Automatic clothing pattern making method and system

Abstract

The invention relates to the technical field of garment tailoring, in particular to an automatic garment pattern making method and system, and the method comprises the following steps: constructing a garment preliminary pattern based on fabric characteristics; the method comprises the following steps: dynamically simulating a preliminary garment model by collecting accurate human body data; carrying out accurate cutting based on a dynamic simulation result; according to the invention, the dynamic response of human body motion data and cloth is considered through the clothing preliminary pattern dynamic simulation module, the clothing pattern is optimized, the clothing is ensured to keep fit and comfort in dynamic motion, and the pattern is dynamically adjusted through a deep learning algorithm to adapt to the motion requirements and human body form changes of different users; the tailoring path is optimized by adopting deep reinforcement learning, and the intelligent agent can continuously adjust and optimize the tailoring path in actual production through interaction with the environment; the deep learning algorithm and the reinforcement learning algorithm can be adaptively adjusted according to feedback, and the optimization precision is gradually improved.

Description

Technical Field

[0001] The invention relates to the technical field of clothing cutting, and in particular to an automatic clothing plate making method and system. Background Art

[0002] In the traditional clothing production process, the design of cutting paths usually relies on manual experience or traditional path optimization algorithms. However, traditional cutting path planning methods often ignore the physical properties of fabrics, such as fabric elasticity, texture direction, etc., resulting in high waste and unreasonable fabric utilization in the cutting process. In addition, the existing technology mainly focuses on the cutting optimization of static dimensions, and does not fully consider factors such as human body movement and dynamic behavior of fabrics, which makes the cutting efficiency low and wasteful under complex clothing pattern design and diversified fabric selection.

[0003] With the development of artificial intelligence and deep learning technology, more and more studies have attempted to apply intelligent algorithms to cutting path optimization, especially to dynamically optimize the cutting path through deep learning algorithms. However, the existing technology still lacks comprehensive consideration of the physical properties of fabrics and dynamic human behavior, and mostly relies on static data, making it difficult to achieve the optimal cutting effect. In order to overcome this limitation, the present invention proposes a cutting path optimization algorithm that combines fabric physical properties, dynamic motion data and deep learning optimization, aiming to improve fabric utilization, reduce cutting waste, and realize automation and precision of clothing production through intelligent and adaptive path planning. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for automatic pattern making of clothing, comprising the following steps:

[0006] Construct a preliminary clothing pattern based on fabric characteristics, specifically:

[0007] By collecting the physical property data of the fabric and the human body size data, and determining the preliminary clothing pattern based on the collected data, the gradient descent algorithm is used to iteratively calculate the determined preliminary clothing pattern, and finally the threshold comparison result is used to determine whether the iteration update is completed;

[0008] By collecting accurate human body data, the preliminary clothing pattern is dynamically simulated, specifically:

[0009] Based on the scanned data, a set of spatial coordinates of human body parts is constructed. At the same time, a set of pattern parameters in the preliminary clothing pattern is collected, and an interactive model is constructed. The constructed interactive model is limited by dynamic functions, specifically:

[0010] According to the displacement of the parts of the human body and the response of the fabric to changes over time during the movement, the two are combined to form the limitation of the interaction model;

[0011] Accurately tailor based on dynamic simulation results, specifically:

[0012] The objective function for minimizing trimming damage is set, and the objective function is deeply optimized through the deep reinforcement learning algorithm, specifically:

[0013] Based on the position points on the cutting path and the fabric waste corresponding to each point on the cutting path, a state space is constructed. At the same time, an action space is constructed based on the position adjustment of the cutting path points. A reward function is established based on the constructed state space and action space. According to the result of the reward function, it is judged whether the adjustment of the cutting path points is a positive optimization for cutting, thereby achieving precise cutting of clothing.

[0014] As a preferred solution of the automatic pattern making method for clothing of the present invention, the preliminary pattern of clothing determined based on the collected data is specifically as follows:

[0015] Based on the processed data, determine the preliminary clothing pattern, specifically:

[0016]

[0017] Among them, E represents the fabric elasticity data, T represents the fabric thickness data, D represents the human body size data, and D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Indicates the target size data of human body parts, E j represents the elastic coefficient of fabric j, T j represents the thickness data of fabric j, ε represents the error term, which is the waste caused by improper fabric cutting in the production process, λ1 and λ2 represent regularization coefficients, P represents the preliminary pattern parameter of clothing, F(E,T,D) represents the preliminary pattern of clothing, and the preliminary pattern of clothing is determined by adjusting the value of parameter P.

