a threshold value, which is acquired in advance. The soil evaluation system according to claim 1 .

10. The acquiring means acquires sediment information indicating an amount of sediment, The spectrum acquisition means acquires the spectrum image based on the image acquired by the acquisition means and the sediment information. The soil evaluation system according to claim 1 .

11. The acquiring means acquires the image using an imaging device capable of imaging pixels on a straight line. The soil evaluation system according to claim 1 .

12. An acquisition step of acquiring an image including sediment; a spectrum acquisition step of acquiring a spectrum image based on the image acquired by the acquisition step; and an evaluation step of evaluating moisture information related to moisture contained in the soil and sand based on the spectral image acquired by the spectrum acquisition step. A method for evaluating soil and sand, comprising:

Images