KR20240016937A - Automatic parametric patterning method of garment pattern using artificial neural network - Google Patents

Abstract

Collected clothing pattern files are converted into pattern data in a predetermined form, edited and processed as necessary, and the types and values of dimensions used in pattern drafting are systematized and stored. Artificial neural network-based learning is performed using labeled learning data that inputs the values of the dimensions used when drafting each of the clothing pattern files and outputs the coordinates of key points with the same name, creating a corresponding artificial neural network for each key point. Create a neural network model. Artificial neural network models that have been learned to generate coordinates of key points required for drafting each clothing pattern are collected and a parametric pattern corresponding to the pattern is automatically created. By inputting the desired body dimensions into the artificial neural network models corresponding to the key points in each parametric pattern, the coordinates of the key points and the coordinates of the curve control points between the key points are calculated, and the coordinate values of the calculated points are calculated. It automatically creates clothing patterns that fit your body measurements.

Description

Automatic parametric patterning method of garment pattern using artificial neural network}

The present invention relates to the field of technology for automating clothing pattern design, and more specifically, to technology for making clothing patterns of fixed dimensions into parametric patterns using artificial neural networks.

In clothing production, it is necessary to create a clothing pattern based on a 2D schematic created for clothing design. Previously, clothing patterns were created using a pattern CAD program. Existing pattern CAD programs provide the ability to create patterns by inputting paper patterns into a computer and modifying the pattern graphics displayed on the screen.

In pattern making, even for clothes of the same shape, the size of the pattern varies depending on the size, so a separate pattern must be created for each size. However, the existing pattern CAD program does not provide connectivity between the elements (dots, lines, etc.) that make up the pattern when creating a pattern, so the pattern components have an independent relationship with each other. Therefore, if patterns of one shape need to be produced with various dimensions, pattern creation work must be done separately for each dimension. In other words, in the existing pattern creation method, even if a pattern for one dimension (basic dimension) is created, in order to create a pattern for another dimension, the previous work must be repeated almost exactly. The content of the work of creating a pattern of the same shape with different dimensions is substantially the same, but repeating the work for each dimension is inefficient. A pattern design method that can solve this inefficiency is needed.

In clothing production, ready-made clothing is mainly produced in various sizes through grading, which has limitations in producing clothing that perfectly fits the various body types of consumers, making it difficult to satisfy all consumers. On the other hand, when clothing is produced using a custom-made method, satisfaction is high because it is possible to produce clothing that fits the consumer's body well, but there is a problem that it takes a lot of time and money to produce the clothing. A method is also needed to significantly reduce the time and cost of clothing production while being suitable for a variety of consumer body types.

There have been several attempts to overcome the limitations of ready-made and custom clothing production methods, while overcoming the inefficiencies of existing pattern design methods. One of those attempts is to design clothing patterns using parametric design methods. Once parametric pattern data is secured, the degree of automation of clothing design, virtual simulation, and actual clothing production work can be greatly increased. Additionally, the parametric pattern design method is effective in creating patterns of various dimensions. However, in order to use the parametric pattern design method, so-called parametric patterning work, which converts existing pattern image data into parametric pattern data, must be performed first. Converting numerous existing pattern image data into parametric pattern data requires a lot of time and specialized skills. A method is needed to perform parametric patterning work effectively and economically.

1. Republic of Korea Patent No. 10-2173900 (Clothing design creation method and system and integrated application program for the same)

One purpose of the present invention is to solve the above-mentioned problems, and to provide a method of automatically converting already prepared clothing patterns of fixed dimensions into parametric patterns using an artificial neural network without the need for parametric formulas.

Another object of the present invention is to provide a method that automatically generates a pattern corresponding to the user's desired body size using an artificial neural network model obtained by performing artificial neural network-based learning with pre-drawn clothing pattern image files. .

Another object of the present invention is to provide a method that can visually evaluate the suitability of a pattern created through 3D clothing drape simulation and reduce the time and effort required in the pattern design and modification process.

The problems to be solved by the present invention are not limited to the above-described problems, and may be expanded in various ways without departing from the spirit and scope of the invention.

The artificial neural network-based automatic parametric patterning method for clothing patterns according to embodiments for realizing an object of the present invention is a method of automatically generating clothing patterns using a computer program running on a computer device, comprising a plurality of methods. Converting clothing pattern files graded to the clothing size into internal pattern data and storing the information edited and processed by the user in a manageable data storage along with the pattern data; Providing a size parameter management function that can store the types and values of dimensions used when drafting each of the clothing pattern files in a data storage in a format that can be systematically managed; The values of the dimensions used when drafting each of the above clothing pattern files are input, and key points having the same name (however, in clothing patterns graded in multiple sizes, key points at the same location are given the same name) Performing artificial neural network-based learning using labeled learning data with coordinates as output to generate a corresponding artificial neural network model for each key point; And it includes a step of automatically generating a parametric pattern corresponding to the pattern by gathering artificial neural network models that have been learned to generate coordinates of key points necessary for drafting each clothing pattern.