[0018] As a preferred solution of the automatic clothing pattern making method of the present invention, the iterative calculation of the determined preliminary clothing pattern by using the gradient descent algorithm is specifically as follows:

[0019] Initialize the initial clothing pattern parameters P0;

[0020] Perform gradient descent iterations on each item in the preliminary clothing pattern, specifically:

[0021] The first gradient calculation is,

[0022]

[0023] Among them, D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Represents the target size data of human body parts, represents the gradient, and by adjusting the value of the pattern parameter P, the error between the actual size data of the human body part and the target data is minimized;

[0024] The second gradient calculation is,

[0025]

[0026] Among them, E j represents the elastic coefficient of fabric j, T j represents the thickness data of fabric j, λ1 represents the regularization coefficient. By adjusting λ1, the elastic coefficient and thickness of the fabric are positively corrected. Represents the gradient, and by adjusting the value of the pattern parameter P, the performance optimization between the elasticity and thickness of the fabric is ensured to be optimal;

[0027] The error term gradient calculation is:

[0028]

[0029] Among them, λ2 represents the regularization coefficient. By adjusting λ2, the error term is positively corrected. ε represents the error term, which is the waste caused by improper fabric cutting in the production process. Represents the gradient, and by adjusting the value of the pattern parameter P, it ensures that the waste is minimized during the production process.

[0030] As a preferred solution of the automatic clothing pattern making method of the present invention, the judgment of whether the iteration update is completed according to the threshold comparison result is specifically as follows:

[0031] The adjustment of the pattern parameter P is based on If adjustments are made,

[0032] P new =P·▽ P F

[0033]

[0034] Among them, P represents the version parameter, represents the gradient of the initial clothing pattern to the pattern parameter P, P new Indicates the adjusted pattern parameters;

[0035] For the update of pattern parameters, the difference between two preliminary clothing patterns is calculated to determine whether the pattern parameter update is completed. Specifically:

[0036] Set the preliminary garment pattern threshold F T , if the difference between the two preliminary garment patterns satisfies the formula F new (E,T,D)-F old (E,T,D)|≤F T , it means that the current pattern parameters have been updated. At this time, F new (E, T, D) is the preliminary pattern of the garment;

[0037] If the difference between the two preliminary clothing patterns satisfies the formula |F new (E,T,D)-F old (E,T,D)|>F T , it means that the current pattern parameter update is not complete, and the pattern parameter values ​​are readjusted until the formula is satisfied.

[0038] As a preferred solution of the automatic clothing pattern making method of the present invention, the construction of the interaction model is specifically as follows:

[0039] The user generates body shape data through the scanning device, and constructs the collected data into a set of spatial coordinates of human body parts H = [h1,h2,..,h n ], where h i Represents the three-dimensional coordinates of the i-th part of the human body;

[0040] Collect the pattern parameters P in the preliminary clothing pattern = [p1, p2, ..., p m ], where p i Represents the value of the i-th iteration of the template parameter;

[0041] Based on the spatial coordinate data of human body parts and pattern parameters, the interactive model has:

[0042]

[0043] Among them, S(H,P) represents the interaction result, h i represents the three-dimensional coordinates of the ith part of the human body, represents the position of the i-th part of the human body in a static state, n represents the upper limit of the number of human body parts, It represents the distance between the actual position and the ideal position of each part of the human body, which is the degree of fit of the clothing in a static state. α represents the motion correction coefficient, which is used to correct the change correction of the interaction between the clothing and the human body in a moving state. S move (H, P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing during movement.

[0044] As a preferred solution of the automatic clothing pattern making method of the present invention, the interactive model constructed by the dynamic function is defined as follows:

[0045] Setting the dynamic position of the human body is the spatial coordinate of the i-th node of the human body at time t, and the spatial coordinate of the target position of the human body is Then, we can establish the dynamic displacement of human body parts over time, and we have:

[0046]

[0047] in, represents the spatial coordinates of the i-th node of the human body at time t, The spatial coordinates representing the target position of the human body, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t;

[0048] According to the determined displacement of the part, the dynamic function is determined, then,

[0049]

[0050] Among them, S move (H,P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing in motion, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t, n represents the upper limit of the number of human body parts, m represents the upper limit of the number of pattern parameters, Δp j (t) represents the pattern parameter p at time t j The response over time is affected by the dynamic change of fabric elasticity data during the dynamic change process, as follows:

[0051] Based on the stress of the fabric, the response of the pattern parameters to changes over time is determined, then:

[0052] F σ (t) = E j ΔL

[0053]

[0054] Among them, F σ (t) represents the stress of the fabric at time t, which is a direct measurement result. ΔL represents the length change of the fabric in motion. L(t) represents the length of the fabric at time t. L0 represents the length at rest. E j Indicates the elastic coefficient of the fabric under motion, which is used to determine the pattern parameters at time t. represents the pattern parameters at time t, which are determined by the gradient calculation formula in the pattern preliminary model, pj Represents the jth pattern parameter in the pattern parameters.

