Turbidity determination system

Abstract

To provide a turbidity determination system capable of stably determining turbidity degree compared to prior art without permanently installing specific equipment underwater.SOLUTION: A turbidity determination system is a system for determining a turbidity degree of a water region based on a measuring image captured of the water region where water flows or is stored, and includes: information entropy calculation means; and turbidity level setting means. The turbidity level setting means sets the turbidity level of the water region by comparing the information entropy with the information entropy reference value. The turbidity degree in the water region can be determined according to the turbidity level.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a technology for determining the turbidity of water in a water area such as a river or pond, and more specifically to a turbidity determination system that can determine the degree of turbidity of a water area based on the pixel values ​​contained in an image taken of the water area. [Background technology]

[0002] River water, the primary source of water for water purification plants, contains clay components derived from mountain sediment, insoluble particles of metals and organic matter, and microorganisms such as plankton, all of which contribute to its turbidity. Meanwhile, when rivers flood due to heavy rain, their turbidity also increases significantly. For example, according to the Tokyo Metropolitan Government, the turbidity of the Tone River and Arakawa River is around 4°C on clear days. However, when Typhoon Hagibis struck the Kanto region in October 2019, the Kanamachi Water Purification Plant, which draws water from the Edogawa River, recorded a maximum temperature of 1,200°C. Here, "turbidity" is an index of the degree of cloudiness of water. For example, when 1 milligram of a turbidity standard substance is contained in 1 liter of water, the turbidity is considered to be 3°C.

[0003] Furthermore, large amounts of water are used in tunnel construction and other projects, and after use, the water is treated before being discharged into rivers, etc. Specifically, treatment equipment for measuring hydrogen ion concentration (pH), suspended solids (SS), naturally occurring heavy metals, etc. is installed at the construction site, and after performing treatment using these facilities to confirm that the turbid water meets the specified management standards, the treated water is discharged into rivers, etc.

[0004] As mentioned above, rivers are used as water sources for water purification plants, or as receptacles for treated water generated by tunnel construction and other projects. It is extremely useful to constantly monitor the degree of turbidity (e.g., turbidity) of river water. For example, when using river water as a water source for a water purification plant, the treatment intensity can be adjusted according to the turbidity, and when receiving treated water from construction work, the contractor can be required to perform appropriate treatment according to the turbidity.

[0005] Conventionally, the main method for determining the degree of turbidity of water in rivers and the like has been to use a turbidity meter such as that disclosed in Patent Document 1. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2003-46991 Summary of the Invention [Problem to be solved by the invention]

[0007] As described in Patent Document 1, turbidity meters can be used in a variety of ways, including a transmitted light method, which measures turbidity by detecting only light transmitted through a sample liquid; a scattered light method, which measures turbidity by detecting only light scattered by particles in the sample liquid; and a transmitted / scattered light method, which measures turbidity by detecting both transmitted and scattered light. However, regardless of the type of turbidity meter used, the turbidity meter must be installed in the river to measure the turbidity of river water. Therefore, there is a risk that the turbidity meter may be washed away, especially during heavy rain, and many people have hesitated to install turbidity meters permanently. Furthermore, it has been pointed out that the turbidity measured by a turbidity meter is merely information from the point where the turbidity meter is installed, and is insufficient to measure the turbidity of the entire river.

[0008] On the other hand, efforts are also being made to use images instead of turbidity meters to determine the level of water turbidity. Images are acquired by photographing an area of ​​water, and the level of turbidity is determined based on the pixel values ​​(RGB values, HSV values, etc.) of each pixel that makes up the image, i.e., hue, saturation, and brightness. However, because the images acquired vary depending on various factors such as the weather at the time of shooting, the surrounding environment that creates shadows, and shooting conditions such as shutter speed and aperture, it has been extremely difficult to stably and quantitatively evaluate the level of water turbidity.

[0009] The object of the present invention is to solve the conventional problems, that is, to provide a turbidity determination system that can determine the degree of turbidity more stably than conventional technology without permanently installing specified equipment in the water. [Means for solving the problem]

[0010] The present invention focuses on the fact that the greater the number of different pixel values ​​contained in an image, the more transparent the water, and the smaller the number of different pixel values, the more cloudy the water, making it possible to stably determine the degree of cloudiness of water, and is an invention based on an idea that has not been seen before.

[0011] The turbidity determination system of the present invention determines the degree of turbidity of a water area based on a measurement image of the water area where water flows or collects, and includes an information entropy calculation means and a turbidity level setting means. The information entropy calculation means calculates the information entropy of the measurement image based on pixel values ​​of the pixels constituting the measurement image. The turbidity level setting means sets a turbidity level representing the degree of turbidity of the water area by comparing the information entropy calculated by the information entropy calculation means with one or more information entropy reference values. The pixel values ​​of the pixels constituting the measurement image are grayscale values, and the information entropy is calculated by dividing the entropy of the measurement image by the maximum entropy (the maximum entropy obtained from the measurement image). The turbidity level is set using the information entropy reference value as a boundary, and is preset so that the closer the information entropy is to 1, the clearer the water is, and the closer the information entropy is to 0, the more turbid the water is. The degree of turbidity of the water area can then be determined based on the turbidity level.