In an exemplary embodiment, in the saving step, for the necessary editing and processing, the user retrieves the clothing pattern file stored in the data storage, deletes unnecessary files, and removes unnecessary pieces, auxiliary lines, and dots from the clothing pattern image. providing functions that support performing a desired task, including a deletion task; Selecting key points to be related to human body dimensions from the imported clothing pattern and providing a function to name each key point; Providing a function that allows the user to determine an origin point that will serve as a standard for aligning clothing pattern images produced with the plurality of clothing dimensions; Providing a function that allows the user to select adjacent points in a clothing pattern image and connect the selected points to form a segment of the pattern; And it may include providing a function for the user to connect a plurality of sides displayed in the clothing pattern image to define the outline, cutting line, seam line, and/or internal line of the pattern.

In an exemplary embodiment, the saving step includes a function of calculating a predetermined ratio A/L and B/L to obtain the coordinates of the curve control points constituting the curve, if the transition curve of the pattern is configured. (Here, L is the distance between the adjacent first and second key points, and B is the length of the first vertical line drawn from the curve control point to the first straight line connecting the first and second key points, A is the distance from the intersection between the first straight line and the first vertical line to the second key point).

In an exemplary embodiment, the main points may include a vertex of the pattern, a point at which the curvature of the curve changes, and a notch point that serves as a reference during sewing.

In an exemplary embodiment, the storing step is such that when patterns of the same shape graded into a plurality of clothing sizes are displayed on a computer device, the origins specified for the patterns of the same shape are all at the same position. A step of providing an automatic sorting function may be further included.

In an exemplary embodiment, the size parameter management function includes: a function for managing equations and variables that define the relationship between dimensions and anthropometric measurements used in clothing pattern drafting; Ability to manage the impact of dimensions used in garment pattern drafting on the coordinates of key points; And it may include a function to manage the influence of the dimensions used in clothing pattern drafting on the coordinates of key points.

In an exemplary embodiment, the parametric pattern may be created in script form and stored in the data storage.

In an exemplary embodiment, the script of the parametric pattern includes definitions of human body dimensions for each pattern, definitions of pattern drafting variables, basic information of the pattern, definitions of key points corresponding to the artificial neural network model, It may contain definitions of subpoints and lines determined by the positions of the main points.

In an exemplary embodiment, the artificial neural network-based clothing pattern automatic parametric patterning method involves inputting desired body dimensions into artificial neural network models corresponding to key points in each parametric pattern, and calculating the coordinates of the key points and It may further include calculating the coordinates of curve control points between the main points and automatically generating a clothing pattern that fits the body size using the coordinate values of the calculated points.

In an exemplary embodiment, the step of automatically generating the clothing pattern includes, when the user executes the pattern creation function and inputs the desired human body size, the input values are entered using a formula defined in the size parameter management function. Converting to dimensions used in pattern drafting; inputting the converted dimensions into the artificial neural network models and outputting coordinates of the key points; calculating the coordinates of a curve control point between two adjacent key points using the relative positions of the key points; And it may include automatically generating patterns that fit the input body size using the calculated coordinates of key points and curve control points.

In an exemplary embodiment, the step of automatically generating the clothing pattern includes, when the position of a first key point, which is one of two adjacent key points, is changed, the coordinates of the changed position of the curve control point are expressed by Equation A It may further include a step of calculating using the formula '=L'×(A/L) and B'=L'×(B/L). Here, L is the distance between the adjacent first and second key points, B is the length of the first vertical line drawn from the curve control point to the first straight line connecting the first and second key points, and A is the distance from the intersection between the first straight line and the first vertical line to the second main point, and B' is the second main point and the first main point whose position has been changed at the new curve control point after changing the position of the first main point. It is the length of the second vertical line drawn from the second straight line connecting the points, and A' is the distance from the intersection between the second straight line and the second vertical line to the second main point.

In an exemplary embodiment, the artificial neural network-based clothing pattern automatic parametric patterning method places automatically generated clothing patterns on a consumer's human body model through 3D clothing drape simulation and specifies sewing conditions to make the clothing. Simulating with; And to evaluate the suitability of the clothing patterns, it may further include calculating clothing pressure and margin distribution and displaying them on the simulated clothing.

In an exemplary embodiment, the clothing pressure (P _pressure ) is expressed by the equation It can be calculated as However, P _strain represents the tensile force, a _i represents the area of the original triangular element before three-dimensional transformation, and A _i represents the area of the triangular element after three-dimensional transformation.

In an exemplary embodiment, the artificial neural network-based automatic parametric patterning method for clothing patterns is a method of limiting or alleviating the influence relationship that a specific body size item has on the coordinates of key points of the pattern according to the suitability evaluation results. A step of modifying artificial neural network models corresponding to key points may be further included.

In an exemplary embodiment, the artificial neural network-based automatic parametric patterning method for clothing patterns creates a file in the industry standard drawing exchange format (DXF) and stores it in a data storage so that the completed clothing pattern can be transmitted to a sewing factory. Additional steps may be included.

According to exemplary embodiments of the present invention, an existing clothing pattern of fixed dimensions can be automatically converted into a parametric pattern whose shape can be changed to suit the body dimensions of various consumers.

In addition, the automatic parametric pattern generation program implemented according to exemplary embodiments of the present invention minimizes matters to be considered at the level of a pattern design expert, such as grading deviation, relationship with human body dimensions, pattern drafting formula, and fabric characteristics, Even general users without expert knowledge can automatically create parametric patterns without pattern formulas. Accordingly, ordinary consumers with various body types can easily create clothing patterns that fit their body types.

