Clothing matching recommendation method and system based on big data

Abstract

The invention relates to the technical field of garment matching recommendation, in particular to a garment matching recommendation method and system based on big data. The method comprises the following steps: acquiring clothes browsing historical data and input body type data of a user on an e-commerce platform, and performing clothes type grouping to obtain clothes type time sequence grouping data; on the basis, color and version information of different garment types is extracted, color-version data between the garment types is generated, the matching effect of the user body type and the version of the different garment types is analyzed according to the data, and matching proportion segmentation effect data is obtained. Secondly, optimizing the visual geometric gravity center deviation to obtain optimized geometric gravity center deviation data; and designing a personalized clothing recommendation strategy in combination with geometric gravity center deviation optimization data and matching effect data, thereby realizing accurate clothing matching recommendation. According to the invention, the garment matching recommendation technology is optimized, so that the garment matching recommendation technology is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of clothing matching recommendation, and in particular to a clothing matching recommendation method and system based on big data. Background Art

[0002] In the field of clothing, personalized recommendations have become an important means to improve user shopping experience and enhance platform competitiveness. Previous clothing recommendation methods mainly rely on user browsing history, purchase history and other data, but these methods often ignore the user's body shape differences, personalized preferences, and clothing style, color and other multi-dimensional matching factors. With the continuous development of big data and artificial intelligence technology, clothing matching recommendation methods based on big data have gradually attracted attention. By analyzing massive user behavior data, body shape data, clothing color, style and other multi-dimensional information, it provides users with more accurate and personalized clothing matching recommendations, thereby improving user satisfaction and purchase conversion rate. The application of big data technology makes clothing recommendation not only limited to simple recommendation algorithms, but also can dig out users' potential personalized needs and preferences by deeply analyzing users' historical browsing data, purchase data and social media behaviors. For example, by analyzing users' past browsing records, it is possible to judge their preference for a certain style, color or material, so as to provide users with clothing recommendations that suit their tastes. However, a traditional clothing matching recommendation method based on big data has the problem of inaccurate recognition of user body shape characteristics, resulting in large errors in clothing matching recommendations. Summary of the invention

[0003] Based on this, it is necessary to provide a clothing matching recommendation method and system based on big data to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a clothing matching recommendation method based on big data is provided, the method comprising the following steps: Step S1: obtaining a user clothing history browsing data set and user input body shape data in an e-commerce platform; grouping the user clothing history browsing data set by clothing categories to obtain clothing category time series grouping data; Step S2: performing color-pattern extraction between different clothing categories on the clothing category time series grouping data to obtain clothing category color-pattern data between different clothing categories; performing pattern matching ratio segmentation effect analysis between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain pattern matching ratio segmentation effect data; performing visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; Step S3: Design a clothing recommendation strategy based on the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0005] Preferably, step S1 comprises the following steps: Step S11: Acquire a user's clothing browsing history data set and user input body shape data in the e-commerce platform; Step S12: arranging the average browsing time sequence of the user's clothing history browsing data set to obtain clothing browsing average time sequence arrangement data; Step S13: grouping the user's clothing history browsing data set by clothing category according to the clothing browsing average time series arrangement data to obtain clothing category time series grouping data.

[0006] Preferably, step S2 comprises the following steps: Step S21: extracting the color-pattern of different clothing categories from the clothing category time series grouping data to obtain clothing category color-pattern data of different clothing categories, wherein the clothing category color-pattern data includes the color-pattern of upper body clothing category and the color-pattern of lower body clothing category; Step S22: performing pattern matching ratio segmentation effect analysis between different clothing types on the user input body shape data according to the clothing type color-pattern data, and obtaining pattern matching ratio segmentation effect data; Step S23: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the clothing type color-pattern data to obtain the visual guidance center of gravity data of the matching; Step S24: performing visual geometric center of gravity deviation optimization processing on the matching visual guidance center of gravity data to obtain geometric center of gravity deviation optimization data.

[0007] Preferably, step S22 includes the following steps: Step S221: calculating the upper and lower body ratios of the user's inputted body shape data to obtain the upper and lower body ratios; Step S222: quantifying the limb morphology differences of the user inputted body shape data according to the ratio of the upper and lower body parts to obtain limb morphology difference quantified data; Step S223: Calculating the pattern shaping contour values ​​between different clothing categories on the clothing category color-pattern data to obtain the pattern shaping contour values ​​between different clothing categories; Step S224: adapting the pattern shaping contour value to the clothing pattern looseness according to the quantitative data of the upper and lower body proportions and the limb morphology differences, to obtain pattern functional looseness adaptation data; Step S225: Analyze the pattern matching ratio segmentation effect between different clothing types based on the pattern functional looseness adaptation data and the quantitative data of the upper and lower body proportions and limb shape differences, and obtain pattern matching ratio segmentation effect data.

[0008] Preferably, step S224 includes the following steps: Calculating the circumference of the limbs based on the quantitative data of the limb morphological differences to obtain the circumference of the limbs; wherein the circumference of the limbs includes: the circumference of the wrist, the circumference of the ankle, the circumference of the elbow, and the circumference of the knee; According to the proportion of the upper and lower body and the circumference of the limbs, the spatial differentiation of bone protrusions between different movement postures was analyzed to obtain the spatial differentiation data of bone protrusions between different movement postures; Based on the spatial differentiation data of bone protrusion, the space reservation amount of the limbs is calculated for the pattern shaping contour value to obtain the space reservation amount of the limbs; The looseness of the clothing pattern is adapted based on the reserved space for the limbs to obtain the functional looseness adaptation data of the pattern.

[0009] Preferably, step S23 includes the following steps: Step S231: extracting colors and patterns between different clothing categories from the clothing category color-pattern data to obtain clothing color data between different clothing categories and clothing pattern data between different clothing categories; Step S232: performing color visual compactness analysis on clothing color data of different clothing types to obtain color visual compactness data; Step S233: quantifying the hue interlaced spatial hierarchy according to the color visual compactness data and the clothing pattern data to obtain hue interlaced spatial hierarchy data; Step S234: identifying the law of visual center of gravity transfer on the pattern matching ratio segmentation effect data based on the color tone interlaced spatial hierarchy data to obtain the law of visual center of gravity transfer data; Step S235: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the visual center of gravity transfer rule data to obtain matching visual guidance center of gravity data.

[0010] Preferably, step S234 includes the following steps: Perform visual landing point analysis on the hue interlaced spatial hierarchy data to obtain hue interlaced visual landing point data; Based on the color tone staggered visual landing point data, the clothing pattern data and the pattern matching ratio segmentation effect data are subjected to the visual expansion perception of the stripe interlaced arc, and the visual expansion perception data of the stripe arc between the matching ratio segmentation is obtained; Performing perception direction guidance analysis on the stripe curvature visual expansion perception data to obtain stripe curvature perception direction guidance data; Based on the stripe arc visual expansion perception data and the stripe arc perception direction guidance data, the visual center of gravity transfer law is identified to obtain the visual center of gravity transfer law data.

[0011] Preferably, step S3 comprises the following steps: Step S31: Divide the geometric center of gravity deviation optimization data into discrete intervals to obtain the geometric center of gravity deviation discrete intervals; Step S32: Design a clothing recommendation strategy based on the geometric center of gravity deviation discrete interval and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0012] Preferably, step S32 includes the following steps: Step S321: Calculate the characteristic distribution weight of the geometric center of gravity deviation discrete interval to obtain geometric center of gravity characteristic weight distribution data; Step S322: performing pattern matching logic learning on the pattern matching ratio segmentation effect data according to the geometric center of gravity feature weight distribution data to obtain pattern matching logic effect learning data; Step S323: Design a clothing recommendation strategy for the pattern matching logic effect learning data based on the strategy gradient algorithm to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0013] Preferably, the present invention also provides a clothing collocation recommendation system based on big data, which is used to execute the clothing collocation recommendation method based on big data as described above, and the clothing collocation recommendation system based on big data comprises: The clothing category grouping module is used to obtain the user clothing history browsing data set and the user input body shape data in the e-commerce platform; group the user clothing history browsing data set into clothing categories to obtain clothing category time series grouping data; The visual center of gravity optimization module is used to extract the color-pattern between different clothing categories from the clothing category time series grouping data to obtain the clothing category color-pattern data between different clothing categories; analyze the pattern matching ratio segmentation effect between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain the pattern matching ratio segmentation effect data; perform visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; The clothing recommendation strategy design module is used to design clothing recommendation strategies based on geometric center of gravity deviation optimization data and pattern matching ratio segmentation effect data to obtain clothing recommendation strategies to perform clothing matching recommendations.