[0012] The turbidity determination system of the present invention may further include a KL information calculation means. This KL information calculation means calculates the KL information based on the pixel values ​​of pixels constituting the reference image and the pixel values ​​of the measured image. In this case, the turbidity level setting means sets the turbidity level by comparing the KL information calculated by the KL information calculation means with one or more KL information reference values, and determines the turbidity level when the turbidity level based on the information entropy matches the turbidity level based on the KL information. Note that the reference image is an image captured when the water area is clear (or turbid), and the turbidity level for the KL information is set in advance using the KL information reference value as a boundary.

[0013] The turbidity determination system of the present invention can also be configured to set the turbidity level based only on the KL information amount. In this case, the turbidity level setting means sets the turbidity level by comparing the KL information amount calculated by the KL information amount calculation means with one or more KL information amount reference values.

[0014] The turbidity determination system of the present invention may further include a JS information amount calculation means. This JS information amount calculation means calculates the JS information amount based on the pixel values ​​of pixels constituting the reference image and the pixel values ​​of the measurement image. In this case, the turbidity level setting means sets the turbidity level by comparing the JS information amount calculated by the JS information amount calculation means with one or more KL information amount reference values. The turbidity level is set in advance using the JS information amount reference values ​​as a boundary.

[0015] The turbidity determination system of the present invention may further include a mutual information calculation means. This mutual information calculation means calculates mutual information based on pixel values ​​of pixels constituting the reference image and pixel values ​​of the measurement image. In this case, the turbidity level setting means sets the turbidity level by comparing the JS information calculated by the mutual information calculation means with one or more KL information reference values. The turbidity level is set in advance using the mutual information reference value as a boundary. [Effects of the Invention]

[0016] The turbidity determination system of the present invention has the following effects. (1) The degree of turbidity in the water area can be determined by using a camera or video camera installed outside the water area. This eliminates the need to permanently install measuring equipment such as a turbidity meter in the water, which means that the risk of the measuring equipment being washed away can be avoided. (2) The degree of turbidity in the water area is evaluated simply by the number of pixel values ​​used, without relying on the color information of the image. As a result, the degree of turbidity can be reliably determined even if the image is affected by various factors such as the weather at the time of shooting. (3) Since the water area is judged based on the pixel values ​​of the entire image, unlike point information such as that of a turbidity meter, it is possible to judge the degree of turbidity of the entire river, for example. [Brief explanation of the drawings]

[0017] [Figure 1] (a) An image showing an on-site image obtained by photographing a river with nearly clear water, and (b) an on-site image obtained by photographing a highly turbid river. [Figure 2] 1 is a block diagram showing the main configuration of a turbidity determination system according to a first embodiment of the present invention. [Figure 3] A mathematical diagram to explain "entropy" and "information entropy" based on measured pixel values. [Figure 4](a) is a frequency distribution diagram for the case where the number of pixels for each grayscale is different, and (b) is a frequency distribution diagram for the case where all grayscales have the same number of pixels. [Figure 5] FIG. 2 is a flowchart showing the main processing flow of the turbidity determination system according to the first embodiment. [Figure 6] FIG. 10 is a block diagram showing the main configuration of a turbidity determination system according to a second embodiment of the present invention. [Figure 7] A mathematical diagram to explain "KL divergence," "JS divergence," and "mutual information." [Figure 8] FIG. 10 is a flowchart showing the main processing flow in the case where the turbidity level is determined based on the KL information amount in the turbidity determination system according to the second embodiment. [Figure 9] FIG. 10 is a flowchart showing the main processing flow in the case where the turbidity level is determined based on the JS information amount in the turbidity determination system according to the second embodiment. [Figure 10] FIG. 10 is a flowchart showing the main processing flow in the case where the degree of turbidity is determined based on mutual information in the turbidity determination system according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0018] An example of an embodiment of the turbidity determination system of the present invention will be described with reference to the drawings. The turbidity determination system of the present invention can determine the degree of turbidity (hereinafter, for convenience, referred to as "degree of turbidity") of water in areas where water flows, such as rivers and streams, or areas where water is stored, such as dams, lakes, and pools (hereinafter, these will be collectively referred to as "water areas"). For convenience, the following description will be given using an example in which the water area is a "river."

[0019] The turbidity determination system of the present invention determines the degree of turbidity of a river using images (hereinafter referred to as "on-site images PS") that capture the state of the river (water area). A means for acquiring on-site images is preferably one that can automatically capture images periodically, such as a digital camera or digital video camera. Furthermore, it is preferable to acquire on-site images using a so-called fixed-point camera so that the angle of view does not change even when the images are taken repeatedly.