[0055] As a preferred solution of the automatic pattern making method for clothing of the present invention, the precise cutting based on the dynamic simulation result is specifically as follows:

[0056] Set the objective function, which is to adjust the position P of each clipping path point i , minimize the trimming loss, specifically:

[0057]

[0058] According to the state space S and the action space A, a reward function is established to determine whether the adjusted action is a positive optimization for trimming. Then,

[0059]

[0060] Among them, P i represents the i-th point on the clipping path, P i represents the i+1th point on the clipping path, P′ i represents the adjusted position of the i-th point on the clipping path, P′ i represents the adjusted position of the i+1th point on the cutting path, η represents the cloth cutting loss correction coefficient, C w,i represents the fabric waste of the i-th point on the cutting path, and R(S,A) represents the reward result, which is used to determine whether the adjustment of the cutting point is positive optimization, specifically:

[0061] If the formula R(S,A)≤0 is satisfied, it means that the adjustment of the cutting path points is positive optimization, which reduces fabric waste. The adjustment nodes are recorded until all nodes are adjusted. All cutting nodes are counted, and the final statistical result is the optimal cutting path for the current garment.

[0062] If the formula R(S,A)>0 is satisfied, it means that the adjustment of the cutting path points is not forward optimization. The adjustment node is proposed, the cutting nodes before adjustment are retained, and the node cutting adjustment is judged one by one until all nodes are judged. All cutting nodes after elimination are statistically eliminated, and the final statistical result is the optimal cutting path for the current garment.

[0063] A clothing automatic pattern making system comprises a clothing preliminary pattern determination module, a clothing preliminary pattern dynamic simulation module and a clothing precise tailoring module; the clothing preliminary pattern determination module comprehensively determines the clothing preliminary pattern based on physical property data of fabrics and human body size data; the clothing preliminary pattern dynamic simulation module constructs an interaction model based on precise human body data and pattern parameters, and establishes a dynamic function to limit the constructed interaction model; the clothing precise tailoring module establishes an objective function, and uses a deep reinforcement learning algorithm to deeply optimize the objective function.

[0064] A computer device comprises a memory and a processor, wherein the memory stores a computer program and the processor implements the steps of a method for automatic garment pattern making when executing the computer program.

[0065] A computer-readable storage medium stores a computer program, which implements the steps of a method for automatic garment pattern making when executed by a processor.

[0066] Beneficial effects of the present invention:

[0067] The present invention comprehensively considers the physical properties of the fabric, dynamically adjusts the cutting path, reduces fabric waste, optimizes the cutting path through an intelligent algorithm, and can accurately calculate the optimal position and direction of each cutting path, thereby maximizing the use efficiency of the fabric and significantly reducing the scraps caused by unreasonable cutting path design.

[0068] The clothing preliminary pattern dynamic simulation module takes into account human motion data and the dynamic response of fabrics, optimizes the clothing pattern, ensures that the clothing remains fit and comfortable during dynamic motion, and dynamically adjusts the pattern through deep learning algorithms to adapt to the motion needs and body shape changes of different users.

[0069] Using deep reinforcement learning to optimize cutting paths, the intelligent agent can continuously adjust and optimize cutting paths in actual production through interaction with the environment. The deep learning algorithm can automatically analyze the complex relationship between fabric characteristics and pattern, and generate the optimal cutting path.

[0070] A dynamic feedback mechanism has been introduced. In the actual production process, cutting results and waste data are collected in real time, and the optimization strategy is continuously adjusted through feedback. The deep learning and reinforcement learning algorithms can adaptively adjust according to the feedback and gradually improve the optimization accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0072] Figure 1 The present invention is a schematic diagram of the overall method steps of a method for automatic clothing pattern making. DETAILED DESCRIPTION

[0073] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0074] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0075] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0076] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0077] Meanwhile, in the description of the present invention, it should be noted that the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0078] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0079] Example 1

[0080] Reference Figure 1 , as an embodiment of the present invention, provides a method for automatic pattern making of clothing, comprising the following steps:

[0081] S1: Construct preliminary garment pattern based on fabric properties.

