Method for extracting analysis region of liquid material image on basis of artificial intelligence

Abstract

The present invention relates to a method for extracting an analysis region of a liquid material image on the basis of artificial intelligence, the method comprising: providing, by a user terminal, a liquid material image obtained by photographing a rectangular chamber in a liquid material tester to a server and requesting a service; receiving the service request from the user terminal; extracting four right-angled shapes corresponding to an ROI corresponding to the region of the chamber from the liquid material image through artificial intelligence; generating point coordinates corresponding to the right-angled shapes extracted by the server; providing the point coordinates to the user terminal in response to the service request; identifying, by the user terminal, the number of the received point coordinates; and determining, by the user terminal, the ROI using the point coordinates if the number of the point coordinates is any one of two, three, and four.

Description

AI-based method for extracting analysis areas from liquid material images

[0001] The present invention relates to a technology for analyzing liquid substances such as sperm or saliva, and more specifically, to a method for extracting an analysis area of ​​a liquid substance image based on artificial intelligence (AI).

[0002] First, the term "liquid substance" in the following is defined to include all types of bodily fluids contained in the human body, such as saliva, blood, semen, sweat, urine, and tears, and more broadly to include various types of similar liquid substances, including organic and inorganic compounds used in scientific research, such as medicine, bio, water quality, and the environment.

[0003] Currently, as disclosed in Korean Patent Nos. 10-2435175 and 10-2288223, personal testing devices that can easily test for bodily fluids such as sperm among liquid substances have been developed and are being distributed.

[0004] These personal testing devices are used in conjunction with smartphones that have an analysis app installed. When the body fluid placed in the chamber of the personal testing device is enlarged and photographed with the smartphone, the analysis app installed on the smartphone finds the area of ​​the chamber, i.e., the ROI (Region of Interest), and analyzes the condition of the body fluid distributed in the ROI.

[0005] However, since the chamber is made of transparent material and is small in size, it is difficult to identify the area of ​​the chamber from the video taken by the analysis app. In particular, in the video taken by ordinary people, not experts, the area of ​​the chamber is often distorted on the screen, making it even more difficult to identify the area of ​​the chamber.

[0006] The purpose of the present invention is to provide a method for extracting a setting analysis area within an image using artificial intelligence (AI).

[0007] According to an embodiment of the present invention for achieving the above object, a method for extracting an analysis area of ​​a liquid substance image is provided, which is performed by a server having artificial intelligence trained to extract four point coordinates corresponding to a ROI in an image, the method comprising the steps of: receiving a service request including a liquid substance image photographing a rectangular chamber in a liquid substance tester from a user terminal; extracting four classes corresponding to ROIs corresponding to an area of ​​the chamber from the liquid substance image through the artificial intelligence; generating point coordinates from the extracted classes; determining the number of point coordinates; determining the ROI corresponding to the number of point coordinates; and providing area information of the ROI determined in response to the service request to the user terminal.

[0008] According to another embodiment of the present invention for achieving the above object, a method for extracting an analysis area of ​​a liquid substance image is provided, which is performed by a server having artificial intelligence trained to extract four point coordinates corresponding to a ROI in an image, the method comprising: receiving a service request including a liquid substance image photographing a rectangular chamber in a liquid substance tester from a user terminal; extracting four classes corresponding to ROIs corresponding to areas of the chamber from the liquid substance image through the artificial intelligence; generating point coordinates from the extracted classes; determining the number of point coordinates; determining the ROI corresponding to the number of point coordinates; generating an inscribed circle for the ROI; generating an inscribed rectangle for the inscribed circle; generating area information for the inscribed rectangle; and providing area information of the inscribed rectangle to the user terminal in response to the service request.

[0009] In a method according to one embodiment or another embodiment of the present invention, the step of extracting four classes corresponding to the ROI may be extracting rectangular shapes included in each class, and the step of generating point coordinates from the extracted classes may be generating point coordinates for vertices of rectangular shapes included in each class.

[0010] A method according to another embodiment of the present invention may further include, prior to the step of generating the inscribed circle, a step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; and a step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the amount of the distortion angle.