[0020] Figure 1 shows on-site images PS obtained by photographing a river, where (a) is an on-site image PS in a state where the water is nearly clear (i.e., low turbidity), and (b) is an on-site image PS in a state where the water is highly turbid (i.e., high turbidity). The on-site image PS shown in Figure 1(a) shows the riverbed because the water is clear, but the on-site image PS shown in Figure 1(b) shows the riverbed because the water is turbid. In this way, it can be said that on-site images PS taken of rivers with low turbidity contain more information than on-site images PS taken of rivers with high turbidity. In other words, on-site images PS with low turbidity use a wide variety of pixel values, while on-site images PS with high turbidity use fewer pixel values.

[0021] The present invention determines the turbidity of the water area by focusing on the number of types of pixel values ​​used, that is, it focuses on the fact that a river in a local image PS with a large number of types of pixel values ​​can be determined to have a low turbidity, and a river in a local image PS with a small number of types of pixel values ​​can be determined to have a high turbidity. Note that pixel values ​​here refer to values ​​possessed by each pixel that makes up the local image PS, and are values ​​that model color and shade so that they can be handled by a computer (electronic calculator), and examples include grayscale, RGB values, HSV values, CMYK values, and NCS values.

[0022] The present invention can be broadly divided into two embodiments: one in which the turbidity of a water area is determined using only the on-site image PS (hereinafter referred to as "Embodiment 1"), and one in which the turbidity of a water area is determined using the on-site image PS and a "reference on-site image" (hereinafter referred to as "Embodiment 2"). Here, the reference on-site image refers to an on-site image PS obtained by photographing an extremely clear water area, or an on-site image PS obtained by photographing an extremely turbid water area. The turbidity determination system 100 of the present invention will be described below, divided into Embodiments 1 and 2.

[0023] 1. Embodiment 1 2 is a block diagram showing the main components of the turbidity determination system 100 of the present invention in embodiment 1. As shown in this figure, the turbidity determination system 100 in embodiment 1 is configured to include information entropy calculation means 101 and turbidity level setting means 102, and can also be configured to include measurement image extraction means 106, on-site image storage means 107, reference value storage means 108, etc.

[0024] The information entropy calculation means 101, turbidity level setting means 102, and measurement image extraction means 106 that constitute the turbidity determination system 100 in embodiment 1 can be manufactured as dedicated devices, or a general-purpose computer device can be used. That is, the processing of the various means is performed by having the computer device execute arithmetic processing using a predetermined program. This computer device is equipped with a processor such as a CPU (Central Processing Unit) or GPU (Graphics Processing Unit), memories such as ROM and RAM, and some also include input means such as a mouse and keyboard, and a display, and can be configured, for example, as a personal computer (PC) or a server.

[0025] The on-site image storage means 107 and the reference value storage means 108 can be configured as a storage device of a general-purpose computer (for example, a personal computer) or can be configured as a database server. When configured as a database server, the database server can be placed on a local network (LAN: Local Area Network) or can be configured as a cloud server that stores data via the Internet.

[0026] Hereinafter, each of the main elements constituting the turbidity determination system 100 in the first embodiment will be described in detail.

[0027] (Measurement image extraction means) The measurement image extraction means 106 reads the on-site image PS from the on-site image storage means 107 (FIG. 2), which stores the on-site image PS, and generates an image (hereinafter referred to as a "measurement image") by extracting (cutting out) only the water area from the on-site image PS. For example, as shown in FIG. 1, the on-site image PS may contain not only the water area but also bank blocks and herbaceous plants. If subsequent processing is performed using the on-site image PS including bank blocks, it may be difficult to properly determine the degree of turbidity in the water area. Therefore, it is recommended that the measurement image extraction means 106 generate a measurement image and use that measurement image for subsequent processing. However, if the on-site image PS captures only the water area, the on-site image PS can be used as the measurement image for subsequent processing without processing the on-site image PS. Alternatively, the turbidity determination system 100 may not include the measurement image extraction means 106 at all, and the on-site image PS can be used as the measurement image for subsequent processing.

[0028] (Method for calculating information entropy) The information entropy calculation means 101 is a means for calculating the "information entropy" of a measurement image based on pixel values ​​(hereinafter, particularly referred to as "measured pixel values") possessed by pixels that constitute the measurement image. Note that the measured pixel values ​​in the first embodiment can be grayscale values. The procedure by which the information entropy calculation means 101 calculates the information entropy will be described below.

[0029] First, the information entropy calculation means 101 calculates "entropy (also called Shannon entropy)" using Equation 1 in Fig. 3. This entropy is also called "average information content" and is a value obtained as the sum of the expected values ​​of individual information contents (self-information contents). Note that the base of the logarithm in Equation 1 in Fig. 3 can be 2 (bit), or any other value such as 10 (digi) or e(nat). In the present invention, the entropy is calculated using the probability distribution (p(x) in Equation 1) of the occurrence of the grayscale (measured pixel value) in the measured image as an explanatory variable.

[0030] FIG. 4 is a graph showing the number of pixels for each grayscale, more specifically, a frequency distribution diagram (bar graph) with the horizontal axis representing the grayscale and the vertical axis representing the number of pixels (frequency). For convenience, only the range of 11 to 30 out of the entire grayscale range (0 to 255) is shown here. In the example of FIG. 4, there are a total of 240 pixels constituting the measured image (grayscales ranging from 11 to 30). FIG. 4(a) shows an example in which the number of pixels differs for each grayscale, while FIG. 4(b) shows an example in which all grayscales have the same number of pixels (i.e., 12 each).