[0082] Specifically, the construction of a preliminary clothing pattern based on fabric characteristics is achieved by establishing a clothing pattern design model, and using factors such as fabric elasticity, thickness, and human body size to construct a corresponding pattern structure. The clothing pattern design model describes the physical deformation of the fabric through a mathematical model, and combines the deformation result with human body size data to generate a preliminary clothing pattern. At the same time, the generated preliminary clothing pattern is dynamically adjusted by formulating an optimization algorithm, which is specifically implemented as follows:

[0083] Obtaining fabric physical property data from fabric samples, including elastic coefficient E and thickness T, and obtaining human body size data D from a standard size library;

[0084] The collected fabric physical property data and human body size data are standardized. The standardized processing is a well-known technology that can be easily associated by people in the field and will not be described in detail in this embodiment. The processed data can ensure that data of different dimensions can be calculated in the same optimization model, thereby completing the generation of subsequent preliminary clothing patterns.

[0085] Based on the processed data, determine the preliminary clothing pattern, specifically:

[0086]

[0087] Among them, E represents the fabric elasticity data, T represents the fabric thickness data, D represents the human body size data, and D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Indicates the target size data of human body parts, E j represents the elastic coefficient of fabric j, T jrepresents the thickness data of fabric j, ε represents the error term, which is the waste part caused by improper fabric cutting in the production process, λ1 and λ2 represent regularization coefficients. By adjusting λ1, the elastic coefficient and thickness of the fabric can be positively corrected. By adjusting λ2, the error term can be positively corrected. P represents the preliminary pattern parameter of clothing. F(E,T,D) represents the preliminary pattern of clothing. The preliminary pattern of clothing is determined by adjusting the value of parameter P.

[0088] It should be noted that the minimum value of the relationship function is iteratively calculated using the gradient descent algorithm, and the specific calculation is as follows:

[0089] Initialize the initial clothing pattern parameters P0;

[0090] Perform gradient descent iterations on each item in the preliminary clothing pattern, specifically:

[0091] The first gradient calculation is,

[0092]

[0093] Among them, D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Represents the target size data of human body parts, represents the gradient, and by adjusting the value of the pattern parameter P, the error between the actual size data of the human body part and the target data is minimized;

[0094] The second gradient calculation is,

[0095]

[0096] Among them, E j represents the elastic coefficient of fabric j, T j represents the thickness data of fabric j, λ1 represents the regularization coefficient. By adjusting λ1, the elastic coefficient and thickness of the fabric can be positively corrected. Represents the gradient, and by adjusting the value of the pattern parameter P, the performance optimization between the elasticity and thickness of the fabric is ensured to be optimal;

[0097] The error term gradient calculation is:

[0098]

[0099] Among them, λ2 represents the regularization coefficient. By adjusting λ2, the error term can be positively corrected. ε represents the error term, which is the waste caused by improper fabric cutting in the production process. PRepresents the gradient, and by adjusting the value of the pattern parameter P, it ensures that the waste is minimized during the production process.

[0100] It should be noted that the adjustment of the pattern parameter P is based on ▽ P If F is adjusted, then we have

[0101]

[0102] Among them, P represents the version parameter, represents the gradient of the initial clothing pattern to the pattern parameter P, P new Indicates the adjusted pattern parameters;

[0103] For the update of pattern parameters, the difference between two preliminary clothing patterns is calculated to determine whether the pattern parameter update is completed. Specifically:

[0104] Set the preliminary garment pattern threshold F T , if the difference between the two preliminary garment patterns satisfies the formula F new (E,T,D)-F old (E,T,D)|≤F T , it means that the current pattern parameters have been updated. At this time, F new (E, T, D) is the preliminary pattern of the garment;

[0105] If the difference between the two preliminary clothing patterns satisfies the formula |F new (E,T,D)-F old (E,T,D)|>F T , it means that the current pattern parameter update is not complete, and the pattern parameter values ​​are readjusted until the formula is satisfied.

[0106] S2: Perform dynamic simulation based on the determined preliminary clothing pattern.

[0107] Specifically, the human body interactive matching based on the determined preliminary clothing pattern is to collect accurate human body data, dynamically simulate the preliminary clothing pattern, and achieve accurate matching with the preliminary clothing pattern, which is specifically implemented as follows:

[0108] The user generates body shape data through the scanning device, and constructs the collected data into a set of spatial coordinates of human body parts H = [h1,h2,..,h n ], where h i Represents the three-dimensional coordinates of the i-th part of the human body;

[0109] Collect the pattern parameters P in the preliminary clothing pattern = [p1, p2, ..., p m ], where p i Represents the value of the i-th iteration of the template parameter;

[0110] Based on the spatial coordinate data of human body parts and pattern parameters, the interactive model has:

[0111]