[0011] Alternatively, a method according to another embodiment of the present invention may further include, prior to the step of generating the inscribed circle, a step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; a step of comparing the distortion angle with a reference angle; and a step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the distortion angle if the distortion angle is greater than or equal to the reference angle.

[0012] Another method of the present invention for achieving the above object is a method for extracting an analysis area of ​​a liquid substance image performed by a user terminal using a server having artificial intelligence trained to extract four point coordinates corresponding to an ROI in an image, the method including the steps of: providing a liquid substance image photographing a chamber of a liquid substance tester to the server to make a service request; receiving point coordinates corresponding to an ROI corresponding to an area of ​​the chamber from the server in response to the service request; determining the number of the received point coordinates; and determining an ROI using the point coordinates when the number of the point coordinates is any one of 2, 3, and 4.

[0013] Another method of the present invention may further include, after the step of determining the ROI, the step of creating an inscribed circle for the ROI; the step of creating an inscribed rectangle for the inscribed circle; and the step of determining the area of ​​the inscribed rectangle as the analysis area.

[0014] Another method of the present invention may further include, before the step of generating the inscribed circle, a step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; and a step of adjusting the screen of the liquid material image so that there is no distortion by rotating the screen by the distortion angle, or may further include, before the step of generating the inscribed circle, a step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; a step of comparing the distortion angle with a reference angle; and a step of adjusting the screen of the liquid material image so that there is no distortion by rotating the screen by the distortion angle if the distortion angle is greater than or equal to the reference angle.

[0015] Another method of the present invention for achieving the above object may include the steps of: a user terminal providing a liquid substance image captured by a rectangular chamber in a liquid substance tester to a server to make a service request; a server receiving the service request from the user terminal; a server extracting four orthogonal shapes corresponding to a ROI corresponding to an area of ​​the chamber from the liquid substance image through artificial intelligence; a server generating point coordinates corresponding to the orthogonal shapes extracted by the server; a server providing the point coordinates generated by the server to the user terminal in response to the service request; a step for the user terminal to determine the number of the point coordinates received; and a step for the user terminal to determine the ROI using the point coordinates when the number of the point coordinates is any one of 2, 3, and 4.

[0016] According to an embodiment of the present invention, the present invention has the advantage of enabling accurate and rapid finding of a set analysis area in an image of a liquid substance using artificial intelligence.

[0017] FIG. 1 is a diagram showing a network environment according to an embodiment of the present invention.

[0018] Figure 2 is a flowchart of a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a first embodiment of the present invention.

[0019] Figure 3 is a flowchart of a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a second embodiment of the present invention.

[0020] FIG. 4 is a diagram illustrating a class for artificial intelligence learning according to an embodiment of the present invention.

[0021] FIG. 5 is a diagram for explaining learning data augmentation according to an embodiment of the present invention.

[0022] FIG. 6 and FIG. 7 are drawings for explaining a method of determining an ROI using point coordinates according to an embodiment of the present invention.

[0023] Figure 8 is a flowchart of a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a third embodiment of the present invention.

[0024] FIG. 9 is a drawing to help understand a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a third embodiment of the present invention.

[0025] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Regardless of the drawing reference numerals, identical or similar components will be assigned the same reference numerals, and redundant descriptions thereof will be omitted. Furthermore, when describing embodiments disclosed in this specification, if a detailed description of a related known technology is judged to obscure the gist of the embodiments disclosed in this specification, the detailed description thereof will be omitted.

[0026] Terms that include ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by these terms. These terms are used solely to distinguish one component from another.

[0027] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0028] In this application, each step described may be performed regardless of the listed order, except in cases where a special causal relationship requires that the steps be performed in the listed order.

[0029] In this application, terms such as “include” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but should be understood not to exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0030] Hereinafter, the present invention will be described with reference to the attached drawings.

[0031] FIG. 1 is a diagram illustrating a network environment according to an embodiment of the present invention. Referring to FIG. 1, the network environment according to an embodiment of the present invention may include a user terminal (10) and a server (100).

[0032] The user terminal (10) is a fixed terminal or a mobile terminal implemented as a computer device, and may include, for example, a smart phone, a mobile phone, a tablet PC, a navigation device, a computer, a laptop, a digital broadcasting terminal, a PDA (Personal Digital Assistant), etc.

