KR20210053692A - Apparatus and method for analyzing microorganism - Google Patents

Abstract

The present invention relates to a sample water analysis device comprising: a chamber in which sample water is received; an imaging unit for imaging sample water in the chamber; and a control unit that analyzes images continuously captured through the imaging unit. The control unit is configured to: use model data generated through machine learning by receiving first tracking images generated by tracking movement paths of living things and inanimate objects, respectively; detect a moving object on the basis of a difference between two consecutive frames; generate a second tracking image by tracking the detected moving object; and compare the model data with the second tracking image to determine whether the moving object is a living organism or an inanimate object. Under a ship environment where vibration or shaking occurs, floating objects in sample water and living aquatic life can be distinguished more accurately.

Description

Sample number analysis apparatus and method {Apparatus and method for analyzing microorganism}

The present invention relates to a sample number analysis apparatus and method, and more particularly, to a sample number analysis apparatus and method capable of more accurately analyzing zooplankton in ballast water so as to be applicable to a ballast water treatment apparatus.

In general, ballast water or ballast water is a ballast tank installed on a ship to maintain balance when a ship is operated in a state where cargo is unloaded from the ship or when the amount of cargo loaded on the ship is very small. It refers to the seawater to fill in.

Since various aquatic organisms inhabit these ballast waters, there is a high concern that serious marine pollution and ecosystem destruction are caused if they are discharged from other areas without any treatment.

Accordingly, the International Maritime Organization (IMO) concluded an international agreement to install equipment necessary for sterilization and purification of ballast water on the ship.

The ballast water treatment system mounted on a ship must be operated after receiving a certificate after undergoing a land test and on-board test in accordance with the standards of the International Maritime Organization (IMO), so the ballast water treated by the ballast water treatment system has to be operated by the International Maritime Organization. There is a need for a system to monitor whether it meets the emission standards specified in the regulations.

However, when measuring live and death of aquatic organisms or analyzing zooplankton to determine whether the discharge standard of the ballast water treatment system is satisfied, it is determined only by the mobility of the object displayed on the screen of the image analysis device. In an environment in which vibration or rolling occurs frequently, it is difficult to distinguish between the movement of zooplankton and the movement of non-living objects (for example, floating objects), and thus there is a problem of inferior accuracy.

In addition, there is a problem that human error occurs because humans must directly count the number of live zooplankton.

Korean Registered Patent Publication No. 10-1531392

The present invention has been conceived to solve the above problems, and in particular, an object thereof is to provide a sample number analysis apparatus and method capable of more quickly and accurately analyzing zooplankton in ballast water.

In order to achieve the above object, the apparatus for analyzing the number of samples according to the aspect of the present invention comprises: a chamber in which the number of samples is accommodated; An imaging unit for photographing the number of samples in the chamber; And a control unit that analyzes the continuously photographed image through the imaging unit, wherein the control unit receives schematic first tracking information generated by tracking movement paths of living and non-living objects, respectively, and generated through machine learning. Using model data, detecting a moving object based on the difference between two consecutive frames, tracking the detected moving object to generate schematic second tracking information, model data and second tracking information Is compared to determine whether the moving object is a living or non-living object.

Here, the schematic first or second tracking information may be composed of a tracking image generated by connecting the positions of each frame with a line in succession.

As another embodiment, the schematic first or second tracking information may be configured as a table in which the position of each frame is displayed in two-dimensional coordinates.

The controller calculates the size of the moving object based on the detected number of horizontal and vertical pixels and the actual size per pixel, and generates second tracking information by tracking only the moving object having a predetermined size or more.

Here, the size of the moving object generating the second tracking information may be 10 μm or more and 50 μm or less, or 50 μm or more.

In an embodiment of the present invention, the first tracking information may be input by labeling a living thing or an inanimate object so that an object whose movement path is tracked is classified.

In addition, in the embodiment of the present invention, machine learning may generate model data using a convolutional neural network (CNN) method.

Meanwhile, a method for analyzing the number of samples according to another aspect of the present invention includes the steps of receiving schematic first tracking information generated by tracking movement paths of living organisms and inanimate objects, respectively, and generating model data through machine learning; Continuously photographing the number of samples in the chamber; Detecting a moving object based on the difference between two consecutive frames, tracking the detected moving object, and generating schematic second tracking information; It includes; comparing the model data and the second tracking information to determine whether the moving object is a living organism or inanimate object.