[0031] For example, in the case of Figure 4(a), there are 16 pixels with a grayscale of 11, with a probability of 1 / 15 (16 / 240), there are 24 pixels with a grayscale of 22, with a probability of 1 / 10 (24 / 240), and there are 0 pixels with a grayscale of 25, with a probability of 0. Once the probability distribution for each grayscale is obtained in this way, the entropy is calculated using <Equation 1> in Figure 3 as described above, and once this entropy is obtained, the information entropy is calculated using <Equation 2> in Figure 3.

[0032] It is known that under certain conditions, the entropy calculated by Equation 1 in FIG. 3 will have a maximum value when all probabilities are the same. For example, under the conditions of FIG. 4 (i.e., grayscales ranging from 11 to 30 and a total pixel count of 240), the largest entropy (hereinafter referred to as "maximum entropy") is calculated when the number of pixels in all grayscales is 12 (probability 1 / 20) as shown in FIG. 4(b). As shown in Equation 2 in FIG. 3, the "information entropy" is calculated by dividing the entropy calculated based on the measured image by the maximum entropy. Therefore, the information entropy calculation means 101 calculates the entropy based on, for example, FIG. 4(a) and the maximum entropy based on FIG. 4(b), and then calculates the information entropy by dividing the entropy by the maximum entropy.

[0033] Assuming that both Figures 4(a) and 4(b) represent measurement images, the measurement image of Figure 4(a) uses fewer types of pixel values ​​(grayscale) (14 types in the figure), while the measurement image of Figure 4(b) uses more types of pixel values ​​(grayscale) (20 types in the figure). The information entropy for the measurement image of Figure 4(a) is calculated as a small value, while the information entropy for the measurement image of Figure 4(b) is calculated as the maximum value (=1.0). As described above, the present invention determines that a river in a measurement image with a large number of types of pixel values ​​has a low level of turbidity, and a river in a measurement image with a small number of types of pixel values ​​has a high level of turbidity. Therefore, the closer the information entropy is to 1, the lower the turbidity (higher transparency) of the river in the measurement image, and the closer the information entropy is to 0, the higher the turbidity (lower transparency) of the river in the measurement image.

[0034] (Means for setting turbidity level) The turbidity level setting means 102 is a means for setting a "turbidity level" that indicates the degree of turbidity of the water area by comparing the information entropy calculated by the information entropy calculation means 101 with an "information entropy reference value." Here, the information entropy reference value is a preset value of 1 or 2 or more, which divides the range of information entropy possible values ​​(0 to 1.0) into two or more ranges. Each range divided by the information entropy reference value is the turbidity level. As mentioned above, the turbidity level should be set so that the closer the information entropy is to 1, the lower the turbidity (higher transparency) of the river in the measurement image, and the closer the information entropy is to 0, the higher the turbidity (lower transparency) of the river in the measurement image. For example, if one "information entropy reference value = 0.5" is preset, two turbidity levels are set, such that the range where information entropy is greater than 0 and less than 0.5 can be set as the "turbid water level," and the range where information entropy is greater than 0.5 and less than 1.0 can be set as the "clear water level." Alternatively, if "information entropy reference value = 0.3" and "information entropy reference value = 0.7" are preset, three turbidity levels are set, such that the range where information entropy is greater than 0 and less than 0.3 can be set as the "strongly turbid water level," the range where information entropy is greater than 0.3 and less than 0.7 can be set as the "turbid water level," and the range where information entropy is greater than 0.7 and less than 1.0 can be set as the "clear water level."

[0035] When the information entropy is calculated by the information entropy calculation means 101, the turbidity level setting means 102 reads the information entropy reference value from the reference value storage means 108 (Fig. 2) that stores the information entropy reference value, compares the information entropy reference value with the calculated information entropy, determines the range that includes the information entropy, and sets the turbidity level corresponding to that range.

[0036] (Processing flow) The main processing of the turbidity determination system 100 will be described in detail below with reference to Fig. 5. Fig. 5 is a flow chart showing the flow of the main processing of the turbidity determination system 100 in embodiment 1, with the central column showing the processing to be performed, the left column showing what is necessary for that processing, and the right column showing what results from that processing.

[0037] To determine the degree of turbidity of a river (water area) using the turbidity determination system 100, first, a measurement image is generated by the measurement image extraction means 106 as shown in Fig. 5 (Step 201 in Fig. 5). Specifically, the on-site image PS is read from the on-site image storage means 107, and only the water area range is extracted from the on-site image PS to generate a measurement image.

[0038] Once the measurement image is generated, the information entropy calculation means 101 calculates the entropy of the measurement image (Step 202 in FIG. 5). Specifically, the entropy is calculated using the measured pixel values ​​constituting the measurement image and Equation 1 in FIG. 3. Then, the information entropy calculation means 101 calculates the information entropy by dividing the entropy related to the measurement image by the maximum entropy (Step 203 in FIG. 5).