[0033] The user terminal (10) can be used by a user. Here, the term "user" may refer to an account registered as a user for a service provided by the server (100). Therefore, for example, the server (100) transmitting certain information to a user may refer to transmitting the information to the user terminal (10) through the user's account registered with the server (100). Accordingly, the user can utilize the ROI service provided by the server (100) through the user terminal (10).

[0034] The ROI service may be a service that provides ROI information from a liquid substance image (e.g., still image or video) provided by a user terminal (10). Here, the ROI information may be one of point coordinates capable of detecting the ROI in the user terminal (10), ROI area information, and analysis area information.

[0035] Such a user terminal (10) may include a communication unit (11), an input unit (12), an output unit (13), a photographing unit (14), a memory (15), and a processor (16).

[0036] The communication unit (11) can provide a function that can communicate with the server (100) in a wired / wireless manner.

[0037] The input unit (12) can receive various information through the user's manipulation and input actions. Such an input unit may be at least one of a touch screen module, a keyboard, a mouse, a button, a camera, a stylus, and a microphone.

[0038] The user terminal (10) can receive user interaction input through the input unit (12). Interaction refers to the user manipulating the input unit (12) to input information reflecting the user's selection or intention into the terminal (10). For example, the interaction may be a touch of a touchscreen, a click of a mouse, typing on a keyboard, sound input from a microphone, image capture from a camera, or motion recognition from a motion sensor.

[0039] The output unit (13) can output various information. The output unit (13) can be a display device, a speaker, a vibration generating device, a tactile generating device, etc. The photographing unit (14) can be a camera module and can photograph a subject, i.e., a liquid substance placed in the chamber.

[0040] The memory (15) functions as a storage medium and may be provided in the form of various storage devices such as, for example, ROM, RAM, flash drive, hard drive, etc. in terms of hardware, or may be provided in the form of web storage. An operating system or at least one program code (for example, an application code for ROI service, i.e., an analysis app (15a)) may be stored in the memory (15). These software components may be loaded from a computer-readable recording medium separate from the memory (15). This separate computer-readable recording medium may include a computer-readable recording medium such as a floppy drive, a disk, a tape, a DVD / CD-ROM drive, a memory card, etc.

[0041] The processor (16) controls the overall operation of the communication unit (11), input unit (12), output unit (13), shooting unit (14), and memory (15), and executes commands received according to the analysis app (15a) stored in the memory (15) to enable ROI information to be provided through the server (100).

[0042] The processor (16) can perform at least an operation according to point coordinates, an operation according to area information of the ROI, and an operation according to analysis area information, depending on the type of ROI information.

[0043] Next, when the server (100) receives a request for an ROI service from a user terminal (10), it can generate ROI information using AI (artificial intelligence) from a liquid substance image included in the service request and provide the information to the user terminal (10).

[0044] For this purpose, the server (100) may include a memory (110), a processor (120), and a communication unit (130).

[0045] The memory (110) functions as a storage medium and can store a plurality of application programs running on the server (100), data for the operation of the server (100), commands, etc. For example, the memory (110) can store artificial intelligence (111), and the artificial intelligence (111) can learn class information of a set ROI from a liquid substance image provided by an administrator or a liquid substance image received from a user terminal (10).

[0046] The processor (120) can control the overall operation of the memory (110) and the communication unit (130) to provide an ROI service to the user terminal (10). For example, the processor (1200) can extract at least one point coordinate for the ROI based on artificial intelligence (111) for the liquid substance image provided by the user terminal (10), determine the area of ​​the ROI using the point coordinate, or determine the set analysis area among the ROI. Accordingly, the processor (120) can provide at least one of the point coordinate, the area information of the ROI, and the analysis area information as ROI information to the user terminal (10).

[0047] The communication unit (130) can communicate with the user terminal (10) via a network in a wired / wireless manner.

[0048] Here, the artificial intelligence-based liquid substance image analysis area extraction system according to an embodiment of the present invention may be configured to include a user terminal (10) and a server (100).

[0049] Hereinafter, a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a first embodiment of the present invention will be described with reference to FIGS. 2, 4 to 7.