In an embodiment of the present invention, the first tracking information and the second tracking information may be measured for substantially the same time.

According to the present invention, it is possible to perform high-precision life and death judgment by making it possible to more accurately distinguish between floating objects in sample water and living aquatic organisms in a ship environment where vibration or shaking occurs.

In addition, according to the present invention, there is an effect of removing human errors by automatically determining live zooplankton using model data generated through machine learning.

In addition, it is configured to automatically count the size of the moving object determined as zooplankton based on 50㎛, so that it is possible to more quickly and accurately determine whether the ballast water treated by the ballast water treatment system satisfies the emission standard of IMO. There is an effect.

1 is a configuration diagram showing a sample number analysis apparatus according to an embodiment of the present invention,
2 is a flow chart showing a sample number analysis method according to an embodiment of the present invention,
3 is a diagram illustrating a process of detecting a moving object in a method for analyzing the number of samples according to an embodiment of the present invention,
4 shows an example of a first tracking image generated by tracking movement paths of living and non-living objects for generating model data to be used in the sample number analysis method according to an embodiment of the present invention,
5 is a diagram illustrating a process of generating a second tracking image by tracking a moving object detected in a method for analyzing the number of samples according to an embodiment of the present invention.
6 is a diagram illustrating a process of classifying zooplankton through a sample number analysis method according to an embodiment of the present invention,
7 shows a display of the sample number analysis apparatus according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to elements of each drawing, it should be noted that the same elements are assigned the same numerals as possible, even if they are indicated on different drawings. In addition, in describing the present invention, when it is determined that a detailed description of a related known configuration or function may obscure the subject matter of the present invention, a detailed description thereof will be omitted. In addition, a preferred embodiment of the present invention will be described below, but the technical idea of the present invention is not limited or limited thereto, and may be modified and variously implemented by a person skilled in the art.

1 is a block diagram showing a sample number analysis apparatus according to an embodiment of the present invention.

As shown in FIG. 1, the sample number analysis apparatus 100 according to an embodiment of the present invention includes a chamber 110 in which the number of samples containing aquatic organisms is accommodated, and an imaging unit for photographing the number of samples in the chamber 110. It includes 120 and a control unit 130 for processing the image captured by the imaging unit 120.

Here, the number of samples in the chamber 110 may be sampled in various fields requiring analysis of underwater organisms, and aquatic organisms included in the sample number may be analyzed.

For example, ballast water is treated in various ways, such as electrolysis or chemical injection during ballasting, and then introduced into the ballast tank and stored, then discharged out of the ship through discharge pipes during deballasting. However, since it is necessary to determine whether the discharged ballast water meets the discharge standards prescribed by the International Maritime Organization, the discharged ballast water is sampled and analyzed for the type of aquatic organisms present in the ballast water, life and death judgment, etc. Will do.

1 shows the sampling of the ballast water, and the sample water flows into the concentrator 20 through the sampling port 12 provided in the pipe 11 through which the ballast water is discharged, and a predetermined size (for example, horizontal and The number of samples containing organisms, floating objects, etc. of a predetermined size or more is filtered by the mesh unit 21 having eyes of 32 μm in the vertical direction, and in this case, the eye size in the diagonal direction is 50 μm). It is configured to flow into ).

Here, since organisms, floating objects, and the like may be trapped in the eyes of the mesh unit 21, it is configured to perform back flushing including the backwashing water supply unit 30 to prevent clogging due to these.

The number of samples that have passed through the mesh unit 21 and the number of samples that have been measured by flowing into the chamber are introduced into the collection unit 40 through respective discharge pipes, and the sample water from the collection unit 40 is again piped (11). Is transferred to.

In the embodiment of FIG. 1, the sampling of ballast water has been described as an example, but the present invention is not limited thereto and may be applied to other fields. As an example, purified water sterilized in a water purification plant may be sampled and analyzed through the sample number analyzer 100 of the present invention.

The chamber 110 is configured such that the number of samples to be analyzed is accommodated and held during the analysis time.

Here, the chamber 110 is configured in the form of a bowl that does not have an inlet (not shown) and an outlet (not shown), so that the experimenter may manually contain water and use it, but as shown in FIG. 1, the sampling port 12 ) May be provided with an inlet (not shown) and an outlet (not shown) so that the sample water concentrated by the concentrating device 20 connected to it is automatically introduced and discharged.

The imaging unit 120 may be installed in an upper or lower direction or a side direction of the chamber 110 to photograph aquatic organisms included in the number of samples of the chamber 110.