[0039] Once the information entropy is calculated, the turbidity level setting means 102 sets the turbidity level (Step 204 in Fig. 5). Specifically, the information entropy reference value is read from the reference value storage means 108 (Fig. 2), and the information entropy is compared with the information entropy reference value to set the turbidity level including the information entropy.

[0040] 2. Embodiment 2 6 is a block diagram showing the main components of a turbidity determination system 100 of the present invention in embodiment 2. As shown in this figure, the turbidity determination system 100 in embodiment 2 is configured to include a turbidity level setting means 102 and a KL information calculation means 103, and can also be configured to include a measurement image extraction means 106, an on-site image storage means 107, a reference value storage means 108, a reference image storage means 109 that stores a reference on-site image, etc. Alternatively, the KL information calculation means 103 can be replaced by a JS information calculation means 104, or the KL information calculation means 103 can be replaced by a mutual information calculation means 105.

[0041] The turbidity level setting means 102, the KL information calculation means 103 (or the JS information calculation means 104 or the mutual information calculation means 105), and the measurement image extraction means 106 that constitute the turbidity determination system 100 in embodiment 2 can be manufactured as dedicated devices, or a general-purpose computer device can be used. That is, the processing of the various means is performed by having the computer device execute arithmetic processing according to a predetermined program. This computer device is equipped with a processor such as a CPU or GPU, memories such as ROM and RAM, and some also include input means such as a mouse and keyboard, and a display, and can be configured, for example, as a personal computer or server.

[0042] Furthermore, the local image storage means 107, the reference value storage means 108, and the reference image storage means 109 can be configured using a storage device of a general-purpose computer or can be configured in a database server. When configured in a database server, they can be placed on a local network or can be a cloud server that stores data via the Internet.

[0043] (KL information amount calculation means) The KL information calculation means 103 calculates the "KL information (Kullback-Leibler information)" of the measurement image based on the "measurement pixel values" of the pixels constituting the measurement image and the pixel values ​​of the pixels constituting the reference image (hereinafter, specifically referred to as "reference pixel values"). Here, the reference image refers to an image in which only the water area is extracted by the measurement image extraction means 106 from a reference on-site image (a on-site image PS captured of a water area in an extremely clear or extremely murky state). Of course, as in the first embodiment, the on-site image PS can be used as the measurement image for subsequent processing. Furthermore, the measurement pixel values ​​in the second embodiment can be grayscale values, or values ​​based on RGB values, HSV values, or the like. For convenience, however, the following description will be given using an example in which the measurement pixel values ​​are grayscale.

[0044] The KL divergence calculation means 103 calculates the "KL divergence" using Equation 3 in FIG. 7. This KL divergence is, so to speak, the sum of the differences in the entropy of two images, and is an index that indicates "how similar the entropy of the two images is." In other words, the KL divergence in the present invention can be said to be a value that indicates how similar the measurement image and the reference image are. To calculate the KL divergence, a probability distribution (Q(x) in Equation 3) for each grayscale (reference pixel value) of the reference image is calculated in advance as described in the first embodiment. Then, a probability distribution (P(x) in Equation 3) for each grayscale (measurement pixel value) of the measurement image is calculated, and the KL divergence is calculated using Equation 3 in FIG. 7 using the probability distribution of the reference image and the probability distribution of the measurement image.

[0045] (Means for setting turbidity level) The turbidity level setting means 102 is a means for setting a "turbidity level" that indicates the degree of turbidity of the water region by comparing the KL information calculated by the KL information calculation means 103 with a "KL information reference value." Here, the KL information reference value is a preset value of 1 or 2 or more, which divides the range that the KL information can take into two or more ranges. Each range divided by the KL information reference value is a turbidity level.

[0046] The closer the KL information value is to 0, the more similar the two pieces of information are; in other words, the closer the KL information value is to 0 in the present invention, the more similar the measured image and the reference image are. Therefore, when the reference on-site image is of a river in an extremely clear state, the turbidity level should be set so that the closer the KL information value is to 0 (i.e., the smaller the value), the less turbid (more transparent) the river in the measured image will be, and the farther the KL information value is from 0 (i.e., the larger the value), the more turbid (less transparent) the river in the measured image will be. On the other hand, when the reference on-site image is of a river in an extremely turbid state, the turbidity level should be set so that the closer the KL information value is to 0, the more turbid (less transparent) the river in the measured image will be, and the farther the KL information value is from 0, the more turbid (less transparent) the river in the measured image will be.

[0047] When the KL information amount is calculated by the KL information amount calculation means 103, the turbidity level setting means 102 reads the KL information amount reference value from the reference value storage means 108 (FIG. 6) that stores the KL information amount reference value, compares the KL information amount with the KL information amount, determines the range that includes the KL information amount, and sets the turbidity level corresponding to that range.

[0048] (Processing flow) The main processing of the turbidity determination system 100 will be described in detail below with reference to Fig. 8. Fig. 8 is a flow diagram showing the main processing flow of the turbidity determination system in embodiment 2 when determining the degree of turbidity based on the KL information amount, with the central column showing the processing to be executed, the left column showing what is necessary for that processing, and the right column showing what results from that processing.

