A hairstyle simulation method based on image recognition

Abstract

The present invention relates to the field of image processing, and particularly to a hairstyle simulation method based on image recognition. The method includes determining hair quality information; simulating a three-dimensional hair image; determining the curling parameters of a target hairstyle; obtaining a curled hair image through a curling model; obtaining a curling fitting offset index; determining whether the curling parameters need to be corrected; and obtaining corrected curling parameters when the curling parameters need to be corrected. The present invention intelligently analyzes the original hair image through a deep learning model, accurately obtains the hair quality information of the user's hair, generates a realistic three-dimensional hair image based on this information, and precisely simulates a personalized hairstyle. By comparing the target hairstyle with the simulated image, fine matching in multiple angles and multiple parameters is achieved. Through an adaptive parameter correction mechanism, key parameters such as curling temperature, time, and gap are dynamically adjusted according to the fitting offset index, approaching the target hairstyle to the greatest extent, and improving the curling efficiency of the curling iron and the sensory quality of the user.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a hairstyle simulation method based on image recognition. Background Art

[0002] With the rapid development of Internet and mobile technologies, the demand for personal image beautification and personalized services is increasing day by day. As an important part of personal image, more and more people hope to obtain more accurate and personalized suggestions and simulation experiences when choosing a hairstyle. Traditional curling methods mainly rely on the operator's experience and subjective judgment, lacking accurate analysis of personal hair quality and personalized adaptation.

[0003] The patent document with the publication number CN113191404A discloses a hairstyle migration model training method, a hairstyle migration method and related devices. Among them, the hairstyle migration model training method includes: obtaining a sample hairstyle image and a sample reference hairstyle image; performing feature extraction and feature fusion on the sample hairstyle image and the sample reference hairstyle image to obtain a sample fusion feature image; performing feature reduction on the sample fusion feature image to obtain a sample migrated hairstyle image; inputting the sample migrated hairstyle image into an image discrimination model and a hairstyle prediction model to obtain a discrimination result and a hairstyle prediction result; based on the sample migrated hairstyle image and the sample hairstyle image, as well as the hairstyles in the hairstyle prediction result and the sample reference hairstyle image, and the discrimination result, performing iterative parameter tuning on the hairstyle migration model, the image discrimination model and the hairstyle prediction model until the hairstyle migration model, the image discrimination model and the hairstyle prediction model all converge.

[0004] Therefore, the following problems can be seen: In the prior art, the actual curling hairstyle is affected by the hair quality, and it is impossible to determine and adjust the operating parameters actually required for curling according to the different hair qualities of users, resulting in poor curling effects. Summary of the Invention

[0005] For this reason, the present invention provides a hairstyle simulation method based on image recognition, which is used to simulate the user's hair and adaptively determine the curling parameters during curling based on the simulated curling image, so as to overcome the problem in the prior art that the operating parameters for curling cannot be determined and adjusted according to the different hair qualities of users, resulting in poor curling effects.

[0006] To achieve the above object, the present invention provides a hairstyle simulation method based on image recognition, including:

[0007] Obtaining the original hair image of the user, and determining the hair quality information based on the original hair image;

[0008] Transmitting the hair quality information to the client, and the client performs hair simulation according to the hair quality information to obtain a three-dimensional hair map;

[0009] Determine the curling parameters of the curling iron corresponding to the target hairstyle;

[0010] Perform curling simulation on the three-dimensional hair image according to the determined curling parameters to obtain a curled hair image;

[0011] Compare the target hairstyle image with the curled hair image to obtain a curling fitting offset index;

[0012] Determine whether it is necessary to correct the curling parameters based on the curling fitting offset index;

[0013] When it is determined that it is necessary to correct the curling parameters, correct the curling parameters to obtain corrected curling parameters;

[0014] The curling parameters include curling temperature, curling time, and curling gap;

[0015] The curling fitting offset index includes a first curling fitting offset index and a second curling fitting offset index.