Here, the imaging unit 120 may include a driving unit (not shown) to move vertically or horizontally, and may be configured to adjust the separation distance from the chamber 110 according to the measurement environment.

In addition, a lamp (not shown) for generating a light source may be further installed on one side of the chamber 110 so that aquatic organisms can be better captured when the image pickup unit 120 captures an image.

On the other hand, the control unit 130 according to an embodiment of the present invention is configured to analyze the images captured through the imaging unit 120, in particular, by detecting a moving object within the number of samples, and tracking the detected moving object It is configured to generate a tracking image and compare it with model data generated through machine learning in advance to analyze whether a moving object is alive or dead. This process will be described in detail below.

2 is a flow chart showing a sample number analysis method according to an embodiment of the present invention. Referring to this, the sample number analysis method of the present invention is a schematic first tracking generated by tracking the movement paths of living and inanimate objects. Steps of receiving each information and generating model data through machine learning (S110), continuously photographing the number of samples accommodated in the chamber 110 (S120), and detecting a moving object within the number of samples (S130) And generating schematic second tracking information by tracking the detected moving object (S140), and determining whether the moving object is alive or dead by comparing the model data with the second tracking information (S150).

In the method for analyzing the number of samples according to an embodiment of the present invention, first, a plurality of first tracking information is generated, inputted as training data, and model data is generated through machine learning (S110). A more detailed process will be described in detail with reference to FIGS. 3 and 4.

Next, in order to analyze the number of samples, the number of samples is introduced into the chamber 110 and the number of samples is photographed by the imaging unit 120 (S120).

Thereafter, the controller 130 analyzes the image continuously photographed by the imaging unit 120 and detects the moving object in the image through the process of FIG. 3 to be described later (S130).

After the moving object is detected, a process of tracking the detected moving object is performed (S140).

Thereafter, the generated tracking image of the moving object is compared with the model data, and based on this, it is determined whether the moving object is alive or dead (S150).

3 is a diagram illustrating a process of detecting a moving object in the sample number analysis method according to an embodiment of the present invention. Referring to FIG. 3, images continuously photographed by the imaging unit 120 are obtained (a) , After extracting the image of two consecutive frames (b), it proceeds to the process of comparing.

When comparing two consecutive frames, a moving area is detected by the difference between the images of each frame (c). In one embodiment of the present invention, the imaging unit 120 continuously captures an image of at least one frame per second. Through this, more reliable image data can be secured and the analysis of aquatic organisms can be performed accurately.

Then, the moving object is extracted from the detected moving area (d).

4 illustrates an example of a first tracking image generated by tracking movement paths of living organisms and inanimate objects for generating model data to be used in the sample number analysis method according to an embodiment of the present invention.

Referring to FIG. 4, the controller 130 of the present invention generates model data through a process of receiving and learning a first tracking image, that is, training data input during machine learning. The model data may be generated by the control unit 130 of the present invention, but may be generated by a separate server or control device and configured to be used by the sample number analysis apparatus of the present invention.

The learning data used in an embodiment of the present invention uses a first tracking image 210 that classifies moving objects into living organisms (for example, zooplankton) and floating objects, respectively.

In an embodiment of the present invention, machine learning is a convolutional neural network, which is a neural network composed of one or several convolutional layers, a pooling layer, and a fully connected layer. , Hereinafter referred to as'CNN') method can be used to generate model data. The CNN method is a structure suitable for learning two-dimensional data, and it is possible to improve accuracy when machine learning the image of an aquatic creature (zooplankton) of the present invention.

The first tracking image 210 is generated by tracking a moving object every predetermined frame (for example, every frame) and connecting each position with a line. Make it possible to classify whether it is a floating object.

In the present invention, the "tracking image" is an embodiment of the tracking information schematized by tracking a moving object, and can be replaced in a variety of ways. As an example, the schematic tracking information may be configured as a diagram in which the position of each frame is displayed in 2D coordinates. In this case, a Cartesian coordinate system composed of (x,y) or a cylindrical coordinate system composed of (r,θ) may be used as the coordinates. In addition, these coordinates may be displayed as numbers, or may be displayed as graphs in which numbers are correspondingly expressed in pictures.

The first tracking image 210 may further include a label 220 to distinguish an image of a living organism (eg, zooplankton) and a floating object, respectively. In another embodiment, each image may be classified by a file name, or a storage space may be classified and classified.