[0014] The beneficial effect of the present invention is that, first of all, it is necessary to collect the user's clothing history browsing data set and the user's body input data from the e-commerce platform. The clothing history browsing data set records the user's browsing behavior on the platform and reflects the user's interest in different types of clothing in different time periods. By analyzing these data, the clothing can be grouped according to the type of clothing, thereby obtaining the time series grouping data of the clothing type. This grouping can help understand the changes in the user's preference for different types of clothing and identify potential fashion trends and individual preferences. At the same time, the body shape data input by the user provides an important personalized basis for subsequent recommendations. The body shape data can include information such as the user's height, weight, shoulder width, waist circumference, hip circumference, etc., laying the foundation for subsequent customization of clothing matching solutions according to user characteristics. The core value of this step is to lay a data foundation for subsequent clothing recommendations through the time series analysis of historical browsing data, and to use personalized body shape information as a core reference to improve the accuracy and matching of recommendations. The color and version features are extracted from the clothing type time series grouping data, and then the color-version data between clothing types are obtained. The color and version of clothing are the most significant features of its appearance, and these two features directly affect the matching effect of clothing and whether it is suitable for a specific body type. By extracting the color and pattern features between different types of clothing, we can understand the similarities and differences in color matching and pattern design between different types of clothing. For example, different types of clothing such as shirts, skirts, and pants have different color matching rules and pattern styles, and these features will have a profound impact on the user's wearing effect. At the same time, combined with the body shape data input by the user, the adaptation effect of different types of clothing on the body shape during the matching process can be analyzed, so as to calculate the pattern matching ratio segmentation effect between different types of clothing. This process not only provides a theoretical basis for clothing recommendation, but also provides quantitative analysis results for personalized recommendation. Through the optimization processing of visual geometric center of gravity deviation, the coordination and beauty of the matching can be further improved, making the recommendation more accurate. Based on the geometric center of gravity deviation optimization data and pattern matching ratio segmentation effect data obtained in the first two steps, a specific clothing recommendation strategy is designed. The geometric center of gravity deviation optimization data optimizes the overall shape of the clothing by analyzing the visual effect after clothing matching, making the matching more in line with the natural aesthetic principles of the human body. These data help identify and adjust the visual center of gravity of clothing, avoid imbalance or abrupt effects in matching, and thus improve the overall image of the wearer. Combined with the data on the effect of pattern matching ratio segmentation, the recommendation strategy can find the most appropriate matching ratio between different types of clothing, ensuring that the clothing can adapt to the user's body shape and conform to fashion trends. Ultimately, these strategies can provide personalized and accurate clothing recommendations based on each user's body shape, clothing preferences, and historical browsing behavior. This step can not only improve the user's wearing experience, but also enhance the user stickiness of the platform, prompting users to purchase and interact more frequently. The effectiveness of the clothing recommendation strategy has brought higher conversion rates and user satisfaction to the platform.Therefore, the present invention is an improvement on a traditional clothing matching recommendation method based on big data, which solves the problem that a traditional clothing matching recommendation method based on big data has inaccurate recognition of user body shape characteristics, thereby causing large errors in clothing matching recommendations, improves the accuracy of recognizing user body shape characteristics, and reduces the errors in clothing matching recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of the steps of a clothing matching recommendation method based on big data; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. DETAILED DESCRIPTION

[0016] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0017] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0018] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0019] To achieve this, please refer to Figures 1 to 3 , a clothing matching recommendation method based on big data, the method comprising the following steps: Step S1: obtaining a user clothing history browsing data set and user input body shape data in an e-commerce platform; grouping the user clothing history browsing data set by clothing categories to obtain clothing category time series grouping data; Step S2: performing color-pattern extraction between different clothing categories on the clothing category time series grouping data to obtain clothing category color-pattern data between different clothing categories; performing pattern matching ratio segmentation effect analysis between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain pattern matching ratio segmentation effect data; performing visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; Step S3: Design a clothing recommendation strategy based on the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0020] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a clothing collocation recommendation method based on big data of the present invention. In this example, the clothing collocation recommendation method based on big data includes the following steps: Step S1: obtaining a user clothing history browsing data set and user input body shape data in an e-commerce platform; grouping the user clothing history browsing data set by clothing categories to obtain clothing category time series grouping data; In the embodiment of the present invention, the process of obtaining the user's clothing browsing history data set and the user's input body shape data in the e-commerce platform first needs to extract the user's browsing data from the user behavior database of the platform. These data usually include the clothing information browsed by the user (such as clothing category, brand, color, price, browsing time, etc.), and the user's body shape information (such as height, weight, waist circumference, chest circumference, hip circumference, shoulder width, etc.). This information is usually obtained through the user's registration information on the platform or the data actively input. The acquired data set needs to be cleaned and standardized to remove invalid or missing records to ensure the integrity and consistency of the data. On this basis, the user's clothing browsing history data set is grouped by clothing type. First, the data is classified according to the classification information of the clothing (such as tops, pants, skirts, etc.). The specific operation is to label all clothing by type to form a clothing type data set. Then, the data of users browsing different types of clothing in different time periods is analyzed in time series. The data is divided according to the user's browsing order using a time series-based grouping method to obtain clothing type time series grouping data. These data reflect the changes in users' preferences for different types of clothing in different time periods. Subsequent steps will rely on these time-series grouped data to further analyze users' matching needs and body shape adaptation.

[0021] Step S2: performing color-pattern extraction between different clothing categories on the clothing category time series grouping data to obtain clothing category color-pattern data between different clothing categories; performing pattern matching ratio segmentation effect analysis between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain pattern matching ratio segmentation effect data; performing visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; In the embodiment of the present invention, the color-pattern extraction between different clothing categories is performed on the clothing category time series grouping data. First, the color and pattern of each category of clothing need to be calibrated. Color extraction can be performed on clothing pictures by image processing technology, using image recognition algorithms (such as K-means clustering algorithm) to perform color analysis on clothing pictures, and extract the main color tones and color distribution features from the pictures. Pattern extraction analyzes the pattern features of clothing through the structural design of clothing (such as the design features of collars, cuffs, waistlines, trouser legs and other parts). For each category of clothing, the pattern features of the upper body and lower body clothing are first extracted, which are divided into "upper body clothing category color-pattern data" and "lower body clothing category color-pattern data", and these two data sets will serve as the basis for subsequent analysis. On this basis, the body shape data input by the user is analyzed according to the clothing category color-pattern data. Specifically, the body shape data of the user needs to be standardized, such as converting the user's height, weight and other information into standard body shape proportions, and dividing the proportions of the upper and lower body through geometric morphological analysis. At this time, mathematical algorithms (such as the least squares method) are used to calculate the pattern matching ratios of different clothing types and user body types. The pattern matching effect between each clothing type and the user's body type is analyzed to obtain the pattern matching ratio segmentation effect data. Then, based on these matching effect data, the geometric center of gravity optimization technology is applied to optimize the pattern matching effect. The specific operation is to calculate the visual center of gravity position of each piece of clothing, and based on the geometric characteristics of the user's body type (such as chest circumference, waist circumference, hip circumference, etc.), adjust the center of gravity position of the clothing through the geometric deviation optimization algorithm to make it more in line with the user's body type, and finally obtain the geometric center of gravity deviation optimization data.