[0050] FIG. 2 is a flowchart for a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a first embodiment of the present invention, FIG. 4 is a diagram for explaining a class for artificial intelligence learning according to an embodiment of the present invention, and FIG. 5 is a diagram for explaining learning data augmentation according to an embodiment of the present invention.

[0051] Referring to FIG. 2, the artificial intelligence (111) of the server (100) can learn a class for a rectangular ROI from a liquid substance image received from an administrator terminal (not shown) or a user terminal (10) in step S201.

[0052] Referring to FIGS. 4 and 5, the learning process of artificial intelligence (111) can be described. The artificial intelligence (111) can create a learning model for a rectangular ROI (40) displayed on the screen (30) of the liquid substance image illustrated in FIG. 4. To this end, the artificial intelligence (111) can define four classes corresponding to each corner of the rectangular ROI (40), that is, class A (C1), class B (C2), class C (C3), and class D (D), and can extract a right-angled shape (50) included in each class (C1 to C4) and learn it as a learning model.

[0053] At this time, the right-angled shape (50) of each class (C1 to C4) has a shape where a horizontal line segment and a vertical line segment meet, and can be determined according to the position of the horizontal line segment with respect to the vertical line segment. For example, the right-angled shape (50) of class A (C1) has a horizontal line segment intersecting the upper end of the vertical line segment and extending vertically to the right from the intersection point, the right-angled shape (50) of class B (C2) has a horizontal line segment intersecting the upper end of the vertical line segment and extending vertically to the left from the intersection point, the right-angled shape (50) of class C (C3) has a horizontal line segment intersecting the lower end of the vertical line segment and extending vertically to the right from the intersection point, and the right-angled shape (50) of class D (C4) has a horizontal line segment intersecting the lower end of the vertical line segment and extending vertically to the left from the intersection point.

[0054] And the artificial intelligence (111) can augment learning data as shown in FIG. 5 to respond to various deformations of the ROI (30). Augmenting learning data can generate various deformations by rotating the rectangular shape (50) of each class (C1 to C4) of the original image by a set angle. For example, FIG. 5 shows an example of generating and learning various deformations by rotating the original rectangular shape (50) to the right by 15° each time, but it can be rotated by various angles such as 5° and 10° instead of 15°, and various deformations can be generated by rotating to the left instead of rotating to the right.

[0055] Returning to FIG. 2, when the server (100) has generated a learning model for the ROI through step S201, the processor (16) of the user terminal (10) activates the analysis app (15a) in step S202 according to a user command input from the input unit (12), and, according to the request of the analysis app (15a), captures the liquid substance placed in the chamber of the liquid substance tester (not shown) through the capture unit (14) to generate an image of the liquid substance. At this time, the chamber may be rectangular in shape, and the area of ​​the chamber may be the ROI.

[0056] In step S203, the processor (16) may provide the captured liquid substance image to the server (100) at the request of the analysis app (15a). The information provided from the user terminal (10) to the server (100) in step S203 may include the liquid substance image and identification information of the user terminal (or user) (e.g., phone number, MAC address, member ID, etc.).

[0057] In step S204, the processor (120) of the server (100) stores the liquid substance image received from the user terminal (10) and then provides the liquid substance image to artificial intelligence (111) so that the learned learning model can be used to extract the rectangular shape of each class (C1 to C4).

[0058] In step S205, the artificial intelligence (111) can determine point coordinates for the intersection of the rectangular shapes of each extracted class (C1 to C4) and provide the determined point coordinates to the processor (120). Here, the maximum number of point coordinates extracted by the artificial intelligence (111) may be 4, and the minimum number may be 0.

[0059] In step S206, the processor (120) can provide the extracted point coordinates as ROI information to the user terminal (10).

[0060] In step S207, the processor (16) of the user terminal (10) can determine the number of point coordinates received from the server (100). In step S208, the processor (16) determines whether the number of point coordinates is two or more, and if it is two or more, the processor can determine the ROI of the rectangle using the point coordinates in step S209.

[0061] On the other hand, the processor (16) can determine that the number of point coordinates is 0 or 1 in step S208 and perform a corresponding response action in step S211. The response action may be an action to notify the user of an extraction failure or to notify the user to take a new picture.