[0016] Further, the determining whether it is necessary to correct the curling parameters based on the curling fitting offset index includes:

[0017] Compare the first curling fitting offset index with the first standard curling fitting offset index to obtain a first comparison result;

[0018] Compare the second curling fitting offset index with the second standard curling fitting offset index to obtain a second comparison result;

[0019] Determine whether it is necessary to correct the curling parameters based on the first comparison result and the second comparison result.

[0020] Further, the determining whether it is necessary to correct the curling parameters based on the first comparison result and the second comparison result includes:

[0021] If the first comparison result is that the first curling fitting offset index is greater than the first standard curling fitting offset index and the second comparison result is that the second curling fitting offset index is greater than the second standard curling fitting offset index, determine that it is necessary to correct the curling parameters when either condition is met.

[0022] Further, the correcting the curling parameters to obtain corrected curling parameters includes:

[0023] Determine the curling parameters that need to be corrected based on the first comparison result and the second comparison result;

[0024] Use a preset correction amplitude to correct the curling parameters that need to be corrected;

[0025] During the process of correcting the curling parameters, the curling image is corrected in real time according to the changes in the curling parameters to obtain a corrected curling image;

[0026] Determine the corrected fitting offset index between the corrected curling image and the target hairstyle image during the process of correcting the curling parameters;

[0027] Determine the corrected difference factor between the corrected fitting offset index and the corresponding first standard curling fitting offset index or second standard curling fitting offset index during the process of correcting the curling parameters;

[0028] Determine the corrected curling parameters according to the corrected difference factor.

[0029] Further, the curling parameters that need to be corrected determined based on the first comparison result and the second comparison result include:

[0030] If the first curling fitting offset index is greater than the first standard curling fitting offset index, adjust the curling temperature or the curling time;

[0031] If the second curling fitting offset index is greater than the second standard curling fitting offset index, adjust the curling gap.

[0032] Further, the determination of the corrected curling parameters according to the corrected difference factor includes:

[0033] Determine the change of the corrected difference factor during the process of correcting the curling. If the corrected difference factor gradually increases during the process of correcting the curling, it is determined that the correction is invalid, and the corrected curling parameters are the curling parameters;

[0034] If the corrected difference factor gradually decreases during the correction process, it is determined that the correction is effective, and the process curling parameters before the increase of the corrected difference factor during the correction process are determined as the corrected curling parameters.

[0035] Further, the determination of the hair quality information based on the original hair image includes:

[0036] Preprocess the original hair image to obtain a hair image to be processed;

[0037] Use a pre-trained deep learning model to analyze the hair image to be processed and determine the hair quality information of the hair image to be processed.

[0038] Further, the obtaining of the curling fitting offset index by comparing the target hairstyle image with the curling image includes:

[0039] Obtain the hairstyle parameters of the curling image, and the hairstyle parameters include the curl rate and the curl length;

[0040] Determine a first curly hair fitting offset index and a second curly hair fitting offset index based on the hairstyle parameters and the standard hairstyle parameters of the target hairstyle.

[0041] Further, the determining the first curly hair fitting offset index and the second curly hair fitting offset index based on the hairstyle parameters and the standard hairstyle parameters of the target hairstyle includes:

[0042] Determine the first curly hair fitting offset index according to the curl curvature and the standard curl curvature of the target hairstyle;

[0043] Determine the second curly hair fitting offset index according to the curl length and the standard curl length of the target hairstyle.

[0044] Further, it also includes:

[0045] When it is determined that the curly hair parameters do not need to be corrected, display the curly hair parameters on the client side;

[0046] When it is determined that the curly hair parameters need to be corrected, display the corrected curly hair parameters on the client side.