[0022] Step S3: Design a clothing recommendation strategy based on the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0023] In the embodiment of the present invention, the operation of designing a clothing recommendation strategy according to the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data first needs to merge the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data to form a user-personalized clothing recommendation basic data set. On this data set, the geometric center of gravity deviation optimization data is first divided into discrete intervals, the purpose of which is to convert the center of gravity deviation data into discrete data with clear distinction, thereby improving the accuracy of subsequent recommendations. The division of this discrete interval is based on a multi-dimensional analysis of the user's body shape, and a hierarchical clustering technology (such as the DBSCAN algorithm) is used to divide different center of gravity deviation features into multiple regions, each corresponding to a user preference. On this basis, according to the pattern matching ratio segmentation effect data, a recommendation strategy design method based on a gradient descent algorithm is used to match different body types of users with clothing types. The recommendation strategy design method identifies the most suitable matching strategy for the user by comparing the matching effect data and the center of gravity optimization data of the user, ensuring that the recommended clothing combination is not only visually coordinated, but also optimized in terms of body shape adaptability. Finally, the clothing recommendation strategy is fine-tuned based on the policy gradient algorithm (such as the Q-learning algorithm in reinforcement learning) to ensure the optimal effect of the recommendation system in practical applications. This recommendation strategy continuously learns from user feedback and matching effects, dynamically optimizes the recommendation results, and achieves accurate clothing matching recommendations.

[0024] Step S1 includes the following steps: Step S11: Acquire a user's clothing browsing history data set and user input body shape data in the e-commerce platform; Step S12: arranging the average browsing time sequence of the user's clothing history browsing data set to obtain clothing browsing average time sequence arrangement data; Step S13: grouping the user's clothing history browsing data set by clothing category according to the clothing browsing average time series arrangement data to obtain clothing category time series grouping data.

[0025] In the embodiment of the present invention, the operation of obtaining the user's clothing history browsing data set and the user's input body shape data in the e-commerce platform first needs to extract the user behavior data from the data warehouse of the e-commerce platform. These data usually contain detailed information of each piece of clothing browsed by the user, including the type, brand, color, material, browsing time, etc. of the clothing, as well as the user's basic body shape data, such as height, weight, waist circumference, hip circumference, shoulder width, etc. In order to ensure the accuracy and completeness of the data, it is necessary to perform preliminary data cleaning processing on the acquired user data to remove invalid data, duplicate data and missing data. In addition, the user's body shape data needs to be standardized, and all body shape data (such as height, weight, chest circumference, waist circumference, etc.) must be converted into a unified quantitative unit (such as centimeters and kilograms) for subsequent analysis. On this basis, the body shape data input by the user is combined with the browsing data to form a comprehensive data set, which provides basic data for subsequent time series analysis and clothing category grouping. The operation of performing browsing average time series arrangement on the user's clothing history browsing data set first needs to arrange the user's browsing data in chronological order. Each user's browsing data contains browsing records of different types of clothing at different time points, and each record contains browsing time, clothing category, brand and other details. In order to eliminate the differences in browsing frequency and time of different users, it is necessary to perform time series averaging on the browsing data of each user. Specifically, it is necessary to first normalize the browsing time of each user and unify the browsing time of different users into the same standard time scale. Then, statistical analysis is performed on the browsing data of each user to calculate the browsing frequency of different types of clothing in time, and the clothing types are arranged in time series based on the browsing frequency. When sorting the browsing frequency of each type of clothing of the user, the sorting algorithm in statistics (such as quick sort or merge sort) can be used to make the clothing with high browsing frequency be ranked in front, forming the average time series arrangement data of clothing browsing. The time series arrangement data reflects the user's clothing browsing preference in different time periods and provides a basis for the subsequent clothing type grouping. The operation of grouping the user's clothing history browsing data set into clothing types according to the average time series arrangement data of clothing browsing is first grouped according to the label of the clothing type. The grouping of clothing categories depends on clothing category labels (such as tops, pants, skirts, shoes, etc.), which can be automatically obtained through the classification information of the product. Each user's browsing history contains the browsing time sequence of different clothing categories. Based on the time sequence arrangement data, the browsing history of the same type of clothing can be clustered to form a time series data set for this type of clothing. In order to ensure the accuracy of the grouping, it is necessary to further apply clustering algorithms to refine the grouping of clothing categories. For example, the K-means clustering algorithm can be used to group clothing categories with similar browsing behavior characteristics.For each group, its browsing frequency and pattern in the time series are calculated, and based on this information, high-frequency and low-frequency clothing categories can be further divided. Finally, the obtained clothing category time series grouping data will contain each user's browsing behavior for various types of clothing in different time periods, and the browsing behavior of each clothing category has been optimized and grouped according to the time series pattern. Through this grouping method, accurate user clothing interest division can be achieved, and reliable input data can be provided for subsequent recommendation and matching analysis.

[0026] Step S2 includes the following steps: Step S21: extracting the color-pattern of different clothing categories from the clothing category time series grouping data to obtain clothing category color-pattern data of different clothing categories, wherein the clothing category color-pattern data includes the color-pattern of upper body clothing category and the color-pattern of lower body clothing category; Step S22: performing pattern matching ratio segmentation effect analysis between different clothing types on the user input body shape data according to the clothing type color-pattern data, and obtaining pattern matching ratio segmentation effect data; Step S23: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the clothing type color-pattern data to obtain the visual guidance center of gravity data of the matching; Step S24: performing visual geometric center of gravity deviation optimization processing on the matching visual guidance center of gravity data to obtain geometric center of gravity deviation optimization data.

[0027] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: extracting the color-pattern of different clothing categories from the clothing category time series grouping data to obtain clothing category color-pattern data of different clothing categories, wherein the clothing category color-pattern data includes the color-pattern of upper body clothing category and the color-pattern of lower body clothing category; In the embodiment of the present invention, the color-pattern extraction between different clothing categories is performed on the clothing category time series grouping data. First, it is necessary to extract the color and pattern features of each clothing category in the clothing category time series grouping data. Color extraction is performed based on the color features of the clothing image, and the color of the clothing is usually represented by the RGB value or HSV value in the color space. In the browsing data of each clothing category in the data set, the main color and color matching of the clothing are recorded. Image processing technology is used, especially by extracting the main color of each clothing image for statistics, and performing standardization processing on it, to obtain the color label of each clothing. For clothing pattern extraction, the clothing pattern category, such as slim, loose, straight, tight, etc., is identified by analyzing the structural features of the clothing and combining the object recognition technology. This process requires analyzing the outline of the clothing according to the design and morphological structure of the clothing, and extracting its specific pattern features. Based on these extracted color and pattern information, the upper body clothing and the lower body clothing are grouped respectively, and these data are integrated with the clothing category time series grouping data to obtain "upper body clothing category color-pattern" and "lower body clothing category color-pattern" data. In this process, color extraction can use color histogram or color clustering algorithm, while pattern recognition can make use of contour analysis and edge detection technology, and further combine the actual contour and shape of the clothing for labeling to ensure the accuracy and completeness of the data.

[0028] Step S22: performing pattern matching ratio segmentation effect analysis between different clothing types on the user input body shape data according to the clothing type color-pattern data, and obtaining pattern matching ratio segmentation effect data; In the embodiment of the present invention, the user input body shape data is analyzed for the pattern matching ratio segmentation effect between different clothing types according to the clothing type color-pattern data. First, the body shape data input by the user needs to be matched with the color and pattern data of the clothing type. In this process, the body shape data of the user is first standardized to ensure that all data items such as height, weight, chest circumference, waist circumference, hip circumference, etc. are within the same scale range. For the pattern data of different clothing types, each clothing pattern is fitted with the body shape data using a corresponding mathematical model to calculate the pattern matching ratio suitable for different body types. For example, based on the geometric structure of body shape data and clothing pattern, by establishing a proportional model, the appropriate ratio of upper body and lower body clothing in pattern matching under different body types is analyzed. This analysis models user groups of different body types through statistical methods such as classification and regression analysis to obtain the optimal matching ratio between clothing pattern and body type. By analyzing the effect of this proportional segmentation, the suitability data of pattern matching for users of different body types when wearing different clothing types can be obtained. The final data on the pattern matching ratio segmentation effect will provide a basis for the next step of matching effect optimization.