[0049] In this case, when determining the degree of turbidity of a river (water area) using the turbidity determination system 100, a measurement image is first generated using the measurement image extraction means 106 as shown in Fig. 8 (Step 301 in Fig. 8). Specifically, the on-site image PS is read from the on-site image storage means 107, and only the water area range is extracted from the on-site image PS to generate a measurement image.

[0050] Once the measurement image is generated, the KL divergence of the measurement image is calculated by the KL divergence calculation means 103 (Step 302 in FIG. 8). Specifically, the probability distribution for each grayscale of the reference image is calculated, and the probability distribution for each grayscale of the measurement image is also calculated, and the KL divergence is calculated by Equation 3 in FIG. 7 using the probability of the reference image and the probability of the measurement image.

[0051] Once the KL information amount is calculated, the turbidity level setting means 102 sets the turbidity level (Step 303 in Fig. 8). Specifically, the KL information amount is read from the reference value storage means 108 (Fig. 6), and the KL information amount is compared with the KL information amount reference value to set the turbidity level including the KL information amount.

[0052] As described above, the turbidity determination system 100 in the second embodiment can also be configured to include the JS information amount calculation means 104 instead of the KL information amount calculation means 103. The turbidity determination system 100 including the JS information amount calculation means 104 will be described below.

[0053] (JS information amount calculation method) The JS information calculation means 104 calculates the "JS information (Jensen-Shannon information)" of the measurement image based on the "measured pixel value" of the measurement image and the "reference pixel value" of the reference image, and calculates the "JS information" using <Equation 4> in FIG. 7. The KL information cannot be treated as the distance between two pieces of information (probability distributions) because the two pieces of information are asymmetric. Therefore, the JS information is an improvement of the KL information so that the two pieces of information can be treated as the distance between the two pieces of information symmetrically. Therefore, like the KL information, the JS information can be said to be a value that indicates the degree of similarity between the measurement image and the reference image. To calculate the JS information, a probability distribution for each grayscale (reference pixel value) of the reference image is calculated in advance as described in the first embodiment. Then, a probability distribution for each grayscale (measured pixel value) of the measurement image (P(x) in <Equation 4>) is calculated, and an average probability distribution (M(x) in <Equation 4>), which is the average of the probability distribution of the reference image and the probability distribution of the measurement image, is calculated. Then, the JS information amount is calculated using the probability distribution of the measured image and the average probability distribution according to Equation 4 in Figure 7.

[0054] (Means for setting turbidity level) In this case, the turbidity level setting means 102 is a means for setting a "turbidity level" that indicates the degree of turbidity in the water area by comparing the JS information calculated by the JS information calculation means 104 with a "JS information reference value." Here, the JS information reference value is a preset value of 1 or 2 or more, which divides the range that the JS information can take into two or more ranges. Each range divided by the JS information reference value is the turbidity level.

[0055] The closer the JS information value is to 0, the more similar the information between the two is; in other words, the closer the JS information value is to 0 in the present invention, the more similar the measured image and the reference image are. Therefore, when the reference on-site image is a photograph of an extremely clear river, the turbidity level should be set so that the closer the JS information value is to 0 (i.e., the smaller the value), the less turbid (more transparent) the river in the measured image will be, and the farther the JS information value is from 0 (i.e., the larger the value), the more turbid (less transparent) the river in the measured image will be. On the other hand, when the reference on-site image is a photograph of an extremely turbid river, the turbidity level should be set so that the closer the JS information value is to 0, the more turbid (less transparent) the river in the measured image will be, and the farther the JS information value is from 0, the more turbid (less transparent) the river in the measured image will be.

[0056] When the JS information amount is calculated by the JS information amount calculation means 104, the turbidity level setting means 102 reads the JS information amount reference value from the reference value storage means 108 (Fig. 6) that stores the JS information amount reference value, compares the JS information amount with the JS information amount, determines the range that includes the JS information amount, and sets the turbidity level corresponding to that range.

[0057] (Processing flow) The main processing of the turbidity determination system 100 will be described in detail below with reference to Fig. 9. Fig. 9 is a flow diagram showing the main processing flow of the turbidity determination system in embodiment 2 when determining the degree of turbidity based on the JS information amount, with the central column showing the processing to be executed, the left column showing what is necessary for that processing, and the right column showing what results from that processing.

[0058] In this case, when determining the degree of turbidity of a river (water area) using the turbidity determination system 100, a measurement image is first generated using the measurement image extraction means 106 as shown in Fig. 9 (Step 401 in Fig. 9). Specifically, the on-site image PS is read from the on-site image storage means 107, and only the water area range is extracted from the on-site image PS to generate a measurement image.