[0112] Among them, S(H,P) represents the interaction result, h i represents the three-dimensional coordinates of the ith part of the human body, represents the position of the i-th part of the human body in a static state, n represents the upper limit of the number of human body parts, It represents the distance between the actual position and the ideal position of each part of the human body, which is the degree of fit of the clothing in a static state. α represents the motion correction coefficient, which is used to correct the change correction of the interaction between the clothing and the human body in a moving state. S move (H,P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing during movement, as follows:

[0113] According to the displacement of the joints and limbs of the human body, the deformation effect of the dynamic model clothing in motion is completed, specifically:

[0114] Setting the dynamic position of the human body is the spatial coordinate of the i-th node of the human body at time t, and the spatial coordinate of the human body at the target position (static position) is Then, we can establish the dynamic displacement of human body parts over time, and we have:

[0115]

[0116] in, represents the spatial coordinates of the i-th node of the human body at time t, Indicates the spatial coordinates of the target position (stationary position) of the human body, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t;

[0117] According to the determined displacement of the part, the dynamic function is determined, then,

[0118]

[0119] Among them, S move (H,P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing in motion, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t, n represents the upper limit of the number of human body parts, Δp j (t) represents the pattern parameter p at time t j The response over time is affected by the dynamic change of fabric elasticity data during the dynamic change process, as follows:

[0120] Based on the stress of the fabric, the response of the pattern parameters to changes over time is determined, then:

[0121] F σ (t) = E j ΔL

[0122]

[0123] Among them, F σ (t) represents the stress of the fabric at time t, which is a direct measurement result. ΔL represents the length change of the fabric in motion. L(t) represents the length of the fabric at time t. L0 represents the length at rest. E j Indicates the elastic coefficient of the fabric under motion, which is used to determine the pattern parameters at time t. Represents the pattern parameters at time t, which are determined by the gradient calculation formula in the pattern preliminary model.

[0124] It should be noted that through the dynamic model, the pattern can be adjusted in real time based on human motion data, which not only optimizes the static size, but also improves the comfort and fit of the clothing during dynamic activities.

[0125] S3: Accurate cutting based on dynamic simulation results of preliminary garment pattern.

[0126] Specifically, the precise cutting based on the dynamic simulation results of the preliminary clothing pattern is to optimize the fabric cutting in depth according to the dynamic simulation results by using a deep learning algorithm to achieve precise fabric cutting, which is specifically implemented as follows:

[0127] Set a starting point P0 of a cloth clipping path and each clipping point P on the path i =[P1,P2,...,P n ];

[0128] Set the objective function, which is to adjust the position P of each clipping path point i , minimize the trimming loss, specifically:

[0129]

[0130] Among them, P i represents the i-th point on the clipping path, P i represents the i+1th point on the clipping path, n represents the total number of points on the clipping path, ||P i -P i+1 || represents the distance between two adjacent points on the cutting path, η represents the cloth cutting loss correction coefficient, C w,i Indicates the fabric waste at the i-th point on the cutting path;

[0131] The deep reinforcement learning algorithm is used to deeply optimize the objective function, as follows:

[0132] Define the state space S, and satisfy the formula S = [P1, P2, .., P n ,C w,1 ,C w,2 ,...,C w,n ], where P i represents the i-th point on the clipping path, C w,i Indicates the fabric waste at the i-th point on the cutting path;

[0133] At the same time, define the action space A and satisfy the formula A=[P′1,P′2,...,P′ n ], where P′ i Indicates the position adjustment of the clipping path point;

[0134] According to the state space and action space, a reward function is established to determine whether the adjusted action is a positive optimization for trimming. Then,

[0135] R(S,A)=(||P i -P i+1 ||+η·C w,i )-(||P′ i -P′ i+1 ||+η·C w,i )

[0136] Among them, P i represents the i-th point on the clipping path, P i represents the i+1th point on the clipping path, P′ i represents the adjusted position of the i-th point on the clipping path, P′ i represents the adjusted position of the i+1th point on the cutting path, η represents the cloth cutting loss correction coefficient, C w,i represents the fabric waste of the i-th point on the cutting path, and R(S,A) represents the reward result, which is used to determine whether the adjustment of the cutting point is positive optimization, specifically:

[0137] If the formula R(S,A)≤0 is satisfied, it means that the adjustment of the cutting path points is positive optimization, which can reduce fabric waste. The adjustment nodes are recorded until all nodes are adjusted. All cutting nodes are counted, and the final statistical result is the optimal cutting path for the current garment.

[0138] If the formula R(S,A)>0 is satisfied, it means that the adjustment of the cutting path points is not forward optimization. The adjustment node is proposed, the cutting nodes before adjustment are retained, and the node cutting adjustment is judged one by one until all nodes are judged. All cutting nodes after elimination are statistically eliminated, and the final statistical result is the optimal cutting path for the current garment.