[0029] Step S23: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the clothing type color-pattern data to obtain the visual guidance center of gravity data of the matching; In the embodiment of the present invention, the visual guidance center of gravity analysis of the matching between different types of clothing is performed according to the clothing type color-pattern data and the pattern matching ratio segmentation effect data. First, the visual center of gravity of each group of clothing needs to be calculated according to the color and pattern distribution of the clothing. The key to the visual center of gravity analysis is to extract the geometric center of gravity of the clothing image, and calculate the spatial center of gravity in combination with the color distribution and pattern form of the clothing. The visual center of gravity of the clothing is weighted based on the color distribution of the clothing and the geometric form of the pattern. Specifically, for each piece of clothing, the visual center of the upper and lower body parts is calculated respectively according to its color and pattern distribution area. Then, the center of gravity of the whole set of clothing matching is obtained by merging the visual center of gravity of each type of clothing. In the matching process, special attention is paid to the balance of the upper and lower body matching, ensuring that the visual center of gravity of the upper and lower body matching is close or reasonably distributed to avoid the situation of incoordination in matching. The technology used in the analysis process includes the calculation method of the geometric center of the image and the color weighted average method to ensure that each clothing matching can achieve visual balance and coordination. Through this analysis, the final data on the center of gravity of visual guidance for matching provides an accurate matching balance reference for the clothing recommendation system.

[0030] Step S24: performing visual geometric center of gravity deviation optimization processing on the matching visual guidance center of gravity data to obtain geometric center of gravity deviation optimization data.

[0031] In the embodiment of the present invention, the visual geometric center of gravity deviation optimization processing is performed on the collocation visual guidance center of gravity data. First, the geometric center of gravity deviation detection is performed on the collocation data of each piece of clothing. According to the clothing type color-pattern data and the collocation visual guidance center of gravity data, the deviation between the visual center of the current collocation and the ideal center of gravity can be calculated. At this time, by comparing the geometric center of gravity differences under different clothing combinations, the geometric optimization algorithm is used to adjust the clothing combination in the collocation to make it more in line with ergonomics and visual aesthetics requirements. Specifically, first, according to the user's body data, combined with the pattern and color distribution of the clothing, an ideal collocation center of gravity area is set. Then, through the optimization algorithm, such as the gradient descent method or the Newton method, the clothing combination is adjusted so that the collocation of the upper and lower body clothing can minimize the center of gravity deviation and achieve a more suitable visual balance. In each adjustment of the optimization process, the influence of clothing pattern and color on the visual center of gravity is taken into account to ensure that each group of collocation can produce a harmonious effect visually. After the optimization process in this step, the final geometric center of gravity deviation optimization data provides a more accurate pairing basis for clothing matching recommendations, ensuring that each piece of clothing can maximize the user's body shape and visual needs.

[0032] Step S22 includes the following steps: Step S221: calculating the upper and lower body ratios of the user's inputted body shape data to obtain the upper and lower body ratios; Step S222: quantifying the limb morphology differences of the user inputted body shape data according to the ratio of the upper and lower body parts to obtain limb morphology difference quantified data; Step S223: Calculating the pattern shaping contour values ​​between different clothing categories on the clothing category color-pattern data to obtain the pattern shaping contour values ​​between different clothing categories; Step S224: adapting the pattern shaping contour value to the clothing pattern looseness according to the quantitative data of the upper and lower body proportions and the limb morphology differences, to obtain pattern functional looseness adaptation data; Step S225: Analyze the pattern matching ratio segmentation effect between different clothing types based on the pattern functional looseness adaptation data and the quantitative data of the upper and lower body proportions and limb shape differences, and obtain pattern matching ratio segmentation effect data.

[0033] In the embodiment of the present invention, the upper and lower body ratios of the user inputted body shape data are calculated. First, the body shape data of the user needs to be standardized to ensure that the quantized values ​​of the dimensions such as height, weight, chest circumference, waist circumference, and hip circumference in the data are of uniform scale. Based on the height, the upper and lower body of the user are respectively divided into proportions. The proportion of the upper body can usually be calculated by measuring the shoulder height and the vertical distance from the chest to the waist, while the proportion of the lower body is determined by measuring the vertical distance from the waist to the sole of the foot. According to these calculated proportions, the body shape characteristics of the user can be further quantified to obtain the "body shape upper and lower body ratio" data. This data provides a basic body shape basis for the subsequent personalized clothing recommendation for the user's body shape. Based on the proportion data, the length distribution of the upper body relative to the lower body can be obtained, which provides preliminary guidance for the optimization of clothing pattern matching. For extreme body shapes (such as a longer upper body or a longer lower body), the deviation value of the proportion can be further used to calculate the adaptability of different types of clothing, thereby affecting the subsequent clothing recommendation. According to the ratio of the upper and lower body of the body, the limb morphology difference of the user's input body data is quantified. First, according to the "body upper and lower body ratio" data obtained in step S221, the morphological difference of the user's limbs needs to be quantified. For the quantification of limb morphology, two dimensions can be used: the length of the upper and lower limbs, and the circumference of the limbs (for example, the circumference of the arms and legs). For the measurement of the length of the upper limbs, it can be determined by measuring the distance from the shoulder to the wrist; the length of the lower limbs is determined by measuring the vertical distance from the hip to the ankle. At the same time, it is also necessary to measure the circumference of the limbs, such as the upper arm circumference, leg circumference, etc., which can help further refine the body shape information. The quantification process uses the formula: upper limb morphology difference quantification = (upper arm circumference / upper limb length), lower limb morphology difference quantification = (leg circumference / lower limb length) By standardizing these differences, "limb morphology difference quantification data" can be obtained, which further reflects the degree of difference between the user's limbs and the overall body shape. These data will provide data support for subsequent clothing pattern optimization and matching suggestions, especially for predicting and adjusting the fit of different parts in clothing design to avoid inharmonious designs in the limbs. The color-pattern data of clothing categories are used to perform numerical calculations on the pattern shaping contours of different clothing categories. First, according to the color-pattern data of clothing categories, it is necessary to extract geometric features of the pattern of each clothing category, especially the outer contour shape of the clothing and its relative proportion to the human body. Through the contour analysis method, the outer contour of the clothing is extracted from the clothing image, and its geometric center is calculated. Next, according to the pattern design of the clothing, the key feature points of each clothing pattern need to be defined, such as the collar position, chest, waist and hip position. Using these key feature points, the morphological transformation of clothing in different parts can be further calculated, for example, the compactness of the upper body of the clothing and the looseness of the lower body.During the numerical calculation of the outline of the pattern shaping, the size comparison method in the geometric model is used to compare the differences in outline shapes between different types of clothing. The outline value of the clothing not only takes into account the design changes of the upper and lower body, but also evaluates whether these designs can be coordinated with the user's body shape. Finally, the obtained "pattern shaping outline value" data will provide data support for the next step of pattern looseness adaptation and clothing matching, helping to ensure that the clothing design meets the actual needs of the user's body shape. According to the quantitative data of the upper and lower body ratio and the limb morphology difference of the body shape, the pattern shaping outline value is adapted to the clothing pattern looseness. First, it is necessary to adjust the pattern looseness of the clothing according to the quantitative data of the upper and lower body ratio and the limb morphology difference calculated in steps S221 and S222. Specifically, the body shape data is compared with the pattern outline value of the clothing to determine the tightness of the clothing. For different body shapes, especially for users with longer upper or lower bodies, the pattern tightness needs to be adapted. For example, if the user's upper body is longer, the upper body part of the clothes should be appropriately loosened in the design of the clothing pattern to maintain the comfort and aesthetics of wearing; on the contrary, if the lower body is longer, the looseness of the lower body part should be appropriately adjusted to make the overall visual proportion more coordinated. The process of looseness adaptation can adjust the overall structure of the clothing pattern through a proportional scaling algorithm to ensure the fit of the clothing in actual wearing. By adapting the looseness of the clothing pattern, the wearing experience of the clothing can be optimized, thereby improving the data accuracy of the matching effect. Finally, the obtained "pattern functional looseness adaptation data" will provide further data support for subsequent clothing matching recommendations. According to the pattern functional looseness adaptation data, the pattern matching ratio segmentation effect between different clothing types is analyzed for the quantitative data of the upper and lower body proportions and limb morphology differences of the body shape. First, it is necessary to comprehensively analyze the "pattern functional looseness adaptation data" obtained in step S224 and the "body upper and lower body proportions" and "limb morphology differences quantitative data" obtained in steps S221 and S222. By using the ratio segmentation algorithm, according to the user's body characteristics and the looseness of the clothing pattern adaptation data, the appropriate ratio of different types of clothing is calculated. This analysis not only considers the looseness of the clothing, but also needs to consider the differences in the user's limb shape and the proportion of the upper and lower body, ensuring that each clothing combination can meet both comfort and the aesthetic requirements of the user's body shape to the greatest extent. In the process of pattern matching ratio segmentation effect analysis, an algorithm based on multivariate regression analysis is used. By inputting the user's body shape data, the pattern and looseness data of the clothing, the best matching ratio is obtained. The analysis result is "pattern matching ratio segmentation effect data", which provides an accurate matching ratio for the clothing recommendation system, which can recommend more suitable clothing combinations according to the differences in body shapes of different users, ensuring that the matching effect of clothing is more visually balanced.