Additionally, clothing patterns automatically generated according to exemplary embodiments of the present invention can be visualized in 3D through 3D clothing drape simulation to evaluate the visual suitability of the patterns. Therefore, the time and effort required for the pattern design and modification process can be reduced. Through this, even general users without specialized knowledge can easily create clothing patterns for consumers of various body types. Furthermore, considering the suitability evaluation results, the artificial neural network can be improved by estimating the relationship between body dimensions and the coordinates of key points of the pattern.

Figure 1 is a flowchart of an automatic parametric patterning algorithm for artificial neural network-based clothing patterns according to the present invention.
FIG. 2 is a flowchart showing the detailed execution procedure of step S10 shown in FIG. 1.
Figure 3 illustrates a case where key points are designated for various patterns.
Figure 4 shows an example in which origins are marked on multiple identical shape patterns graded to multiple sizes and the origins are overlapped and aligned.
FIG. 5 is a flowchart showing the detailed procedure of step S20 shown in FIG. 1.
FIG. 6 is a flowchart showing the detailed procedure of step S30 shown in FIG. 1.
Figure 7 shows an example of parametric pattern data stored in script form.
Figure 8 schematically shows the definition of curve control points.
Figure 9 schematically shows a method for calculating clothing pressure.
Figure 10 shows an example of visualizing the distribution of clothing pressure and allowance by simulating a drape on a 3D human body model using the calculated clothing pressure and allowance.
Figure 11 illustrates the configuration of a computer device for performing a method of automatic parametric patterning of a clothing pattern using an artificial neural network by executing the automatic parametric patterning computer program according to the embodiment of the present invention described above.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the attached drawings. The same reference numerals are used for the same components in the drawings, and duplicate descriptions for the same components are omitted.

Figure 1 shows a flow chart of an automatic parametric patterning algorithm for artificial neural network-based clothing patterns according to the present invention.

An algorithm for automatic parametric patterning of clothing patterns can be implemented as a computer program. The automatic parametric patterning computer program may be stored in a storage medium readable by a computer device. The processor of the computer device can automatically convert the clothing pattern into parametric pattern data using an artificial neural network by executing the automatic parametric patterning computer program stored in the storage medium.

Referring to FIG. 1, when the automatic parametric patterning computer program is executed in the processor of the computer device, it supports the processor to perform internally the computer device and/or perform various tasks required by the user. It can provide several functions.

To explain its functions in detail, first, the automatic parametric patterning computer program has the function of collecting, editing, processing, and storing existing clothing pattern data graded into various clothing sizes in a data storage so that the user can manage it. can be provided. The edited and processed clothing pattern data can be stored, for example, in a database (step S10).

To explain step S10 in more detail, generally, one garment may be composed of a combination of multiple patterns. The pattern combination is created not only for one dimension (size), but usually for multiple dimensions (sizes: for example, for women's clothing, sizes 44, 55, 66, 77, etc.; for men's clothing, tops 95, 100, 105, etc. Size 110, etc., bottoms 30, 31, 32, 33, 34, 35 inches, etc.) can be drawn with separate pattern combinations for each. A user can collect multiple pattern image files relating to a combination of multiple patterns drawn at various dimensions for a single garment. Collected pattern image files (e.g. CAD files) graded in various sizes are read from a computer device and converted into internal pattern data used in an automatic parametric patterning computer program, and then go through various editing and processing operations to create the final result. For example, it can be created as a file in the standard drawing exchange format of the clothing pattern industry (Drawing Exchange Format (DXF)) and stored in a database for systematic management. Even after saving, various editing and processing operations can be performed on the pattern data.

The flowchart of FIG. 2 shows the detailed execution procedure of step S10 shown in FIG. 1.

Referring to FIG. 2, functions that allow a user to load a pattern image file stored in a database and perform various desired operations on the file can be provided (step S110).

The user can perform various arrangements on the pattern image files retrieved from the database (step S120). For example, while performing tasks such as editing or processing a pattern image file of one size, a pattern image file of a different size can be added or an unnecessary pattern image file can be deleted. You can also change the order of loaded pattern image files. You can also perform tasks such as deleting unnecessary pieces, auxiliary lines, or points from clothing patterns.

In addition, the user can select key points from each retrieved pattern that will be related to human body dimensions and name them (step S130). Main points can be the vertex of the pattern, the point where the curvature of the curve changes, or the notch point that serves as a standard for sewing.

Figure 3 illustrates a case where key points are designated for various patterns. Referring to Figure 3, each pattern (10, 12, 14, 16, 18) is a plurality of patterns (10 ₁ -10 _n ), (12 ₁ - It may include 12 _n ), (14 ₁ -14 _n ), (16 ₁ -16 _n ), and (18 ₁ -18 _n ). In each of the illustrated patterns (10, 12, 14, 16, and 18), the dots in small square boxes are examples of main points. Drawing reference numbers 10-P1, 10-Pn, 12-P1, 12-Pn, 14-P1, 14-Pn, 16-P1, 16-Pn, 18-P1, 18-Pn, etc. indicate some key points.

Additionally, the user can set an origin point that will serve as a standard for aligning patterns made with various clothing dimensions (step S140). The origin is a reference point that makes it easy to check the shape when placing multiple graded patterns. The origin can be set to an arbitrary point according to the user's convenience.