[0139] Example 2

[0140] A second embodiment of the present invention provides a clothing automatic pattern making system, comprising a clothing preliminary pattern determination module, a clothing preliminary pattern dynamic simulation module and a clothing precise cutting module;

[0141] Specifically, the clothing preliminary pattern determination module comprehensively determines the clothing preliminary pattern based on the physical property data of the fabric and the human body size data; the clothing preliminary pattern dynamic simulation module constructs an interactive model based on the precise human body data and pattern parameters, and establishes a dynamic function to limit the constructed interactive model; the clothing precise tailoring module establishes an objective function, and uses a deep reinforcement learning algorithm to deeply optimize the objective function.

[0142] Furthermore, the clothing preliminary pattern determination module comprehensively determines the preliminary clothing pattern based on the physical property data of the fabric and the human body size data; it is the starting link of the entire clothing automatic pattern making system, laying the foundation for subsequent dynamic simulation and precise cutting; specifically:

[0143] According to different fabric characteristics and human body data, data standardization and normalization are performed to facilitate subsequent processing and calculation. Rule-based algorithms (such as geometric modeling and template-based design) are used in combination with the elastic coefficient of the fabric and human body size data to infer the preliminary clothing pattern. At the same time, combined with the target human body size, the pattern of key parts of the clothing (such as shoulder width, chest circumference, sleeve length, etc.) is adjusted to make the pattern match the human body size as much as possible.

[0144] It should be noted that the clothing preliminary pattern determination module provides basic data (such as preliminary pattern size and fabric characteristics) for dynamic simulation. Subsequent dynamic simulation will further optimize the pattern based on this data and consider adaptability during exercise;

[0145] The clothing preliminary pattern dynamic simulation module is based on the precise data of the human body and the preliminary pattern parameters, constructs an interactive model and establishes a dynamic function to limit it. Through dynamic simulation, the performance of clothing during exercise can be evaluated, the pattern can be optimized, and the comfort and fit of clothing can be ensured. Specifically:

[0146] Based on accurate human body data and preliminary patterns, an interaction model between the human body and clothing is established. This model can reflect the adaptation relationship between different parts of the human body and clothing in static and dynamic conditions.

[0147] Design dynamic functions to simulate the dynamic changes of the human body in different motion states (such as walking, bending, raising hands, etc.), and calculate the strain and stretching of clothing during movement. The setting of dynamic functions takes into account factors such as the elasticity and mechanical properties of the fabric;

[0148] Based on motion simulation feedback, adjust the local shape of the garment (such as shoulders, chest, waist, etc.) to ensure that the garment remains comfortable and fits well in various motion states;

[0149] It should be noted that the dynamic simulation module of the preliminary clothing pattern provides accurate pattern data through dynamic simulation and optimized pattern. The optimized pattern will be used for actual cutting, thereby improving the efficiency of fabric use and reducing cutting waste.

[0150] The clothing precision cutting module uses a deep reinforcement learning algorithm to deeply optimize the objective function and accurately cut the clothing pattern to ensure the maximum utilization of fabric and the minimum waste. By intelligently optimizing the cutting path, precise clothing cutting is finally achieved, specifically:

[0151] Receive optimized pattern data from the preliminary pattern dynamic simulation module of the garment, including the adjusted pattern and motion adaptability, and define the objective function of cutting optimization, the main goal of which is to minimize cutting waste and optimize fabric utilization;

[0152] Apply deep reinforcement learning algorithms to deeply optimize the cutting path through reinforcement learning. Based on the optimized pattern and fabric characteristics, the intelligent algorithm automatically generates the optimal cutting path. The path is dynamically adjusted to adapt to the elasticity and texture direction of the fabric, thereby reducing waste and improving production efficiency.

[0153] It should be noted that the clothing preliminary pattern determination module provides preliminary pattern data for the subsequent dynamic simulation and precise cutting modules, ensuring that the pattern meets the actual size requirements of the human body from the beginning. The clothing preliminary pattern dynamic simulation module, based on the optimization of the preliminary pattern, simulates the interaction between the human body and the clothing in different dynamic states to ensure that the clothing can provide comfort in a dynamic environment and provide optimized pattern parameters. The clothing precise cutting module optimizes the cutting path through deep reinforcement learning, and automatically generates the optimal cutting path based on the dynamically optimized pattern data and fabric characteristics, thereby maximizing fabric utilization and reducing waste.

[0154] Furthermore, if the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0155] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0156] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0157] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention) may not be described.

[0158] It will be appreciated that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but will be a routine task of design, fabrication, and production for those of ordinary skill having the benefit of this disclosure without undue experimentation.