[0034] Step S224 includes the following steps: Calculating the circumference of the limbs based on the quantitative data of the limb morphological differences to obtain the circumference of the limbs; wherein the circumference of the limbs includes: the circumference of the wrist, the circumference of the ankle, the circumference of the elbow, and the circumference of the knee; According to the proportion of the upper and lower body and the circumference of the limbs, the spatial differentiation of bone protrusions between different movement postures was analyzed to obtain the spatial differentiation data of bone protrusions between different movement postures; Based on the spatial differentiation data of bone protrusion, the space reservation amount of the limbs is calculated for the pattern shaping contour value to obtain the space reservation amount of the limbs; The looseness of the clothing pattern is adapted based on the reserved space for the limbs to obtain the functional looseness adaptation data of the pattern.

[0035] In an embodiment of the present invention, the circumference of the limbs is calculated based on the quantitative data of the limb morphology difference provided by the user, specifically including the circumference of the upper arm and the leg. The circumference of the limbs involved includes: the circumference of the wrist, the circumference of the ankle, the circumference of the elbow, and the circumference of the knee. For each circumference, a conventional measurement method is used to quantify it to ensure that the circumference calculation of each part has high accuracy. Taking the wrist circumference as an example, a measuring tool (such as a soft ruler) is used to measure the circumference of the thinnest part of the wrist to obtain the circumference of the wrist; similarly, the circumference of the ankle, elbow, and knee is measured. In practical applications, the body shape data provided by the user needs to be standardized to ensure that each measurement value has a consistent dimension and accuracy. The measured circumference data of the wrist, ankle, elbow, and knee are organized into a whole to obtain the "circumference of the limbs" data. This data is the basis for the design of the clothing pattern in the subsequent steps, and can provide the size characteristics of each part of the user's limbs, and provide a quantitative basis for the space reservation and looseness adjustment in clothing design. Based on the data of "circumference size of limbs" and the calculated "proportion of upper and lower body", the spatial differentiation of bone protrusion under sports posture is analyzed. In this analysis, it is first necessary to model the three-dimensional space of bones based on the user's body characteristics and common sports postures (such as standing, walking, running, sitting, etc.). According to the biomechanical model of the human body, the human bones (especially the limbs) will protrude to varying degrees in space under different sports postures. Using the three-dimensional bone model of the human body, combined with the specific circumference size of the limbs and body proportions of the user, the spatial differences of the protruding parts of the bones under each sports posture can be obtained. In this process, the specific implementation method uses three-dimensional coordinate calculation, calculates the displacement and deformation of the protruding parts of the bones (such as the knees and elbows) by adjusting the posture angle, and compares them with the standard human body model to obtain the quantitative data of the protruding space difference. For each sports posture, the calculated "spatial differentiation data of bone protrusion" characterizes the spatial distribution changes of the limb bones during the movement process, and provides a basis for the subsequent pattern adjustment to ensure that the clothing can adapt to the spatial requirements under different sports postures. After obtaining the spatial differentiation data of bone protrusion, the data is further used to optimize the pattern shaping of the clothing. First, the space requirements of limb parts (such as wrists, ankles, elbows and knees) during movement are evaluated based on the differences in bone protrusion space in each movement posture. For each part, its protrusion value is calculated, that is, the space requirement caused by the difference in bone or muscle morphology in a specific posture. By comparing with the contour values ​​of the clothing pattern, the additional space required for each limb part in the clothing pattern design is determined. This calculation can be performed as follows: Based on the differences in bone protrusion, the reserved space for each limb part is estimated to ensure that these differences are taken into account in the clothing design. By quantifying the space requirements of the bone protrusion part, the clothing reserve for each part is set.For example, for the knee, in a squatting or kneeling posture, the clothing fabric at the knee needs to be extra loose, and comfort is usually ensured by adjusting the cutting design of the knee area. Based on the calculation of this reserve, the "space reserve" required for each limb in different movement postures is obtained. This data provides a quantitative basis for the next step of loose adaptation of the clothing pattern. According to the obtained space reserve of the limbs, these reserved spaces are directly integrated into the design process of the clothing pattern. For example, in the knee area, the looseness of the clothing fabric is increased to adapt to the activity space of the knee joint; similarly, the wrist and ankle parts also need to be adjusted according to the looseness of the space reserve. Through physical and mechanical methods, the elasticity and resilience of the clothing fabric are analyzed, and the looseness of the clothing is calculated in combination with the space requirements of the limbs to ensure that the clothing can adapt to the movements of the limbs in different movement postures. In the design of clothing pattern, the looseness of the fabric is optimized by using appropriate cutting technology, so that the clothing can provide sufficient activity space without sacrificing comfort and aesthetics. This process can be accomplished by adjusting the sewing points of the garment, adjusting the looseness of the garment, or by designing elastic fabric areas. Finally, through the above steps, the "functional looseness adaptation data of the pattern" is obtained, which provides key data support for subsequent clothing recommendations, ensuring that the clothing can maintain a balance between comfort and beauty in different sports postures, and meet the user's personalized clothing needs.

[0036] Step S23 includes the following steps: Step S231: extracting colors and patterns between different clothing categories from the clothing category color-pattern data to obtain clothing color data between different clothing categories and clothing pattern data between different clothing categories; Step S232: performing color visual compactness analysis on clothing color data of different clothing types to obtain color visual compactness data; Step S233: quantifying the hue interlaced spatial hierarchy according to the color visual compactness data and the clothing pattern data to obtain hue interlaced spatial hierarchy data; Step S234: identifying the law of visual center of gravity transfer on the pattern matching ratio segmentation effect data based on the color tone interlaced spatial hierarchy data to obtain the law of visual center of gravity transfer data; Step S235: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the visual center of gravity transfer rule data to obtain matching visual guidance center of gravity data.