Figure 4 shows an example in which origins are marked on multiple identical shape patterns graded to multiple sizes and the origins are overlapped and aligned.

Referring to FIG. 4, (A) and (B) are a plurality of first patterns (20 ₁ -20 _n ) drawn in multiple sizes with the same pattern shape aligned by overlapping their origins (20-P0). It is shown. The origin (20-P0) of the first patterns (20 ₁ -20 _n ) shown in (A) is located at the upper left side of the pattern, so that the larger the size of the first patterns (20 ₁ -20 _n ), the lower right. A portion of the pattern is exposed, and the origin (22-P0) of the plurality of first patterns (20 ₁ -20 _n ) shown in (B) is located at the lower left vertex, so that the first patterns (20 ₁₎ -20 _n ) The larger the size, the more part of the pattern is exposed to the upper right. In addition, (C) and (D), like (A) and (B), have a plurality of second patterns (24 ₁ -24 _m ) drawn in multiple sizes with the same pattern shape at their origin (24-P0). ) is shown overlapping and sorting. (C) is a plurality of second patterns (24 ₁ -24 _m ) arranged by overlapping them at the origin (24-P0) located at the lower left corner of the pattern, so the size of the second patterns (24 ₁ -24 _m ) The larger is, the part of the pattern is exposed to the upper right, and (D) is a plurality of second patterns (24 ₁ -24 _m ) arranged by overlapping the origin (26-P0) located at the bottom center, so that the second pattern The larger the size of the field (24 ₁ -24 _m ), the more part of the pattern is exposed upward. In this way, the alignment of the patterns varies depending on the location of the origin.

In addition, a function may be provided that allows the user to select adjacent points in the pattern image and connect the selected points to form a segment of the pattern. If the side is a curve, a predetermined ratio to obtain the coordinates of the point (curve control point) constituting the curve can also be calculated (the predetermined ratio is defined as the ratio A/L in the explanation related to equation (1) described later. and ratio B/L. For further details, please refer to the explanation of equation (1) below).

Additionally, a function may be provided where the user connects a plurality of sides displayed in the pattern image to define the outline, cutting line, seam line, and/or internal line of the pattern. Information about the constructed sides and various lines defined by connecting the sides (i.e., coordinate information) can be stored in a data storage. Preferably, this information can be stored in a database format so that users can conveniently and systematically manage it.

The automatic parametric patterning computer program may include a function that supports all functions related to pattern data management mentioned above to be performed manually by the user. However, when the automatic parametric patterning computer program displays patterns of the same shape graded into multiple clothing sizes on the display of a computer device, it is difficult to align the origin points specified for the patterns of the same shape so that they are all at the same position. It can be configured to perform automatically. In other words, the user does not move and sort.

The read pattern image file (e.g. CAD file) is converted into internal pattern data, and information (e.g. key point information, Origin information, outline/cutting line/seam line/internal line information, etc.), deleted and/or changed information, etc. may be stored in a data storage so that they can be managed together in association with the pattern data.

Next, the automatic parametric patterning computer program allows the user to determine the type of dimensions used when drafting each garment pattern (e.g., chest circumference, waist circumference, hip circumference, sleeve length, hip length, coat length, height, etc.) It can provide the ability to store values in a data storage in a form that can be systematically managed. The types and values of the body dimensions can be stored, for example, in a database. For this purpose, an automatic parametric patterning computer program can provide 'size parameter management capability' (step S20). Figure 5 is a flowchart showing the detailed procedure of step S20.

Referring to Figures 1 and 5, the size parameter management function may include a function of receiving the type and value of the body size used when drafting the pattern of each clothing size and managing it in a database (step S210). For example, the types and values of body measurements used when drawing a size 55 pattern for a women's skirt may be waist circumference 70cm, hip circumference 91cm, hip length 19cm, skirt length 72cm, etc.

The size parameter management function can provide a function for managing a list of variables for variable learning and a formula that defines the relationship between the dimensions used in clothing pattern drafting and anthropometric measurements (step S220). For example, a dimension (e.g., side length) that is a value used when drafting a pattern but is not a value obtained from anthropometric measurements can be defined as a relational expression that includes an anthropometric measurement value (e.g., height) as a variable. For example, the side length used in pattern drafting can be defined as 'side length = height x 0.3' using height, which is a human body measurement.

Additionally, the size parameter management function may include a function to manage the influence of dimensions used in clothing pattern drafting on the coordinates of key points. Using this function, the user can edit the relationship between variables and anthropometric measurements (step S230). For example, the user can also define how certain dimensions are related to the x and/or y coordinates of key points. For example, when a vertex (main point) is moved along the x-axis and/or y-axis, the You can set a definition for the direction of movement (the amount of movement is not related to height, but only to waist circumference). For example, in the case of a top pattern, the The amount of movement in the x-axis direction of vertex P1 and the degree of correlation between height, waist circumference, and chest circumference can be set to 0, 1, and 0.3, respectively.

The types and values of dimensions used when drawing patterns of each size can be obtained from the person who drew the pattern directly. Using these functions, users can store the type values of dimensions used when drafting each pattern in a form that can be systematically managed, for example, by converting it into a database and storing it in a data storage. In other words, the type and value of the dimensions used when drafting the DXF file of each size pattern and the pattern of the corresponding size can be matched to each other in the database so that it becomes interconnected information.