[0062] In step S210, the processor (16) can perform analysis on the liquid substance distributed in the ROI determined in step S209 using the analysis app (15a).

[0063] Hereinafter, an AI-based liquid substance image analysis area extraction method according to a second embodiment of the present invention will be described with reference to FIGS. 3 to 5. FIG. 3 is a flowchart of an AI-based liquid substance image analysis area extraction method according to a second embodiment of the present invention.

[0064] Referring to FIG. 3, the artificial intelligence (111) of the server (100) can learn a class for a rectangular ROI from a liquid substance image received from an administrator terminal (not shown) or a user terminal (10) in step S301. Here, step S301 is identical to step S201 described above.

[0065] While the server (100) has created a learning model for the ROI through step S301, the processor (16) of the user terminal (10) can capture the chamber through steps S302 and S303 to create a liquid substance image and provide the liquid substance image to the server (100).

[0066] In step S304, the processor (120) of the server (100) stores the liquid substance image received from the user terminal (10) and then provides the liquid substance image to artificial intelligence (111) so that the learned learning model can be used to extract the rectangular shape of each class (C1 to C4).

[0067] In step S205, the artificial intelligence (111) can determine point coordinates for the intersection of the rectangular shapes of each extracted class (C1 to C4) and provide the determined point coordinates to the processor (120). Here, the maximum number of point coordinates extracted by the artificial intelligence (111) may be 4, and the minimum number may be 0.

[0068] In step S206, the processor (120) can determine the number of point coordinates extracted from the artificial intelligence (111). Then, in step S307, the processor (120) determines whether the number of point coordinates is two or more, and if so, can determine the ROI of the rectangle using the point coordinates in step S308.

[0069] On the other hand, the processor (120) may determine that the number of point coordinates is 0 or 1 in step S307, and perform a corresponding response action in step S311. The response action may be an action to notify the user of an extraction failure or to notify the user to take a new picture.

[0070] In step S309, the processor (120) can provide area information for the ROI determined in step S308 to the user terminal (10) as ROI information. At this time, the area information of the ROI may be the coordinates of the four vertices of the ROI or the border coordinates of the ROI.

[0071] In step S310, when the processor (16) of the user terminal (10) receives ROI information from the server (100), it determines an ROI in the liquid substance image based on the ROI information, and performs analysis on the liquid substance distributed in the determined ROI using the analysis app (15a).

[0072] Hereinafter, a method for determining an ROI using point coordinates in a user terminal (10) or a server (100) will be described in more detail with reference to FIGS. 6 and 7. The method for determining an ROI using point coordinates in a user terminal (10) may correspond to steps S207 to S209, and the method for determining an ROI using point coordinates in a server (100) may correspond to steps S306 and S308.

[0073] FIGS. 6 and 7 are diagrams illustrating a method for determining an ROI using point coordinates according to an embodiment of the present invention. Referring to FIG. 6 , the processor (16 or 120) can determine the number of point coordinates in step S601. The number of point coordinates determined in step S601 may be one of 0, 1, 2, 3, and 4.

[0074] If the number of point coordinates is 0 or 1, the processor (16 or 120) can notify the user of an extraction failure or to take a new picture in step S602.

[0075] If the number of point coordinates is two, the processor (16 or 120) can determine the positional status of the point coordinates, i.e., whether the two point coordinates are adjacent or whether the two point coordinates are located in the diagonal direction, in step S603.

[0076] When two point coordinates are positioned in a vertical or horizontal direction, the processor (16 or 120) can determine the length of one side (X side or Y side) of a straight line drawn between two coordinates (P1 and P2 or P1 and P3) as shown in (a) and (b) of FIG. 7 in step S604, draw a line perpendicular to the straight line drawn from each of the two point coordinates by the length of the one side determined, and connect one end of the drawn two straight lines to determine a quadrangular ROI. Here, the ROI can be a square.

[0077] If the processor (16 or 120) has two point coordinates (P2, P3) positioned diagonally, the processor (16 or 120) can determine the ROI by drawing a diagonal line by connecting the two point coordinates (P2, P3) with a straight line as shown in (3) of FIG. 7 in step S605, and using the Pythagorean definition, by making the lines extending vertically from each of the two point coordinates (P2, P3) and meeting each other form 90° to each other.