[0037] In the embodiment of the present invention, the color and pattern information of different types of clothing are extracted by analyzing the color-pattern data of clothing types. For each type of clothing, its color components are first identified. This process uses a color extraction algorithm to extract the main color distribution in the clothing picture through an image analysis tool, and specifically uses an RGB (red, green, and blue) color space model to separate and grade colors to obtain the main color and auxiliary color of each type of clothing. By quantifying the proportion of color areas, the "clothing color data" of each type of clothing is finally generated. At the same time, the pattern of the clothing is extracted by using image processing technology, and the texture, print or other pattern elements on the clothing are identified by a pattern recognition algorithm. These patterns can be geometric shapes, stripes, patterns, etc. During the pattern extraction process, the characteristics of the pattern are extracted by edge detection and graphic recognition algorithms of the image. For each type of clothing, the corresponding "clothing color data" and "clothing pattern data" are obtained respectively. These data will serve as the basis for guiding clothing matching in subsequent analysis, and can provide basic information for clothing visual perception. After obtaining the color data of different types of clothing, "color visual compactness analysis" is performed next. This analysis aims to evaluate the visual aggregation of each clothing color and the matching effect of the color. First, according to the color data of the clothing, by quantifying the visual distribution of each clothing color, calculating the proportion of each color on the clothing, and combining the principles of visual psychology, the compactness of the clothing color is analyzed. The compactness of the color reflects whether the color is evenly distributed in the clothing and whether the color matching is harmonious. Specifically, the color contrast calculation and color position distance calculation are used to measure the colors of different clothing through the "similarity of color space" algorithm. If the two colors are close, they are considered to be visually compact; if the two colors are in strong contrast, they are considered to be visually not compact. The "color visual compact data" calculated in this step can provide a basis for subsequent matching to ensure the color unity and visual sense balance of clothing when matching. According to the "color visual compact data" obtained in step S232 and the "clothing pattern data" extracted in step S231, "tonal interlacing spatial hierarchical quantification" is performed. The core goal of this step is to analyze the interactive relationship between clothing color and pattern and its hierarchical structure in space. First, by performing spatial distribution analysis on the clothing pattern data, the position and frequency of the pattern in the clothing are quantified, and the degree of tonal interlacing of each clothing pattern is obtained in combination with the color distribution. Specifically, for clothing with complex patterns, it is necessary to identify the multi-level color layout of the pattern, divide the pattern into different color blocks or color bands through hierarchical analysis, and then evaluate the degree of interlacing between different color blocks or color bands based on the color distribution. Using the spatial hierarchical analysis method, the interlacing relationship between clothing colors and patterns is quantified to obtain the spatial hierarchical data of the tonal interlacing of clothing.This data can clearly reflect how the colors and patterns of clothing are combined in space, thus providing a more refined hierarchical reference for subsequent clothing matching. After obtaining the "hue interlaced spatial hierarchical data", continue to identify the law of visual center of gravity transfer in clothing matching based on this data. Through the relationship between hue interlacing and pattern hierarchy, the law of visual center of gravity transfer in clothing matching can be analyzed. The specific operation is to first analyze the visual balance of the matching effect of clothing color and pattern. When there is a large difference between the color and pattern in the clothing, the visual center of gravity will tend to the more conspicuous part; when the color and pattern are more balanced, the visual center of gravity is more stable. This process calculates the visual center of gravity position of each clothing match through factors such as the matching ratio, position and contrast with the human body of the color and pattern. Based on these analyses, the regular data of center of gravity transfer in clothing matching is obtained, that is, when the clothing is under different matching schemes, how the visual center of gravity shifts with different matching. This data provides a scientific basis for subsequent matching recommendations, which can help users better understand how to guide the visual focus through the matching of colors and patterns, thereby optimizing the matching effect of clothing. Finally, the visual guidance center of gravity of different types of clothing is analyzed based on the "visual center of gravity transfer law data". This analysis evaluates the visual guidance center of gravity in clothing matching by comprehensively considering the color, pattern and version of clothing. First, through the visual center of gravity transfer law data obtained from the previous analysis, combined with the matching schemes of different types of clothing, it is evaluated how each clothing combination affects the guidance of the visual center. Specifically, it analyzes how the colors and patterns of tops, bottoms and accessories in different clothing matching work together to form the final visual focus. For example, for a combination of brighter tops and darker bottoms, the visual focus is often concentrated on the upper body; while for a combination of balanced colors of tops and bottoms, the visual focus will be more even. By evaluating these matching effects, the "matching visual guidance center of gravity data" is obtained, which provides a basis for the final decision of clothing matching recommendations. This data can help the system identify the visual effects required by users and optimize their visual effects and wearing experience by reasonably matching clothing.

[0038] Step S234 includes the following steps: Perform visual landing point analysis on the hue interlaced spatial hierarchy data to obtain hue interlaced visual landing point data; Based on the color tone staggered visual landing point data, the clothing pattern data and the pattern matching ratio segmentation effect data are subjected to the visual expansion perception of the stripe interlaced arc, and the visual expansion perception data of the stripe arc between the matching ratio segmentation is obtained; Performing perception direction guidance analysis on the stripe curvature visual expansion perception data to obtain stripe curvature perception direction guidance data; Based on the stripe arc visual expansion perception data and the stripe arc perception direction guidance data, the visual center of gravity transfer law is identified to obtain the visual center of gravity transfer law data.

[0039] In the embodiment of the present invention, first, the hue interlaced spatial hierarchy data is used as input to perform visual landing point analysis. The goal of this analysis is to identify the visual focus and its landing point position in the color and pattern matching of clothing. The analysis process starts with color distribution, pattern morphology and clothing pattern structure, and quantifies the color area of ​​clothing, the local distribution of patterns and the visual performance of different parts of clothing. Specifically, the center of gravity position of clothing color is calculated by using the "color distribution center of gravity method", and then the center of gravity of the pattern is calculated in combination with the edge detection technology of the pattern. Further, the image segmentation algorithm is used to distinguish different color blocks in the clothing pattern to determine the local visual center of gravity of each color block or pattern. Through the comprehensive analysis of these algorithms, the concentrated area or "landing point" of the visual focus in clothing matching is obtained, that is, the visual focus area of ​​each clothing matching. The obtained "hue interlaced visual landing point data" includes the visual concentration and specific location of different clothing color and pattern interlaced areas, providing data support for subsequent matching optimization. After obtaining the "hue interlaced visual landing point data", the next step is to perform stripe interlaced arc visual expansion perception on the clothing pattern data and the pattern matching ratio segmentation effect data based on the data. Stripe interweaving refers to the visual effect produced by the interweaving of different stripe patterns and other pattern elements in clothing matching. This step uses the "stripe curve transformation algorithm" to deform and stretch the stripes in the clothing pattern according to the center of gravity and segmentation ratio of the pattern, simulating the visual expansion effect of the stripes and the clothing pattern after interweaving. By linearly transforming the geometric form of the stripe pattern, the visual expansion effect of the stripes interweaving in different pattern matching is obtained. In the specific operation, by analyzing the data of the clothing pattern, the stripe elements in the clothing pattern are identified using edge recognition technology, and then the geometric form of the stripes is stretched and bent through a specific curvature control algorithm, so as to achieve the visual expansion perception of stripe interweaving. The output of this step is "stripe curvature visual expansion perception data", which provides dynamic stripe interweaving curvature information for the visual guidance of each clothing matching, which can help determine the perceptual expansion effect of stripes in matching. Next, based on the "stripe curvature visual expansion perception data" obtained in the previous step, the perceptual direction guidance analysis is performed. The core of this step is to analyze the guiding role of the curvature and interweaving mode of the stripes on the visual perception direction. Combined with the curvature of the stripes and the distribution of the clothing pattern, the direction of the stripe pattern is quantitatively analyzed. Specifically, first, the inclination angle, curvature and interlaced relationship of the stripes are calculated, and then the visual guidance direction of the stripe pattern on the clothing is determined through geometric analysis. Using the projection method, the curvature change of the stripe pattern is converted into a guiding direction on a two-dimensional plane. Through these data, the perceived direction of the stripe pattern in the clothing combination is obtained. Furthermore, by calculating the relationship between the curvature of the stripes and the visual guidance path, the final direction of the stripes in each clothing combination is obtained.The result of this analysis is the "stripe arc perception direction guidance data", which is used to describe how the stripe pattern guides the flow of sight and thus affects the overall visual effect of clothing. Finally, after obtaining the "stripe arc visual expansion perception data" and "stripe arc perception direction guidance data", the law of visual center of gravity transfer is identified. The purpose of this step is to analyze the influence of the stripe pattern and its curved shape on the visual center of gravity of clothing, and how to guide the transfer of the visual center of gravity by adjusting the curvature and direction of the stripes. First, by comprehensively analyzing the "stripe arc visual expansion perception data" and "stripe arc perception direction guidance data" of clothing matching, the center of gravity transfer model is used to calculate the change of the visual center of gravity position under different stripe matching. Specifically, for each clothing matching, the relationship between the shape of the stripe pattern and the clothing version is analyzed, and how the stripes affect the distribution of the visual center of gravity in the matching process is calculated. By calculating the driving effect of the shape of the stripes on the center of gravity position in each matching, the law of center of gravity transfer of each clothing matching scheme is obtained. This law helps to identify the guiding effect of the stripe pattern on the visual focus in clothing matching, and finally obtain the "visual center of gravity transfer law data". These data provide a scientific basis for clothing matching recommendations, helping to design matching plans with a reasonable distribution of visual center of gravity and in line with aesthetic needs.