Next, the automatic parametric patterning computer program uses training data that labels the body dimensions used when drawing each pattern and the coordinates of key points on the pattern as input and output, respectively, and creates one artificial neural network for each key point. Learning can be performed by specifying (step S30).

Figure 6 is a flowchart showing the detailed procedure of step S30.

Referring to Figure 6, the dimensions (e.g., chest circumference, waist circumference, sleeve length, etc.) used to draw each of the plurality of patterns graded into a plurality of sizes (e.g., top sizes 95, 100, 105, 110, etc.) You can prepare labeled learning data with values as input and coordinates of key points with the same name (in clothing patterns graded in multiple sizes, key points at the same location are given the same name) as output. The automatic parametric patterning computer program can perform learning based on an artificial neural network using the labeled learning data prepared in this way (step S310).

For artificial neural network learning, weights can be set for each variable (step S320). An artificial neural network can be composed of three layers: input layer, hidden layer, and output layer, and the number of neurons in each layer can be adjusted. This artificial neural network-based learning can be performed for each key point.

Through such artificial neural network-based learning, a corresponding artificial neural network model (i.e., a pair of models that output x-coordinate values and y-coordinate values, respectively) can be created for each key point (step S330). Each artificial neural network model learned in this way can be configured as part of an automatic parametric patterning computer program. When the dimensions used to draw the corresponding clothing pattern are input into the input layer of each learned artificial neural network model, the coordinates of key points corresponding to the corresponding artificial neural network model can be output.

Once the learned artificial neural network models for each key point of each clothing pattern are secured through artificial neural network-based learning, the learned artificial neural network models that generate the coordinates of the key points required for drafting each clothing pattern are collected and responded to the pattern. A parametric pattern can be created (step S40).

In one embodiment, a set of artificial neural network models that generate the coordinates of each key point required for pattern drafting can generate a parametric pattern. Here, parametric pattern refers to a method of drawing a pattern using several parameters (body dimensions). According to one prior art, a parametric pattern was created by formalizing the coordinates of each point constituting the pattern, whereas in the method proposed in the embodiment of the present invention, the user has the desired human body size for a specific type of pattern (i.e. If you input a person's measurements (chest circumference, height, waist circumference, hip circumference, etc.), you can use an artificial neural network that has been trained to output the coordinates of the points that make up the pattern. When there are k key points required for pattern drafting (however, k is a natural number of 3 or more), a parametric pattern requires k artificial neural networks, so a parametric pattern can be created using a set of these artificial neural networks. Therefore, the entity of the parametric pattern can be an artificial neural network model with as many main points. When each artificial neural network model inputs body dimensions (parameters), it can output the coordinates of each point that defines the shape of the pattern based on the learned content.

In order to automatically generate such parametric patterns, an automatic parametric patterning computer program can provide a function to collect data from a trained artificial neural network model and create parametric pattern data. Each parametric pattern, which can be comprised of a set of artificial neural network models, can be saved in script form. Parametric patterns in the form of scripts can be configured so that the corresponding artificial neural network model data can be read in existing parametric pattern design software.

An example of parametric pattern data stored in script form is illustrated in FIG. 7.

In an exemplary embodiment, referring to FIG. 7, the parametric pattern script includes definitions of human body dimensions for each pattern, definitions of pattern drafting variables, basic information of the pattern, and key points corresponding to the artificial neural network model. It may include definitions of points, subpoints and lines determined by the positions of main points, etc. The parametric pattern script may include information for each pattern repeatedly as many times as the total number of patterns.

In the parametric pattern script, the definition of the dimensions (size) for each pattern is information about the number of sizes (for example, if there are 5 sizes of men's tops: 95, 100, 105, 110, and 115, the number of sizes is 5. ) and information on human body dimensions (height, waist circumference, chest circumference, etc.) defined for each defined overall size. Definitions of human body dimensions can be included as many times as there are sizes.

The definition of the variables for pattern drafting may include information about the number of variables for pattern drafting and a definition for each of the total defined variables. Definitions of variables for pattern drafting can be included repeatedly as many variables as there are.

The basic information of the pattern may include information on the number of patterns, and basic pattern information regarding the width, height, coordinates of the origin, etc. for each of the entire patterns. The definition of the artificial neural network points may include information about the neural network definition for the x coordinate and the neural network definition for the y coordinate for each of the main points. Definitions of these artificial neural network points can be included repeatedly as many as the number of key points. Since subpoints and lines can be determined by the location of the main point, definitions of subpoints and lines may include information about each of such subpoints and each of the entire lines.

By inputting the desired body dimensions into the artificial neural network models corresponding to the main points in each parametric pattern of a given clothing, the coordinates of new points are output, and this can be used to create a clothing pattern that fits the body size (S50 step).

In step S50, the pre-secured artificial neural network model can output the coordinates of a point that defines the shape of the pattern based on the learned content by inputting body dimensions (parameters). An automatic parametric patterning computer program uses the artificial neural network models in the parametric pattern to output coordinates of new points when a user (or consumer) inputs body measurements for pattern design, and uses these to use the user's (or consumer's) body measurements. You can create a pattern that fits your dimensions. As explained earlier, parametric patterns are composed of artificial neural network models with a number of key points. Therefore, if you input the body dimensions of the user (or consumer) here, each artificial neural network model can output the coordinates of key points, and using the coordinates of key points obtained in this way, you can create a pattern that fits the consumer's dimensions. .