[0078] And, if the number of point coordinates is 3, the processor (16 or 120) can determine the ROI by drawing a straight line of one side length in the vertical direction for each of the two point coordinates where two straight line segments are not connected, that is, two point coordinates where only one straight line segment is connected, so that they meet each other.

[0079] Finally, the processor (16 or 120) can determine the ROI by connecting each of the four point coordinates in step S607 if the number of point coordinates is three.

[0080] Hereinafter, a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a third embodiment of the present invention will be described with reference to FIGS. 8 and 9.

[0081] FIG. 8 is a flowchart for a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a third embodiment of the present invention, and FIG. 9 is a drawing to help understand a method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to a third embodiment of the present invention.

[0082] The method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to the third embodiment of the present invention is for extracting an analysis area to analyze a liquid substance contained within an area inside the ROI, i.e., an analysis area, without analyzing the liquid substance within the ROI.

[0083] In the case of liquid substances with mobility, such as sperm, the liquid substances distributed near the edges of the ROI (i.e., the chamber) and those distributed in the center exhibit different activities. Specifically, liquid substances distributed near the edges of the ROI exhibit low activity due to blocked movement paths, while liquid substances distributed in the center exhibit high activity due to unrestricted movement.

[0084] Therefore, the method for extracting an analysis area of ​​an artificial intelligence-based liquid substance image according to the third embodiment of the present invention can extract an analysis area of ​​a central portion showing high activity among ROIs.

[0085] The method for extracting an analysis region of an artificial intelligence-based liquid substance image according to the third embodiment of the present invention may include an additional process performed between the step of determining the ROI and the step of analyzing the liquid substance when performing the method for extracting an analysis region of an artificial intelligence-based liquid substance image according to the first or second embodiment of the present invention.

[0086] In step S801, the processor (16) of the user terminal (10) or the processor (120) of the server (100) can determine the ROI.

[0087] Thereafter, at step S802, the processor (16 or 120) can calculate the distortion angle by which the ROI of the rectangle is distorted. The distortion angle may be the distortion angle, i.e., the rotation angle, of the liquid material image relative to the screen.

[0088] A first method for calculating a distortion angle is to create a reference ROI (30a) having a distortion of 0° on the screen of a liquid material image as shown in (b) of FIG. 9, then draw a first straight line connecting one vertex of the reference ROI (30a) from the center point (a) of the screen, and then draw a second straight line connecting one vertex of the created ROI (30) corresponding to one vertex of the reference ROI (30a) from the center point (a) of the screen, and then calculate the angle between the first straight line and the second straight line as the distortion angle. Here, corresponding to one vertex of the reference ROI (30a) may mean that it is a vertex of the same right angle shape.

[0089] Alternatively, a second method for calculating the distortion angle is to draw a horizontal straight line (L1) passing through the center point (a) of the screen, as shown in (e) of FIG. 9, and draw a third straight line on one side of the ROI (30) where the straight line (L1) does not intersect, and calculate the angle between the straight line (L1) and the third straight line as the distortion angle.

[0090] Alternatively, a third method for calculating the distortion angle is to draw a horizontal straight line (L1) passing through the center point (a) of the screen, as illustrated in (e) of FIG. 9, and then move the straight line (L1) upward or downward to create a straight line (L2) that intersects one vertex of the ROI (30) (for example, it may be the point coordinate corresponding to P4 in FIG. 7), and then calculate the angle with the corner intersecting the straight line (L2) as the distortion angle. Here, the corner intersecting the straight line (L2) may be the corner connected to the point coordinate (P3) when the straight line (L2) is located downward from the straight line (L1), and may be the corner connected to the point coordinate (P2) when the straight line (L2) is located upward from the straight line (L1).

[0091] In step S803, the processor (16, 120) can compare the misalignment angle calculated in step S802 with a reference angle. The reference angle may be ±5°, ±10°, ±15°, etc. Here, a + (plus) angle may mean that the ROI (30) is misaligned by rotating to the right with respect to the screen, and a - (minus) angle may mean that the ROI (30) is misaligned by rotating to the left with respect to the screen. Here, the reference angle may be a misalignment angle that does not affect the analysis of the liquid substance.