[0040] Step S3 includes the following steps: Step S31: Divide the geometric center of gravity deviation optimization data into discrete intervals to obtain the geometric center of gravity deviation discrete intervals; Step S32: Design a clothing recommendation strategy based on the geometric center of gravity deviation discrete interval and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0041] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: Divide the geometric center of gravity deviation optimization data into discrete intervals to obtain the geometric center of gravity deviation discrete intervals; In the embodiment of the present invention, the "geometric center of gravity deviation optimization data" is first divided into discrete intervals. The geometric center of gravity deviation optimization data reflects the difference between the visual center of gravity and the actual physical center of gravity of each part of the clothing pattern in the matching, and the purpose is to ensure that the visual center of gravity of the matching is more in line with the natural center of gravity distribution law of the human body through optimization and adjustment. In the specific implementation, the geometric center of gravity deviation data is first mapped to a coordinate system, and this coordinate system uses the upper body and the lower body of the clothing as two main segmentation areas. Then, according to the characteristics of the clothing pattern, the geometric center of gravity deviation data is divided into several discrete intervals using the "interval division method". This method is based on the distribution characteristics of the center of gravity deviation in clothing matching, and the data is segmented according to the key points in the clothing matching to ensure that the center of gravity deviation changes in each section are consistent. In each interval, the data is further subdivided using a uniform distribution method to ensure that the value range within each interval is not too wide, and the center of gravity deviation information in each clothing matching can be accurately reflected. After the discrete interval division is completed, the data of each interval will serve as the basis for subsequent analysis and recommendation. Through this discretization process, the subsequent data analysis and recommendation process can more accurately handle the center of gravity deviation of each clothing combination and avoid errors caused by too continuous data. The output of this step is the "geometric center of gravity deviation discrete interval", which will provide a specific basis for the subsequent clothing recommendation strategy design.

[0042] Step S32: Design a clothing recommendation strategy based on the geometric center of gravity deviation discrete interval and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0043] In the embodiment of the present invention, a clothing recommendation strategy is designed based on the "geometric center of gravity deviation discrete interval" and "pattern matching ratio segmentation effect data" obtained in the previous step. The goal of clothing recommendation strategy design is to recommend appropriate clothing matching according to different geometric center of gravity deviation intervals and the pattern matching ratio segmentation effect. First, the "weighted average method" is used to combine the matching ratio segmentation effect data in different pattern matching with the data of the geometric center of gravity deviation discrete interval. The weighted average method scores the effect of clothing matching according to the center of gravity deviation data of each interval, and weighted averages the scoring results to evaluate the effectiveness of different matching strategies. This method automatically adjusts the weight of each matching according to the difference in clothing pattern and matching ratio, thereby giving different priorities to each matching recommendation scheme. In the specific implementation process, the "geometric center of gravity deviation discrete interval" data is first grouped according to the clothing pattern classification, and then the corresponding pattern matching ratio segmentation effect data is also divided according to the type of clothing. Next, different recommendation strategies are designed according to the characteristics of each clothing pattern. For example, for loose clothing styles, a recommendation strategy can be designed that focuses on optimizing the visual center of gravity deviation, while for tight styles, it focuses on adjusting the proportion segmentation effect to ensure the balance between comfort and visual effect when wearing. The design of each strategy will fine-tune the matching proportion segmentation effect based on the results of the previous center of gravity deviation analysis, so that each recommended clothing matching scheme can achieve the best balance between visual and actual wearing effects. On this basis, the personalized characteristics of the user's body shape need to be considered when designing the recommendation strategy. By establishing a "body shape-clothing adaptation" matrix, the recommendation strategy is further refined based on the user's body shape data. This process matches the user's body shape characteristics with the evaluation results of the clothing matching scheme to obtain the most suitable matching recommendation for the user. For example, for users with a fuller body shape, the tightness of the clothing can be appropriately increased when recommending matching, and the version ratio can be adjusted to ensure that the visual effect after matching meets the user's aesthetic and comfort requirements. Finally, through this design process, the generated "clothing recommendation strategy" can specifically guide each user to choose the best clothing matching according to their body shape, visual center of gravity deviation and clothing version matching ratio. These recommendation strategies can achieve accurate clothing matching recommendations based on big data, optimize the user's wearing experience, and improve the personalization of matching effects.

[0044] Preferably, step S32 includes the following steps: Step S321: Calculate the characteristic distribution weight of the geometric center of gravity deviation discrete interval to obtain geometric center of gravity characteristic weight distribution data; Step S322: performing pattern matching logic learning on the pattern matching ratio segmentation effect data according to the geometric center of gravity feature weight distribution data to obtain pattern matching logic effect learning data; Step S323: Design a clothing recommendation strategy for the pattern matching logic effect learning data based on the strategy gradient algorithm to obtain a clothing recommendation strategy to perform clothing matching recommendations.

[0045] In the embodiment of the present invention, the feature distribution weight of the "geometric center of gravity deviation discrete interval" is first calculated to obtain the "geometric center of gravity feature weight distribution data". The purpose of this step is to calculate the relative importance of each interval in clothing matching based on the data characteristics in the discrete interval, and then provide a basis for the subsequent clothing matching recommendation. The implementation process first analyzes the distribution characteristics of each interval based on the geometric center of gravity deviation data in the discrete interval using statistical methods. For example, by calculating the standard deviation, mean and data concentration of each interval, the weight of different intervals in the visual center of gravity deviation is evaluated. Next, based on these statistics, the feature weights of each interval are standardized by normalization to ensure that their sum is 1. This standardization process can prevent certain intervals from having an excessive impact on the final result, so that the weight of each interval can accurately reflect its importance in clothing matching. In specific operations, the key feature data (such as center of gravity deviation value, clothing version features, etc.) of each geometric center of gravity deviation discrete interval are first extracted, and then these features are weighted using the weighted average method to obtain the feature weight of each interval. Through this method, the influence of each interval can be clearly calculated and used as the basis for subsequent analysis. The "geometric center of gravity feature weight distribution data" finally obtained in this step will be used to guide the logical learning and recommendation strategy design of clothing matching, ensuring that more accurate adjustments can be made according to the characteristics of different intervals during the matching process. Based on the "geometric center of gravity feature weight distribution data", the "pattern matching ratio segmentation effect data" is used to learn the pattern matching logic to obtain the "pattern matching logic effect learning data". The key task of this step is to optimize the logical structure of the matching scheme by learning the pattern matching ratio in clothing matching and its relationship with the geometric center of gravity deviation. In the implementation process, the "geometric center of gravity feature weight distribution data" obtained in the previous step is first matched with the pattern matching ratio of each clothing matching scheme to clarify the impact of different pattern matching ratios on the geometric center of gravity deviation in each matching scheme. On this basis, the matching logic learning is carried out by using methods such as "logistic regression analysis" or "support vector machine" combined with the effect data of different matching schemes.