To be more specific, the user (or consumer) executes the pattern creation function provided by the automatic parametric patterning computer program to measure the desired human body size (i.e., chest circumference, height, waist circumference, hip circumference, etc. as human dimensions). ), these values can be converted to dimensions used in pattern drafting using the formula defined in the 'Size Parameter Management Function' described above. And by inputting the converted dimensions into the artificial neural network model created to obtain the coordinates of the main point, the coordinates of the main point can be obtained.

In order to draw a pattern, in addition to the main points, it is necessary to obtain the coordinates of curve control points. Curve control points are points between two adjacent key points and can define a curve connecting the two key points. When two adjacent major points are connected by a straight line, there are no curved control points between the two major points. The outline of the pattern can be drawn by connecting all the obtained key points and curve control points.

Figure 8 schematically shows the definition of curve control points.

Referring to FIG. 8, the coordinates of the curve control point can be obtained using the relative positions of the main points. Since the curve control points are defined according to the same principle as schematically shown in FIG. 8, the shape of the curve can be similarly transformed even if the position of the main point is changed. That is, when the position of one of the two key points (KP1, KP2), that is, the first key point (KP1), changes, the curve control point between the first key point (KP1) and the second key point (KP2) The position of (CP1) must also be changed, and the position of the changed curve control point (CP1') can be determined by the relationship below.

A'=L'×(A/L),

B'=L'×(B/L) ......(One)

Here, L is the initial distance between two adjacent key points (KP1, KP2), and B connects the two key points (KP1, KP2) at the curve control point (CP1) before changing the position of the first key point (KP1). is the length of the first vertical line drawn from the first straight line, and A is the distance from the intersection between the first straight line connecting the two key points (KP1 and KP2) and the first vertical line to the second key point (KP2). B' is the second curve drawn from the second straight line connecting the changed first key point (KP1) and the second key point (KP2) at the new curve control point (CP1') after changing the position of the first key point (KP1). It is the length of a vertical line, and A' is the distance from the intersection between the second straight line and the second vertical line to the second key point (KP2).

As explained previously, main points can be specified in the pattern image included in the DXF file, and in the pattern image, when connecting adjacent points to form the sides of the pattern, if the side is a curve, the first main point before the position change Since the coordinates of (KP1), the coordinates of the second main point (KP2), and the coordinates of the curve control point (CP1) are known, the values of L, A, and B, which are the ratios for calculating the curve control point using these coordinates, are also known. It can be calculated and stored together. Since the coordinates of the first key point (KP1') after the position change are also known, the value of L' can be obtained. By substituting the obtained values of L, A, B, and L' into equation (1), the values of A' and B' can be obtained. Therefore, a new curve control point (CP1) whose position will change in accordance with the change in the position of the main point can be obtained. ') can be obtained. In this way, in one embodiment, if the corresponding artificial neural network models are obtained through artificial neural network-based learning for only the main points of the pattern, the position information to which the curve control points will move as the main points move is obtained through Equation (1). It can be calculated automatically. It has the advantage of not requiring artificial neural network-based learning even for curve control points other than key points.

Once the main points of a pattern and the curve control points between the main points are obtained, the automatic parametric patterning computer program can automatically generate clothing patterns that fit the input body size using the obtained points. . The user (or consumer) can use the above pattern creation function provided by the automatic parametric patterning computer program to create the entire patterns that make up the design target garment in the above manner.

Next, the suitability of the patterns generated using the above method can be evaluated. In other words, clothing patterns automatically generated through the process described above can be displayed by placing them on the consumer's human body model through 3D clothing drape simulation and turning them into clothing by specifying sewing conditions. At this time, the clothing pressure or allowance distribution can be visualized to determine the suitability of the clothing patterns (step S60). The suitability of the pattern can be judged visually through simulated images,

To be more specific, Figure 9 schematically shows a method for calculating clothing pressure.

Referring to Figure 9, the generated patterns are planar. When the flat patterns are sewn into a three-dimensional form, some parts may receive more tension depending on the shape of the human body, and some parts may be in close contact with the human body and some may be slightly distant. Since the part that receives tensile force can be indirectly known from the degree of deformation of the area of the triangular elements that make up the garment, the relative distribution of tensile force on the garment can be known by expressing this in color. A large tensile force can be interpreted as meaning that that part is under pressure, so the tensile force distribution can be thought of as the clothing pressure distribution.

To calculate clothing pressure, first calculate the areas of the triangular elements (a ₁ , a ₂ , a ₃ , ..., a ₆ ) around point P on the planar pattern. Then, when a 3D garment is made, the areas (A ₁ , A ₂ , A ₃ , ..., A ₆ ) of the triangular elements around point p on the 3D garment are calculated. Clothing pressure P _pressure can be calculated using equation (2) below.

......(2)

However, P _strain represents the tensile force, a _i represents the area of the original triangular element before three-dimensional transformation, and A _i represents the area of the triangular element after three-dimensional transformation.