In addition, the first tracking image 210 is generated by connecting the positions at multiple points of the moving object acquired for a predetermined time, and the predetermined time is set to a time sufficient to express the characteristics of the first tracking image 210 It should be. Through a number of experiments, the inventors have confirmed that it is desirable to configure the first tracking image 210 based on the position acquired for about 20 seconds when the image of the number of samples is captured at 60 frames per second.

In FIG. 4 of the present invention, three first tracking images 210 of zooplankton and floating matter were each learned, and each tracking image 210 was generated by tracking a moving object for approximately 20 seconds, but the present invention is limited thereto. It doesn't work. That is, more tracking images 210 may be used, and tracking time may also be more or less than 20 seconds, which enables the optimal number of necessary information and tracking time to be determined through accumulation of experiments and experiences. In addition, it is possible to determine the number of necessary information for additional organisms in addition to zooplankton.

The first tracking image 210 generated as described above is input to generate model data through machine learning, and the sample number analysis apparatus of the present invention uses this model data to determine the life or death of a moving object. Will be performed.

5 is a diagram illustrating a process of generating a second tracking image by tracking a moving object detected in a method for analyzing the number of samples according to an embodiment of the present invention.

5, in order to track the position of the detected moving object, an identifier (ID) is assigned to the detected moving object, and the difference between two consecutive frames is continuously calculated and based on this, The second tracking image is detected.

That is, the second tracking image of the moving object is generated by tracking and recording the position of the moving object detected in the previous frame and the position of the moving object detected in the current frame every frame.

In an embodiment of the present invention, the control unit 130 converts the detected moving object to the size of the moving object based on the number of horizontal and vertical pixels of the detected moving object and the actual size per pixel, and tracks only moving objects having a predetermined size or more to perform second tracking. You can also create an image. For example, by setting the size of the moving object generating the second tracking image to be 10 μm or more and 50 μm or less, or 50 μm or more, it is possible to check whether the result of ballast water sterilization, which is the number of samples, satisfies the IMO standard.

In an embodiment of the present invention, the first tracking image and the second tracking image may be measured for substantially the same time period, and thus may be configured to analyze and compare the tracking images under the same time condition.

Meanwhile, in an embodiment of the present invention, the controller 130 determines whether the moving object is a living zooplankton or a floating object by comparing the model data and the tracking image of the moving object, and whether the determined result is true or false is analyzed. It is judged based on the accuracy of the result. That is, when the accuracy of the determined result is greater than or equal to a predetermined threshold, it may be configured to recognize that the determined result is correct. In an exemplary embodiment of the present invention, when the accuracy of the results of a plurality of experiments is 50% or more, the probability that the judgment result is correct is very high, and therefore, it is recognized that the judgment result is correct when it is 50% or more.

The characteristics of zooplankton of 10 μm or more and 50 μm or 50 μm or more among living aquatic organisms are reflected in the tracking image because the actual movement path changes rapidly and the movement is irregular. Based on this characteristic information, it is possible to distinguish between dead or floating objects and living organisms (especially, zooplankton) to determine the life or death of a moving object.

6 is a diagram illustrating a process of classifying zooplankton through a sample number analysis method according to an embodiment of the present invention.

Referring to FIG. 6, a plurality of moving objects were detected in the upper image, but as a result of comparing the tracking images of the detected moving objects with machine-learned model data, one movement having an accuracy equal to or greater than a predetermined threshold as disclosed in the lower image. Only the subject was judged to be zooplankton.

Meanwhile, the apparatus 100 for analyzing the number of samples according to an embodiment of the present invention may further include a display unit 140 for displaying a captured image, as shown in FIG. 7.

In the display unit 140, an image unit 141 showing an image of an aquatic creature is located on the left side, and on the right side, the tracking image and model data of the moving objects 141a, 141b, and 141c displayed by the image unit 141 As a result of comparative analysis, zooplankton having an accuracy of more than a predetermined threshold is disclosed.

Here, the disclosed zooplankton may be disclosed for each type.In order to recognize the type of zooplankton, the model data generated by machine learning each type of tracking image in advance and the tracking image of the moving object are compared with the type image 142 Analysis information such as, type name 143, accuracy 144 and size 145 may be provided.

In this case, only the accuracy 144 having a predetermined value (for example, 50%) or more may be displayed on the display unit 140.