[0046] Specifically, the relationship between the version ratio segmentation effect and the geometric center of gravity deviation of different clothing combinations is fitted, and a regression analysis model is used to take the characteristics of the matching ratio and the geometric center of gravity deviation as input and the matching effect as output for model training. By learning the clothing matching logic, it can be clear which matching ratios have the most significant optimization effect on the geometric center of gravity deviation. The "pattern matching logic effect learning data" finally obtained in this step will serve as an important basis for designing recommendation strategies to ensure the rationality and practicality of the recommendation results. Based on the "pattern matching logic effect learning data", the "strategy gradient algorithm" is used to design clothing recommendation strategies to obtain the final "clothing recommendation strategy". The policy gradient algorithm is a common method in reinforcement learning for solving the optimal strategy. In this step, the policy gradient algorithm will dynamically optimize the clothing matching strategy by simulating the effects of different matching schemes based on the "pattern matching logic effect learning data". First, based on the "pattern matching logic effect learning data", an initial clothing matching strategy is set and used to generate a set of clothing matching schemes. Then, by evaluating the effect of each matching scheme (for example, from multiple dimensions such as visual center of gravity deviation, comfort, matching effect, etc.), the reward value of each scheme is calculated. Next, the policy gradient algorithm gradually adjusts the matching strategy to continuously improve its ability to select appropriate matches. The specific operation is: perform a gradient update on the current strategy and adjust the matching ratio parameters in the strategy so that each matching solution selected can get a higher reward. This process is completed through repeated iterations. Each update will adjust the strategy according to the current matching effect, so that the final matching solution can minimize the geometric center of gravity deviation and achieve the optimal matching effect. Through the policy gradient algorithm, it can be ensured that the clothing recommendation strategy can adaptively make personalized recommendations based on different body characteristics, pattern matching ratios and geometric center of gravity deviations. Finally, after the policy gradient algorithm training is completed, the obtained "clothing recommendation strategy" will be applied to the actual clothing matching recommendation to help users find the most suitable matching solution among the massive clothing choices.

[0047] The present invention also provides a clothing collocation recommendation system based on big data, which is used to execute the clothing collocation recommendation method based on big data as described above. The clothing collocation recommendation system based on big data includes: The clothing category grouping module is used to obtain the user clothing history browsing data set and the user input body shape data in the e-commerce platform; group the user clothing history browsing data set into clothing categories to obtain clothing category time series grouping data; The visual center of gravity optimization module is used to extract the color-pattern between different clothing categories from the clothing category time series grouping data to obtain the clothing category color-pattern data between different clothing categories; analyze the pattern matching ratio segmentation effect between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain the pattern matching ratio segmentation effect data; perform visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; The clothing recommendation strategy design module is used to design clothing recommendation strategies based on geometric center of gravity deviation optimization data and pattern matching ratio segmentation effect data to obtain clothing recommendation strategies to perform clothing matching recommendations.

[0048] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0049] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A clothing matching recommendation method based on big data, characterized in that: The following steps are involved: Step S1: Obtain the user clothing browsing history dataset and user input body shape data in the e-commerce platform; Group the user's clothing history browsing data set by clothing category to obtain clothing category time series grouping data; Step S2: performing color-pattern extraction between different clothing categories on the clothing category time series grouping data to obtain clothing category color-pattern data between different clothing categories; performing pattern matching ratio segmentation effect analysis between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain pattern matching ratio segmentation effect data; performing visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; Step S3: Design a clothing recommendation strategy based on the geometric center of gravity deviation optimization data and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

2. The clothing matching recommendation method based on big data according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Acquire a user's clothing browsing history data set and user input body shape data in the e-commerce platform; Step S12: arranging the average browsing time sequence of the user's clothing history browsing data set to obtain clothing browsing average time sequence arrangement data; Step S13: grouping the user's clothing history browsing data set by clothing category according to the clothing browsing average time series arrangement data to obtain clothing category time series grouping data.

3. The clothing matching recommendation method based on big data according to claim 1 is characterized in that: Step S2 includes the following steps: Step S21: extracting the color-pattern of different clothing categories from the clothing category time series grouping data to obtain clothing category color-pattern data of different clothing categories, wherein the clothing category color-pattern data includes the color-pattern of upper body clothing category and the color-pattern of lower body clothing category; Step S22: performing pattern matching ratio segmentation effect analysis between different clothing types on the user input body shape data according to the clothing type color-pattern data, and obtaining pattern matching ratio segmentation effect data; Step S23: performing a visual guidance center of gravity analysis of different types of clothing on the pattern matching ratio segmentation effect data according to the clothing type color-pattern data to obtain the visual guidance center of gravity data of the matching; Step S24: performing visual geometric center of gravity deviation optimization processing on the matching visual guidance center of gravity data to obtain geometric center of gravity deviation optimization data.

4. The clothing matching recommendation method based on big data according to claim 3 is characterized in that: Step S22 includes the following steps: Step S221: calculating the upper and lower body ratios of the user's inputted body shape data to obtain the upper and lower body ratios; Step S222: quantifying the limb morphology differences of the user inputted body shape data according to the ratio of the upper and lower body parts to obtain limb morphology difference quantified data; Step S223: Calculating the pattern shaping contour values ​​between different clothing categories on the clothing category color-pattern data to obtain the pattern shaping contour values ​​between different clothing categories; Step S224: adapting the pattern shaping contour value to the clothing pattern looseness according to the quantitative data of the upper and lower body proportions and the limb morphology differences, to obtain pattern functional looseness adaptation data; Step S225: Analyze the pattern matching ratio segmentation effect between different clothing types based on the pattern functional looseness adaptation data and the quantitative data of the upper and lower body proportions and limb shape differences, and obtain pattern matching ratio segmentation effect data.

5. The clothing matching recommendation method based on big data according to claim 1 is characterized in that: Step S3 includes the following steps: Step S31: Divide the geometric center of gravity deviation optimization data into discrete intervals to obtain the geometric center of gravity deviation discrete intervals; Step S32: Design a clothing recommendation strategy based on the geometric center of gravity deviation discrete interval and the pattern matching ratio segmentation effect data to obtain a clothing recommendation strategy to perform clothing matching recommendations.

6. The clothing matching recommendation method based on big data according to claim 5 is characterized in that: Step S32 includes the following steps: Step S321: Calculate the characteristic distribution weight of the geometric center of gravity deviation discrete interval to obtain geometric center of gravity characteristic weight distribution data; Step S322: performing pattern matching logic learning on the pattern matching ratio segmentation effect data according to the geometric center of gravity feature weight distribution data to obtain pattern matching logic effect learning data; Step S323: Design a clothing recommendation strategy for the pattern matching logic effect learning data based on the strategy gradient algorithm to obtain a clothing recommendation strategy to perform clothing matching recommendations.

7. A clothing matching recommendation system based on big data, characterized in that: Used to execute the clothing matching recommendation method based on big data as claimed in claim 1, the clothing matching recommendation system based on big data comprises: The clothing category grouping module is used to obtain the user clothing history browsing data set and the user input body shape data in the e-commerce platform; group the user clothing history browsing data set by clothing category to obtain clothing category time series grouping data; The visual center of gravity optimization module is used to extract the color-pattern between different clothing categories from the clothing category time series grouping data to obtain the clothing category color-pattern data between different clothing categories; analyze the pattern matching ratio segmentation effect between different clothing categories on the user input body shape data according to the clothing category color-pattern data to obtain the pattern matching ratio segmentation effect data; perform visual geometric center of gravity deviation optimization processing on the pattern matching ratio segmentation effect data to obtain geometric center of gravity deviation optimization data; The clothing recommendation strategy design module is used to design clothing recommendation strategies based on geometric center of gravity deviation optimization data and pattern matching ratio segmentation effect data to obtain clothing recommendation strategies to perform clothing matching recommendations.

Images