[0059] Once the measurement image is generated, the JS information amount calculation means 104 calculates the JS information amount of the measurement image (Step 402 in FIG. 9). Specifically, the probability distribution of the reference image is obtained, and the probability distribution (P(x) in <Equation 4>) for each grayscale (measured pixel value) related to the measurement image is also obtained, and then the average probability distribution (M(x) in <Equation 4>) which is the average of the probability distribution of the reference image and the probability distribution of the measurement image is obtained. Then, the JS information amount is calculated using <Equation 4> in FIG. 7 using the probability distribution of the measurement image and the average probability distribution.

[0060] Once the JS information amount is calculated, the turbidity level is set by the turbidity level setting means 102 (Step 403 in Fig. 9). Specifically, the JS information amount is read from the reference value storage means 108 (Fig. 6), and the JS information amount is compared with the JS information amount reference value to set the turbidity level including the JS information amount.

[0061] As described above, the turbidity determination system 100 in the second embodiment can also be configured to include a mutual information calculation means 105 instead of the KL information calculation means 103. The turbidity determination system 100 including the mutual information calculation means 105 will be described below.

[0062] (Mutual information calculation means) The mutual information calculation means 105 calculates the "mutual information" of the measurement image based on the "measured pixel value" of the measurement image and the "reference pixel value" of the reference image, and calculates the "mutual information" using <Equation 5> in FIG. 7. This mutual information is a value that represents the degree to which one piece of information (probability distribution) depends on the other, and is minimum (=0) when the two are independent (i.e., least dependent), and the greater the degree of dependence, the larger the value. Therefore, the mutual information in the present invention can be said to be a value that indicates how similar the measurement image and the reference image are. In other words, the larger the mutual information value, the closer the measurement image and the reference image are to being similar, while the closer the mutual information is to 0, the less similar the measurement image and the reference image are to being similar.

[0063] To calculate the mutual information, the probability distribution (P(y) in <Equation 5>) for each grayscale (reference pixel value) of the reference image is calculated in advance in the manner described in embodiment 1. Then, the probability distribution (P(x) in <Equation 5>) for each grayscale (measurement pixel value) of the measurement image is calculated, and the joint probability distribution (P(x,y) in <Equation 5>) for the reference pixel value and the measurement pixel value is calculated. Then, the probability distribution of the measurement image, the average probability distribution, and the joint probability distribution are used to calculate the mutual information according to <Equation 5> in FIG. 7.

[0064] (Means for setting turbidity level) In this case, the turbidity level setting means 102 is a means for setting a "turbidity level" that represents the degree of turbidity in the water region by comparing the mutual information calculated by the mutual information calculation means 105 with a "mutual information reference value." Here, the mutual information reference value is a preset value of 1 or more, which divides the range in which the mutual information can be taken into two or more ranges. Each range divided by the mutual information reference value is a turbidity level.

[0065] As described above, the larger the mutual information value, the more similar the measured image and the reference image are. Therefore, when the reference on-site image is an image of a river in an extremely clear state, the turbidity level should be set so that the larger the mutual information value, the lower the turbidity (higher transparency) of the river in the measured image, and the smaller the mutual information value, the higher the turbidity (lower transparency) of the river in the measured image. On the other hand, when the reference on-site image is an image of a river in an extremely turbid state, the turbidity level should be set so that the larger the mutual information value, the higher the turbidity (lower transparency) of the river in the measured image, and the smaller the mutual information value, the lower the turbidity (higher transparency) of the river in the measured image.

[0066] When the mutual information is calculated by the mutual information calculation means 105, the turbidity level setting means 102 reads the mutual information reference value from the reference value storage means 108 (FIG. 6) that stores the mutual information reference value, compares the mutual information with the mutual information, determines the range that includes the mutual information, and sets the turbidity level corresponding to that range.

[0067] (Processing flow) The main processing of the turbidity determination system 100 will be described in detail below with reference to Fig. 10. Fig. 10 is a flow diagram showing the main processing flow of the turbidity determination system in embodiment 2 when determining the degree of turbidity based on mutual information, with the central column showing the processing to be performed, the left column showing what is necessary for that processing, and the right column showing what results from that processing.

[0068] In this case, when determining the degree of turbidity of a river (water area) using the turbidity determination system 100, a measurement image is first generated using the measurement image extraction means 106 as shown in Fig. 10 (Step 501 in Fig. 10). Specifically, the on-site image PS is read from the on-site image storage means 107, and only the water area range is extracted from the on-site image PS to generate a measurement image.

[0069] Once the measurement image is generated, the mutual information calculation means 105 calculates the mutual information of the measurement image (Step 502 in FIG. 10). Specifically, the probability distribution (P(y) in <Equation 5>) for each grayscale (reference pixel value) of the reference image is calculated, and the probability distribution (P(x) in <Equation 5>) for each grayscale (measurement pixel value) of the measurement image is calculated, and further the joint probability distribution (P(x, y) in <Equation 5>) for the reference pixel value and the measurement pixel value is calculated. Then, using the probability distribution of the measurement image, the average probability distribution, and the joint probability distribution, the mutual information is calculated according to <Equation 5> in FIG. 7.

[0070] Once the mutual information is calculated, the turbidity level setting means 102 sets the turbidity level (Step 503 in Fig. 10). Specifically, the mutual information is read from the reference value storage means 108 (Fig. 6), and the mutual information is compared with the mutual information reference value to set the turbidity level including the mutual information.