[0159] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for automatic pattern making of clothing, characterized in that: The following steps are included: Construct a preliminary clothing pattern based on fabric characteristics, specifically: By collecting the physical property data of the fabric and the human body size data, and determining the preliminary clothing pattern based on the collected data, the gradient descent algorithm is used to iteratively calculate the determined preliminary clothing pattern, and finally the threshold comparison result is used to determine whether the iteration update is completed; By collecting accurate human body data, the preliminary clothing pattern is dynamically simulated, specifically: Based on the scanned data, a set of spatial coordinates of human body parts is constructed. At the same time, a set of pattern parameters in the preliminary clothing pattern is collected, and an interactive model is constructed. The constructed interactive model is limited by dynamic functions, specifically: According to the displacement of the parts of the human body and the response of the fabric to changes over time during the movement, the two are combined to form the limitation of the interaction model; Accurately tailor based on dynamic simulation results, specifically: The objective function for minimizing trimming damage is set, and the objective function is deeply optimized through the deep reinforcement learning algorithm, specifically: Based on the position points on the cutting path and the fabric waste corresponding to each point on the cutting path, a state space is constructed. At the same time, an action space is constructed based on the position adjustment of the cutting path points. A reward function is established based on the constructed state space and action space. According to the result of the reward function, it is judged whether the adjustment of the cutting path points is a positive optimization for cutting, thereby achieving precise cutting of clothing.

2. A method for automatic pattern making of clothing as claimed in claim 1, characterized in that: Determining the preliminary pattern of the garment based on the collected data is as follows: Based on the processed data, determine the preliminary clothing pattern, specifically: Among them, E represents the fabric elasticity data, T represents the fabric thickness data, D represents the human body size data, and D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Indicates the target size data of human body parts, E j represents the elastic coefficient of fabric j, T j represents the thickness data of fabric j, ε represents the error term, which is the waste caused by improper fabric cutting in the production process, λ1 and λ2 represent regularization coefficients, P represents the preliminary pattern parameter of clothing, F(E,T,D) represents the preliminary pattern of clothing, and the preliminary pattern of clothing is determined by adjusting the value of parameter P.

3. A method for automatic pattern making of clothing as claimed in claim 2, characterized in that: The iterative calculation of the determined preliminary clothing pattern by using the gradient descent algorithm is specifically as follows: Initialize the initial clothing pattern parameters P0; Perform gradient descent iterations on each item in the preliminary clothing pattern, specifically: The first gradient calculation is, Among them, D i Indicates the actual size data of human body parts, including chest circumference, waist circumference and hip circumference. i represents the serial number of the human body part, and n represents the upper limit of the number of human body parts. Represents the target size data of human body parts, ▽ P represents the gradient, and by adjusting the value of the pattern parameter P, the error between the actual size data of the human body part and the target data is minimized; The second gradient calculation is, Among them, E j represents the elastic coefficient of fabric j, T j represents the thickness data of fabric j, λ1 represents the regularization coefficient. By adjusting λ1, the elastic coefficient and thickness of the fabric are positively corrected. P Represents the gradient, and by adjusting the value of the pattern parameter P, the performance optimization between the elasticity and thickness of the fabric is ensured to be optimal; The error term gradient calculation is: ▽ P (λ2·e)=λ2·(▽ P e) Among them, λ2 represents the regularization coefficient. By adjusting λ2, the error term is positively corrected. ε represents the error term, which is the waste caused by improper fabric cutting in the production process. P Represents the gradient, and by adjusting the value of the pattern parameter P, it ensures that the waste is minimized during the production process.

4. A method for automatic pattern making of clothing as claimed in claim 3, characterized in that: The determination of whether the iteration update is completed according to the threshold comparison result is specifically as follows: The adjustment of the pattern parameter P is based on ▽ P If F is adjusted, then we have P new =P·▽ P F Among them, P represents the pattern parameters, ▽ P F represents the gradient of the initial clothing pattern to the pattern parameter P, P new Indicates the adjusted pattern parameters; For the update of pattern parameters, the difference between two preliminary clothing patterns is calculated to determine whether the pattern parameter update is completed. Specifically: Set the preliminary garment pattern threshold F T , if the thresholds of the two preliminary clothing patterns satisfy the formula F new (E,T,D)-F old (E,T,D)|≤F T , it means that the current pattern parameters have been updated. At this time, F new (E, T, D) is the preliminary pattern of the garment; If the difference between the two preliminary clothing patterns satisfies the formula |F new (E,T,D)-F old (E,T,D)|>F T , it means that the current pattern parameter update is not complete, and the pattern parameter values ​​are readjusted until the formula is satisfied.