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows: through the intelligent analysis of the original hair image by the deep learning model, the hair quality information of the user's hair can be accurately obtained, and a realistic three-dimensional hair map can be generated based on this information. According to the original hair images of different users, personalized hairstyles can be accurately simulated. By comparing the target hairstyle with the simulated image, fine matching in multiple angles and multiple parameters can be achieved. Through the adaptive parameter correction mechanism, key parameters such as the curling temperature, time, and gap can be dynamically adjusted according to the fitting offset index, approaching the target hairstyle to the greatest extent. The client can preview the simulation results in real time, helping the user to more intuitively understand the effects achieved by curling with different curly hair parameters, determine the optimal curly hair parameters for forming the target hairstyle, and improve the curling efficiency of the curling iron.

[0048] Further, by introducing a comparison method of the first curly hair fitting offset index and the second curly hair fitting offset index, multi-dimensional precise evaluation of hairstyle simulation is realized, the differences between the simulated hairstyle and the target hairstyle can be analyzed more comprehensively and meticulously, the accuracy and reliability of hairstyle matching are greatly improved. By strictly comparing the actual curly hair fitting offset index with the preset standard curly hair fitting offset index, the parameters that need to be adjusted can be quickly and accurately identified, the efficiency of hairstyle simulation is greatly improved, the time and cost of manual intervention are reduced, and the scientificity and reliability of hairstyle simulation are significantly enhanced.

[0049] Furthermore, by setting flexible judgment conditions, subtle deviations in hairstyle simulation can be captured in a timely manner, parameters that need to be adjusted can be automatically identified, and parameter adjustment can be carried out in a timely manner, thereby minimizing the difference between the simulated hairstyle and the target hairstyle, ensuring that the simulation result can be as close as possible to the target hairstyle, greatly reducing the workload of manual intervention, and improving the efficiency and accuracy of hairstyle simulation.

[0050] Furthermore, by accurately identifying the specific parameters that need to be corrected and updating the curly hair image in real time during the parameter correction process, users can observe in real time the impact of parameter adjustment on the hairstyle simulation effect. This interactive optimization method greatly improves the user experience of hairstyle design. By presetting the correction amplitude and analyzing the dynamic difference factor, the parameter correction strategy can be flexibly adjusted according to the characteristics of different hair qualities and different hairstyles to adapt to various complex hairstyle simulation scenarios. By judging the correction difference factor, it is ensured that the simulation result can continuously approach the target hairstyle, significantly improving the accuracy and real-time performance of hairstyle simulation.

[0051] Furthermore, by establishing an accurate mapping relationship between the curly hair fitting offset index and specific curly hair parameters, precise control of hairstyle simulation parameters is achieved. The offset indexes in different dimensions directly correspond to specific parameter adjustment strategies, improving the scientificity and pertinence of parameter correction. The corresponding curly hair parameters can be automatically identified and precisely adjusted according to the offset indexes in different dimensions, significantly reducing the cost of manual intervention, improving the efficiency of hairstyle simulation, and ensuring that the simulation result is more comprehensive and accurate.

[0052] Furthermore, by dynamically monitoring the change trend of the correction difference factor, the effect of parameter correction can be evaluated in real time according to the real-time change of the difference factor, meaningless parameter correction can be effectively reduced, and the optimal parameter combination closest to the target hairstyle can be finally found. This not only saves computing resources but also can more accurately lock the parameter configuration closest to the target hairstyle, making the hairstyle simulation process more efficient.

[0053] Furthermore, by performing professional preprocessing on the original hair image, including image enhancement and denoising, the quality and analyzability of the image are significantly improved, laying a solid data foundation for subsequent deep learning analysis. By using a pre-trained deep learning model for hair quality information analysis, hair images with different skin colors and different hair qualities can be processed, and the subtle features of the hair can be accurately identified, making the hair quality information analysis more comprehensive and accurate and improving the accuracy of the simulation.

[0054] Furthermore, by obtaining two key hairstyle parameters, namely the curl rate and the curl length, and strictly comparing the actual hairstyle parameters with the standard hairstyle parameters, the deviation degree of hairstyle simulation can be accurately quantified, providing a scientific and accurate data basis for subsequent parameter optimization and contributing to the precision of hairstyle simulation.