[0071] The turbidity determination system 100 in the second embodiment may further include a KL information amount calculation means 103 and an information entropy calculation means 101. In this case, the turbidity level is set based on the KL information amount calculated by the KL information amount calculation means 103, and the turbidity level is set based on the information entropy calculated by the information entropy calculation means 101. Then, when the turbidity level based on the KL information amount and the turbidity level based on the information entropy match, the turbidity level can be determined. Note that when the turbidity level based on the KL information amount and the turbidity level based on the information entropy do not match, it is preferable to output, for example, "undeterminable" or output both turbidity levels. [Industrial Applicability]

[0072] The turbidity determination system of the present invention is particularly useful for river administrators, including national and local governments. According to the present invention, the degree of turbidity of river water can be stably determined, and as a result, when the water is used as a water source for a water purification plant, appropriate treatment can be carried out according to the degree of turbidity, which ultimately leads to the provision of high-quality drinking water to citizens. Considering this, the present invention can be expected to be not only industrially applicable but also to make a significant contribution to society. [Explanation of symbols]

[0073] 100 Turbidity determination system of the present invention 101 (Turbidity Determination System) Information Entropy Calculation Method 102 (Turbidity determination system) turbidity level setting means 103 KL information calculation method (for turbidity determination system) 104 JS information calculation method (for turbidity determination system) 105 Mutual information calculation means (for turbidity determination system) 106 (Turbidity determination system) measurement image extraction means 107 (Turbidity Determination System) On-site Image Storage Means 108 (Turbidity determination system) reference value storage means 109 (Turbidity Judgment System) Reference Image Storage Means PS Local images

Claims

1. A system for determining the degree of turbidity of a water area based on a measurement image of the water area where water flows or water is collected, an information entropy calculation means for calculating an information entropy of the measurement image based on pixel values ​​of pixels constituting the measurement image; and a turbidity level setting means for setting a turbidity level representing the degree of turbidity of the water region by comparing the information entropy calculated by the information entropy calculation means with one or more information entropy reference values, the pixel values ​​are in grayscale; the information entropy is a value obtained by dividing the entropy related to the measurement image by the maximum entropy obtained by the measurement image, The turbidity level is set with the information entropy reference value as a boundary, and is set in advance so that the closer the information entropy is to 1, the clearer the water is, and the closer the information entropy is to 0, the cloudier the water is. The degree of turbidity of the water area can be determined according to the turbidity level. A turbidity determination system characterized by:

2. a KL information amount calculation means for calculating a KL information amount based on the pixel values ​​of pixels constituting the reference image and the pixel values ​​of the measurement image, The reference image is an image captured when the water area is clear or cloudy, the turbidity level setting means sets the turbidity level by comparing the KL information calculated by the KL information calculation means with one or more KL information reference values; The turbidity level is set in advance using the KL information criterion value as a boundary, Furthermore, the turbidity level setting means determines the turbidity level when the turbidity level based on the information entropy and the turbidity level based on the KL information amount match.

2. The turbidity determination system according to claim 1.

3. A system for determining the degree of turbidity of a water area based on a measurement image of the water area where water flows or water is collected, a KL information calculation means for calculating a KL information based on a measurement pixel value of a pixel constituting the measurement image and a reference pixel value of a pixel constituting a reference image; a turbidity level setting means for setting a turbidity level representing the degree of turbidity of the water region by comparing the KL information calculated by the KL information calculation means with one or more KL information reference values, The reference image is an image captured when the water area is clear or cloudy, The turbidity level is set in advance using the KL information criterion value as a boundary, The degree of turbidity of the water area can be determined according to the turbidity level. A turbidity determination system characterized by:

4. A system for determining the degree of turbidity of a water area based on a measurement image of the water area where water flows or water is collected, a JS information amount calculation means for calculating a JS information amount based on a measurement pixel value of a pixel constituting the measurement image and a reference pixel value of a pixel constituting a reference image; a turbidity level setting means for setting a turbidity level representing the degree of turbidity of the water region by comparing the JS information calculated by the JS information calculation means with one or more JS information reference values, The reference image is an image captured when the water area is clear or cloudy, The turbidity level is set in advance using the JS information reference value as a boundary, The degree of turbidity of the water area can be determined according to the turbidity level. A turbidity determination system characterized by:

5. A system for determining the degree of turbidity of a water area based on a measurement image of the water area where water flows or water is collected, a mutual information calculation means for calculating mutual information based on measurement pixel values ​​of pixels constituting the measurement image and reference pixel values ​​of pixels constituting a reference image; a turbidity level setting means for setting a turbidity level representing the degree of turbidity of the water region by comparing the mutual information calculated by the mutual information calculation means with one or more mutual information reference values, The reference image is an image captured when the water area is clear or cloudy, The turbidity level is set in advance using the mutual information reference value as a boundary, The degree of turbidity of the water area can be determined according to the turbidity level. A turbidity determination system characterized by:

Images