[0092] At step S804, the processor (16, 120) can determine whether the misalignment angle is greater than or equal to the reference angle.

[0093] And the processor (16, 120) determines that even if the ROI (30) with the current twist angle is used when the twist angle is smaller than the reference angle, it does not affect the analysis of the liquid substance, and accordingly, in step S805, as shown in (f) of FIG. 9, an inscribed circle (b) is formed in the ROI (30), and in step S806, an inscribed rectangle (c) is formed in the inscribed circle (b), and the inscribed rectangle (c) can be determined as the analysis area.

[0094] In the case where the inscribed rectangle (c) is determined as the analysis area, the processor (120) of the server (100) may provide the four point coordinates for the analysis area (c) as ROI information to the user terminal (10), or may provide the corner coordinates for the analysis area (c) as ROI information to the user terminal (10).

[0095] Alternatively, the processor (16) of the user terminal (10) determines an inscribed rectangle (c) as an analysis area and then performs analysis on the liquid substance distributed in the analysis area, or determines the analysis area (c) in the liquid substance image using four point coordinates for the analysis area (c) received from the server (100) or corner coordinates for the analysis area (c), and performs analysis on the liquid substance distributed in the determined analysis area (c) as shown in (d) of FIG. 9.

[0096] Meanwhile, if the processor (16, 120) determines that the misalignment angle is greater than the reference angle in step S804, it can create an ROI without misalignment on the screen by rotating the ROI (30) in the opposite direction of the misalignment angle by the misalignment angle as shown in (c) of FIG. 9 in step S807.

[0097] Thereafter, the processor (16, 120) can sequentially perform steps S805 and S806 for a ROI without distortion to determine an analysis area (C).

[0098] The technical features disclosed in each embodiment of the present invention are not limited to that embodiment, and, unless they are mutually incompatible, the technical features disclosed in each embodiment may be combined and applied to different embodiments.

[0099] Therefore, although each embodiment focuses on its own technical features, each technical feature can be applied in combination with each other as long as they are not mutually incompatible.

[0100] The present invention is not limited to the above-described embodiments and the attached drawings, and various modifications and variations are possible within the scope of those skilled in the art. Therefore, the scope of the present invention should be defined not only by the claims of this specification but also by equivalents thereof.

Claims

1. A method for extracting an analysis area of ​​a liquid substance image, performed on a server with artificial intelligence trained to extract four point coordinates corresponding to the ROI within the image. A step of receiving a service request including a liquid substance image captured in a rectangular chamber in a liquid substance tester from a user terminal; A step of extracting four classes corresponding to the ROI corresponding to the area of ​​the chamber from the liquid material image using the artificial intelligence; A step of generating point coordinates from the extracted above class; A step of determining the number of above point coordinates; A step of determining the ROI corresponding to the number of the above point coordinates; and Including a step of providing the area information of the ROI determined in response to the service request to the user terminal. A method for extracting analysis areas from liquid material images based on artificial intelligence.

2. A method for extracting an analysis area of ​​a liquid substance image performed on a server with artificial intelligence trained to extract four point coordinates corresponding to the ROI within the image. A step of receiving a service request including a liquid substance image captured in a rectangular chamber in a liquid substance tester from a user terminal; A step of extracting four classes corresponding to the ROI corresponding to the area of ​​the chamber from the liquid material image using the artificial intelligence; A step of generating point coordinates from the extracted above class; A step of determining the number of above point coordinates; A step of determining the ROI corresponding to the number of the above point coordinates; A step of creating an inscribed circle for the above ROI; A step of generating an inscribed rectangle for the above inscribed circle; a step of generating area information for the above inscribed rectangle; and Including a step of providing area information of the inscribed rectangle to the user terminal in response to the service request. A method for extracting analysis areas from liquid material images based on artificial intelligence.

3. In paragraph 1 or 2, The step of extracting the four classes corresponding to the above ROI is to extract the rectangular shapes included in each class. The step of generating point coordinates from the above extracted classes is to generate the point coordinates for the vertices of the rectangular shapes included in each of the above classes. A method for extracting analysis areas from liquid material images based on artificial intelligence.