Clothing pressure P _pressure can be estimated by calculating the degree of change in the area of triangular elements. Additionally, the margin can be calculated by finding the shortest distance from a point on the clothing to an adjacent human body.

Automatic parametric patterning computer programs can provide garment fit assessment capabilities. The automatic parametric patterning computer program generates a pattern in a computer device that fits the human body size entered by the user (i.e., chest circumference, height, waist circumference, hip circumference, etc. as human dimensions) and retrieves human body model data. After arriving, you can place the created patterns in space and specify sewing conditions to simulate them as clothing. From the simulation, clothing pressure (P _pressure ) and margin can be calculated. Using the calculated clothing pressure and allowance, the distribution of clothing pressure and allowance can be visualized in the simulated clothing, and through this, pattern suitability can be judged. Depending on the evaluation results, the suitability of the pattern can be improved by adjusting the parameters input to the artificial neural network.

Figure 10 shows an example of visualizing the distribution of clothing pressure and allowance by simulating a drape on a 3D human body model using the calculated clothing pressure and allowance.

Referring to Figure 10, red represents the part where the clothing pressure is lowest, blue represents the portion where the clothing pressure is the highest, and green represents the clothing pressure in the middle of the two. The simulation image shown shows that there is relatively higher garment pressure in areas such as the thighs, hips, calves, and knees.

Next, according to the suitability evaluation results, the artificial neural network model corresponding to the main points can be modified by limiting or relaxing the influence relationship that a specific body size item has on the coordinates of the main points of the pattern (step S70).

By testing how much a dimension item affects the coordinates of a key point, it is possible to determine the influence a specific body size item has on the coordinates of a key point in a pattern. Based on the identified impact relationships, actions can be taken to limit or mitigate the impacts.

Finally, the completed clothing pattern can be created as a file in the industry standard drawing exchange format (DXF) and saved in the data storage so that it can be transmitted to the sewing factory (step S80). The sewing factory can use the received DXF file to cut actual clothing patterns and produce clothing through sewing work.

Figure 11 illustrates the configuration of a computer device for performing a method of automatic parametric patterning of a clothing pattern using an artificial neural network by executing the automatic parametric patterning computer program according to the embodiment of the present invention described above.

Referring to FIG. 11, the computer device 100 includes hardware resources such as a user interface unit 110, a processor unit 120, a memory unit 130, and a data storage unit 140, and trained artificial intelligence corresponding to key points. A neural network model 150 and a database 160 may be provided.

The user interface unit 110 is connected to the processor unit 120 and may support a function that allows a user to input data or view processed results. The processor unit 120 executes the automatic parametric patterning program described above to automatically generate a parametric pattern by performing various tasks shown in the flowchart of FIG. 1, and automatically generates a clothing pattern using the parametric pattern, Various computational processes necessary for suitability evaluation of the generated clothing pattern can be performed. The memory unit 130 may provide a workspace for the processor unit 120 to process such calculations. The data storage 140 includes trained artificial neural network models 150 corresponding to key points of the pattern, and various programs and data (automatic parametric patterning program, parametric pattern, Pattern data, DXF files, etc.) can be saved. The data storage 140 is connected to the processor unit 120, and the processor unit 120 reads necessary programs and data from the data storage 140 through read and write operations or data processed in the data storage 140. can be saved. The computer device 100 having this configuration may be a general-purpose computer device or a computer system manufactured specifically for the present invention.

The method according to the embodiment described above is written as an automatic parametric patterning program that can be performed through various computer means, and can be converted into program command form and recorded on other types of computer-readable media in addition to the data storage 140. there is. The computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc.

The present invention can be used in the field of clothing pattern design.

Although the embodiments have been described with limited drawings as described above, those skilled in the art can make various modifications and changes to the present invention without departing from the spirit and scope of the present invention as set forth in the following patent claims. You will understand that it is possible. Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims described below.

Claims (17)