Here, the displayed zooplankton size 145 may be calculated based on the number of horizontal and vertical pixels and the actual size per pixel of the moving objects 141a, 141b, and 141c. For example, when the number of pixels in the area is 18*25 and the actual size per pixel is 2.5mm, the actual size is 45mm*62.5mm, and this actual size is displayed on the display unit 140.

In addition, the display unit 140 may include a size display unit 146 that displays whether the size of the zooplankton is 50 μm or more based on the sizes of the moving objects 141a, 141b, and 141c. As an example of the size display unit 146, FIG. 7 is configured to display how many individuals are 50 μm or more and how many individuals are less than 10 to 50 μm.

In the example of FIG. 7, it was determined that three kinds of zooplankton exist in water, and all of the sizes were analyzed to be 50 μm or more.

As described above, the sample number analysis apparatus 100 of the present invention receives an image of the number of samples photographed by the imaging unit 120 by the control unit 130, and separates and analyzes floating matters and living organisms in the sample water, thereby sterilizing ballast water. It is possible to judge whether the result satisfies the IMO criteria. If the standard is satisfied, the deballasting operation is performed to discharge the treated water to the coastal waters. If the standard is not satisfied, the discharge of the treated water is stopped and a reprocessing process for sterilization of aquatic organisms is performed.

In particular, the sample number analysis apparatus and method according to an embodiment of the present invention, when applied to a ballast water treatment apparatus installed on a ship where vibration or shaking occurs, the actual tracking image of the moving object and the previously generated model data It is possible to perform high-precision live and death discrimination by automatically distinguishing live plankton from floating objects by using the system. In addition, the size of moving objects determined as zooplankton is automatically counted on the basis of 50 μm to be used in the ballast water treatment system. It is possible to more quickly and accurately determine whether the ballast water processed by this method satisfies the emission standards of IMO.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the technical field to which the present invention belongs can make various modifications, changes, and substitutions within the scope not departing from the essential characteristics of the present invention. will be. Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are not intended to limit the technical idea of the present invention, but are for illustrative purposes, and the scope of the technical idea of the present invention is not limited by these embodiments and the accompanying drawings. . The scope of protection of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

11: plumbing
12: sampling port
20: thickening device
100: sample number analysis device
110: chamber
120: imaging unit
130: control unit

Claims (11)

A chamber in which the number of samples is accommodated;
An imaging unit for photographing the number of samples in the chamber; And
Including; a control unit for analyzing the image continuously photographed through the imaging unit,
The control unit,
Each of the schematic first tracking information generated by tracking the movement paths of living and inanimate objects is input, and model data generated through machine learning is used,
Detects a moving object based on the difference between two consecutive frames, tracks the detected moving object to generate schematic second tracking information,
A sample number analysis device that compares model data and second tracking information to determine whether a moving object is a living organism or inanimate object.
The method according to claim 1,
Schematic first or second tracking information,
A sample number analysis device consisting of a tracking image generated by connecting the positions of each frame with a continuous line.
The method according to claim 1,
Schematic first or second tracking information,
A sample number analysis device consisting of a chart showing the position of each frame in two-dimensional coordinates.
The method according to claim 1,
The control unit,
The number of samples analyzing apparatus for generating second tracking information by converting the size of the moving object based on the number of horizontal and vertical pixels of the detected moving object and the actual size per pixel, and tracking only moving objects having a predetermined size or more.
The method of claim 4,
The size of the moving object that generates the second tracking information is 10 μm or more and 50 μm or less, or 50 μm or more.
The method according to claim 1,
The first tracking information,
A device for analyzing the number of samples, in which living or non-living objects are labeled and input so that objects whose movement paths are tracked are distinguished.
The method according to claim 1,
Machine learning,
A sample number analysis device that generates model data using the CNN (Convolutional Neural Network) method.
Generating model data through machine learning by receiving each of the schematic first tracking information generated by tracking the movement paths of living and non-living objects;
Continuously photographing the number of samples in the chamber;
Detecting a moving object based on the difference between two consecutive frames, and generating schematic second tracking information by tracking the detected moving object;
Comprising the step of comparing the model data and the second tracking information to determine whether the moving object is a living organism or inanimate object; containing, sample number analysis method.
The method of claim 8,
Schematic first or second tracking information,
A sample number analysis device consisting of a tracking image generated by connecting the positions of each frame with a continuous line.
The method of claim 8,
Schematic first or second tracking information,
A sample number analysis device consisting of a chart showing the position of each frame in two-dimensional coordinates.
The method of claim 8,
The first tracking information and the second tracking information are measured for substantially the same time period.

Images