4. In paragraph 2, Before the step of creating the above inscribed circle, A step of calculating the distortion angle of the ROI with respect to the screen of the liquid material image; and Further comprising a step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the above distortion angle. A method for extracting analysis areas from liquid material images based on artificial intelligence.

5. In paragraph 2, Before the step of creating the above inscribed circle, A step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; A step of comparing the above-mentioned distortion angle with a reference angle; and If the above distortion angle is greater than the reference angle, the step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the amount of the distortion angle is further included. A method for extracting analysis areas from liquid material images based on artificial intelligence.

6. In a method for extracting an analysis area of ​​a liquid substance image performed on a user terminal using a server having artificial intelligence trained to extract coordinates of four points corresponding to the ROI within the image, A step of requesting a service by providing a liquid substance image captured in the chamber of a liquid substance tester to the server; A step of receiving point coordinates corresponding to the ROI corresponding to the area of ​​the chamber from the server in response to the service request; A step of determining the number of received point coordinates; and Including a step of determining an ROI using the point coordinates when the number of the point coordinates is one of 2, 3, and 4. A method for extracting analysis areas from liquid material images based on artificial intelligence.

7. In paragraph 6, After the step of determining the above ROI, A step of creating an inscribed circle for the above ROI; A step of generating an inscribed rectangle for the above inscribed circle; Further comprising the step of determining the area of ​​the inscribed rectangle as the analysis area, A method for extracting analysis areas from liquid material images based on artificial intelligence.

8. In paragraph 7, Before the step of creating the above inscribed circle, A step of calculating the distortion angle of the ROI with respect to the screen of the liquid material image; and Further comprising a step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the above distortion angle. A method for extracting analysis areas from liquid material images based on artificial intelligence.

9. In paragraph 7, Before the step of creating the above inscribed circle, A step of calculating a distortion angle of the ROI with respect to the screen of the liquid material image; A step of comparing the above-mentioned distortion angle with a reference angle; and If the above distortion angle is greater than the reference angle, a further step of adjusting the screen of the liquid material image so that there is no distortion by rotating it by the amount of the distortion angle is included. A method for extracting analysis areas from liquid material images based on artificial intelligence.

10. A step in which a user terminal requests a service by providing a liquid substance image captured in a square chamber of a liquid substance tester to the server; A step in which the server receives the service request from the user terminal; A step in which the server extracts four orthogonal shapes corresponding to the ROI corresponding to the area of ​​the chamber from the liquid material image through artificial intelligence; A step of generating point coordinates corresponding to the rectangular shape extracted by the server; A step of providing the point coordinates generated by the server to the user terminal in response to a service request; A step of determining the number of point coordinates received by the user terminal; and The user terminal includes a step of determining an ROI using the point coordinates when the number of the point coordinates is any one of 2, 3, and 4. A method for extracting analysis areas from liquid material images based on artificial intelligence.

11. In paragraph 10, After the step of determining the above ROI, A step in which the user terminal generates an inscribed circle for the ROI; A step in which the user terminal generates an inscribed rectangle for the inscribed circle; The user terminal further includes a step of determining the area of ​​the inscribed rectangle as an analysis area. A method for extracting analysis areas from liquid material images based on artificial intelligence.

12. In paragraph 11, Before the step of creating the above inscribed circle, A step in which the user terminal calculates the distortion angle of the ROI with respect to the screen of the liquid material image; and The step of adjusting the screen of the liquid material image so that there is no distortion by rotating the user terminal by the distortion angle is further included. A method for extracting analysis areas from liquid material images based on artificial intelligence.

13. In paragraph 11, Before the step of creating the above inscribed circle, A step in which the user terminal calculates the distortion angle of the ROI with respect to the screen of the liquid material image; The step of the user terminal comparing the misalignment angle with a reference angle; and The user terminal further includes a step of adjusting the screen of the liquid material image so that there is no distortion by rotating the screen by the distortion angle if the distortion angle is greater than the reference angle. A method for extracting analysis areas from liquid material images based on artificial intelligence.

Images