5. A method for automatic pattern making of clothing as claimed in claim 4, characterized in that: The construction of the interaction model is as follows: The user generates body shape data through the scanning device, and constructs the collected data into a set of spatial coordinates of human body parts H = [h1,h2,..,h n ], where h i Represents the three-dimensional coordinates of the i-th part of the human body; Collect the pattern parameters P in the preliminary clothing pattern = [p1, p2, ..., p m ], where p i Represents the value of the i-th iteration of the template parameter; Based on the spatial coordinate data of human body parts and pattern parameters, the interactive model has: Among them, S(H,P) represents the interaction result, h i represents the three-dimensional coordinates of the ith part of the human body, represents the position of the i-th part of the human body in a static state, n represents the upper limit of the number of human body parts, It represents the distance between the actual position and the ideal position of each part of the human body, which is the degree of fit of the clothing in a static state. α represents the motion correction coefficient, which is used to correct the change correction of the interaction between the clothing and the human body in a moving state. S move (H, P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing during movement.

6. A method for automatic pattern making of clothing as claimed in claim 5, characterized in that: The interactive model constructed by the dynamic function is limited as follows: Setting the dynamic position of the human body is the spatial coordinate of the i-th node of the human body at time t, and the spatial coordinate of the target position of the human body is Then, we can establish the dynamic displacement of human body parts over time, and we have: in, represents the spatial coordinates of the i-th node of the human body at time t, The spatial coordinates representing the target position of the human body, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t; According to the determined displacement of the part, the dynamic function is determined, then, Among them, S move (H,P) represents a dynamic function, which is used to dynamically simulate the deformation effect of clothing in motion, Δh i (t) represents the dynamic displacement of the ith part of the human body at time t, n represents the upper limit of the number of human body parts, m represents the upper limit of the number of pattern parameters, Δp j (t) represents the pattern parameter p at time t j The response over time is affected by the dynamic change of fabric elasticity data during the dynamic change process, as follows: Based on the stress of the fabric, the response of the pattern parameters to changes over time is determined, then: F σ (t)=E j ·ΔL Among them, F σ (t) represents the stress of the fabric at time t, which is a direct measurement result. ΔL represents the length change of the fabric in motion. L(t) represents the length of the fabric at time t. L0 represents the length at rest. E j Indicates the elastic coefficient of the fabric under motion, which is used to determine the pattern parameters at time t. represents the pattern parameters at time t, which are determined by the gradient calculation formula in the pattern preliminary model, p j Represents the jth pattern parameter in the pattern parameters.

7. A method for automatic pattern making of clothing as claimed in claim 6, characterized in that: The precise tailoring based on the dynamic simulation results is specifically as follows: Set the objective function, which is to adjust the position P of each clipping path point i , minimize the trimming loss, specifically: According to the state space S and the action space A, a reward function is established to determine whether the adjusted action is a positive optimization for trimming. Then, R(S,A)(||P i -P i+1 ||+η·C w,i )-(||P i ′-P′ i+1 ||+η·C w,i ) Among them, P i represents the i-th point on the clipping path, P i represents the i+1th point on the clipping path, P i ′ represents the adjusted position of the i-th point on the clipping path, P i ′ represents the adjusted position of the i+1th point on the cutting path, η represents the cloth cutting loss correction coefficient, C w,i represents the fabric waste of the i-th point on the cutting path, and R(S,A) represents the reward result, which is used to determine whether the adjustment of the cutting point is positive optimization, specifically: If the formula R(S,A)≤0 is satisfied, it means that the adjustment of the cutting path points is positive optimization, which reduces fabric waste. The adjustment nodes are recorded until all nodes are adjusted. All cutting nodes are counted, and the final statistical result is the optimal cutting path for the current garment. If the formula R(S,A)>0 is satisfied, it means that the adjustment of the cutting path points is not forward optimization. The adjustment node is proposed, the cutting nodes before adjustment are retained, and the node cutting adjustment is judged one by one until all nodes are judged. All cutting nodes after elimination are statistically eliminated, and the final statistical result is the optimal cutting path for the current garment.

8. A clothing automatic plate making system, applied to a clothing automatic plate making method according to any one of claims 1 to 7, characterized in that: Including, clothing preliminary pattern determination module, clothing preliminary pattern dynamic simulation module and clothing precise cutting module; The clothing preliminary pattern determination module comprehensively determines the clothing preliminary pattern based on the physical property data of the fabric and the human body size data; The clothing preliminary pattern dynamic simulation module constructs an interactive model based on accurate human body data and pattern parameters, and establishes a dynamic function to limit the constructed interactive model; The clothing precision tailoring module establishes an objective function and uses a deep reinforcement learning algorithm to deeply optimize the objective function.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images