A method of automatically generating a clothing pattern using a predetermined computer program running on a computer device,
Converting clothing pattern files graded into a plurality of clothing sizes into internal pattern data of the predetermined computer program and storing the information edited and processed by the user together with the pattern data in a data storage in a manageable manner;
Providing a size parameter management function that can store the types and values of dimensions used when drafting each of the clothing pattern files in a data storage in a format that can be systematically managed;
The values of the dimensions used when drafting each of the above clothing pattern files are input, and key points with the same name (however, in clothing patterns graded in multiple sizes, key points at the same location are given the same name) Performing artificial neural network-based learning using labeled learning data with coordinates as output to generate a corresponding artificial neural network model for each key point; and
Artificial neural network-based clothing pattern automatic parameter comprising a step of automatically generating a parametric pattern corresponding to the pattern by collecting artificial neural network models learned to generate coordinates of key points required for drafting each clothing pattern. How to pattern a rig. The method of claim 1, wherein the storing step includes:
For the necessary editing and processing, the user can load the garment pattern file stored in the data storage and perform desired operations, including deleting unnecessary files and deleting unnecessary pieces, auxiliary lines, and dots from the garment pattern image. providing supporting functions;
Selecting key points to be related to human body dimensions from the imported clothing pattern and providing a function to name each key point;
Providing a function that allows the user to determine an origin point that will serve as a standard for aligning clothing pattern images produced with the plurality of clothing dimensions;
Providing a function that allows the user to select adjacent points in a clothing pattern image and connect the selected points to form a segment of the pattern; and
Artificial neural network-based clothing pattern automatic parameters comprising providing a function for the user to define the outline, cutting line, seam line, and/or internal line of the pattern by connecting a plurality of sides displayed in the clothing pattern image. How to pattern a rig. The method of claim 2, wherein the storing step includes a function for calculating a predetermined ratio A/L and B/L to obtain the coordinates of a curve control point constituting the curve, if the transition curve of the pattern is configured ( Here, L is the distance between the adjacent first and second key points, B is the length of the first vertical line drawn from the curve control point to the first straight line connecting the first and second key points, and A is the distance from the intersection between the first straight line and the first vertical line to the second main point). The method of claim 2, wherein the main points include a vertex of the pattern, a point at which the curvature of the curve changes, and a notch point that serves as a standard for sewing. Automatic parametric patterning of clothing patterns based on an artificial neural network. method. The method of claim 2, wherein the storing step is performed automatically so that when patterns of the same shape graded into multiple clothing sizes are displayed on a display of a computer device, the origins specified for the patterns of the same shape are all located at the same position. An artificial neural network-based automatic parametric patterning method for clothing patterns, further comprising providing a sorting function. The method of claim 1, wherein the size parameter management function includes: a function for managing formulas and variables that define the relationship between dimensions used in clothing pattern drafting and anthropometric measurements; Ability to manage the impact of dimensions used in garment pattern drafting on the coordinates of key points; And an artificial neural network-based automatic parametric patterning method for clothing patterns, comprising a function for managing the influence of the dimensions used in clothing pattern drafting on the coordinates of key points. The method of claim 1, wherein the parametric pattern is generated in script form and stored in the data storage. According to claim 7, the script of the parametric pattern includes definitions of human body dimensions for each pattern, definitions of pattern drafting variables, basic information of the pattern, definitions of key points corresponding to the artificial neural network model, and key points. An artificial neural network-based automatic parametric patterning method for clothing patterns, characterized by including definitions of accessory points and lines determined by the positions of the points. The method of claim 1, wherein desired body dimensions are input into artificial neural network models corresponding to key points in each parametric pattern to calculate the coordinates of the key points and the coordinates of curve control points between the key points, An artificial neural network-based automatic parametric patterning method for clothing patterns, further comprising the step of automatically generating a clothing pattern appropriate for body dimensions using the coordinate values of the calculated points. The method of claim 9, wherein the step of automatically generating the clothing pattern includes,
When the user executes the pattern creation function and inputs the desired human body size, converting the input values into dimensions used in pattern drafting using a formula defined in the size parameter management function; inputting the converted dimensions into the artificial neural network models and outputting coordinates of the key points; calculating the coordinates of a curve control point between two adjacent key points using the relative positions of the key points; And an artificial neural network-based automatic parametric patterning method for clothing patterns, comprising the step of automatically generating patterns that fit the input body size using the calculated coordinates of key points and curve control points. The method of claim 10, wherein in the step of automatically generating the clothing pattern, when the position of a first key point, which is one of two adjacent key points, is changed, the coordinates of the position to be changed of the curve control point are expressed by equation A' = L'×(A/L) and the formula B'=L'×(B/L), where L is the distance between adjacent first and second key points , B is the length of the first vertical line drawn from the first straight line connecting the first and second main points at the curve control point, and A is the second main point at the intersection between the first straight line and the first vertical line. is the distance to the point, and B' is the length of the second vertical line drawn from the second straight line connecting the second key point and the changed first key point at the new curve control point after the position of the first key point is changed, A ' is the distance from the intersection between the second straight line and the second vertical line to the second main point. An artificial neural network-based clothing pattern automatic parametric patterning method. The method of claim 9, further comprising: placing automatically generated clothing patterns on a consumer's human body model through 3D clothing drape simulation and specifying sewing conditions to simulate them as clothing; And an artificial neural network-based automatic parametric patterning method for clothing patterns, further comprising calculating clothing pressure and allowance distribution and displaying them on the simulated clothing to evaluate the suitability of the clothing patterns. The method of claim 12, wherein the clothing pressure (P _pressure ) is expressed by the formula (where P _strain is the tensile force, a _i is the area of the original triangular element before 3-dimensional conversion, and A _i represents the area of the triangular element after 3-dimensional conversion). Automatic clothing pattern based on artificial neural network. Parametric patterning method. The method of claim 12, further comprising modifying artificial neural network models corresponding to key points in a manner to limit or alleviate the influence relationship of a specific body size item on the coordinates of key points of the pattern according to the suitability evaluation results. An artificial neural network-based automatic parametric patterning method for clothing patterns. The method of claim 12, further comprising the step of converting the completed clothing pattern into a file in the industry standard drawing exchange format (DXF) and storing it in a data storage so that it can be transmitted to a sewing factory. Parametric patterning method. A computer-executable program stored in a computer-readable recording medium to perform the artificial neural network-based automatic parametric patterning method for clothing patterns according to any one of claims 1 to 15. A computer-readable recording medium on which a computer program for performing the artificial neural network-based automatic parametric patterning method for clothing patterns according to any one of claims 1 to 15 is recorded.