[0055] Furthermore, by simultaneously considering two key parameters, namely the curl curvature and the curl length, and precisely comparing the actual hairstyle parameters with the standard hairstyle parameters, the deviation degree of hairstyle simulation can be more accurately quantified, the differences between the simulated hairstyle and the target hairstyle can be evaluated more comprehensively and meticulously, and the accuracy and reliability of hairstyle simulation are significantly improved.

[0056] Furthermore, by real-time displaying the curl parameters or correcting the curl parameters on the client side, an intuitive and transparent hairstyle simulation interaction experience is provided for users. Users can understand the key parameters of hairstyle simulation in real time. By presenting detailed parameter information, more references and guidance for hairstyle customization are provided for users. Users can adjust the optimal curl parameters of the curling iron based on the displayed parameters to ensure the curling effect. Description of the Drawings

[0057] Figure 1 It is a flowchart of the method for hairstyle simulation based on image recognition according to an embodiment of the present invention;

[0058] Figure 2 It is a flowchart of determining whether the curl parameters need to be corrected according to an embodiment of the present invention;

[0059] Figure 3 It is a flowchart of obtaining the corrected curl parameters according to an embodiment of the present invention;

[0060] Figure 4 It is a structural diagram of the curling iron applied in an embodiment of the present invention. Detailed Embodiments

[0061] In order to make the objectives and advantages of the present invention clearer, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0062] The preferred embodiments of the present invention will be described below with reference to the drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0063] It should be noted that in the description of the present invention, the terms indicating the direction or positional relationship such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the present invention.

[0064] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and defined, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be a direct connection or an indirect connection through an intermediate medium, and it may be the communication inside two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0065] Please refer to Figure 1 , as Figure 1 shown, which is a flowchart of the hairstyle simulation method based on image recognition according to an embodiment of the present invention;

[0066] Specifically, the hairstyle simulation method based on image recognition provided by the embodiment of the present invention includes:

[0067] Obtain the original hair image of the user, and determine the hair quality information based on the original hair image;

[0068] Transmit the hair quality information to the client, and the client performs hair simulation according to the hair quality information to obtain a three-dimensional hair map;

[0069] Determine the curling parameters of the curling iron corresponding to the target hairstyle;

[0070] Perform curling simulation on the three-dimensional hair map according to the determined curling parameters to obtain a curling image;

[0071] Compare the target hairstyle map with the curling image to obtain a curling fitting offset index;

[0072] Determine whether it is necessary to correct the curling parameters based on the curling fitting offset index;

[0073] When it is determined that it is necessary to correct the curling parameters, correct the curling parameters to obtain corrected curling parameters;

[0074] The curling parameters include curling temperature, curling time, and curling gap;

[0075] The curling fitting offset index includes a first curling fitting offset index and a second curling fitting offset index.

[0076] Specifically, the hair quality information is hair information such as hair dryness, hair diameter, hair length, and hair hardness that affects the curling effect. In this embodiment, it is hair dryness, hair diameter, hair length, and hair hardness. The client is a device for displaying a simulated hairstyle, which can be a computer, a mobile phone, etc., as long as it can display the simulated hair image, and it is not limited here. Using three-dimensional graphics rendering technology, based on the hair quality information, a particle system simulation, a hair physical simulation algorithm, curve interpolation reconstruction, and a simulation of hair fiber structure are used to construct a hair model. According to the target hairstyle, corresponding parameters are selected. The target hairstyle is the hairstyle of the curling effect that the user determines to achieve, which can be selected in the client, and the curling parameters of the corresponding curling iron can also be determined in the client. The curling simulation uses the curling parameters of the determined target image to simulate the curling of each hair in the three-dimensional hair image, and simulates the curling deformation of the hair to obtain a curled hair image. The curling fitting offset index is a parameter that characterizes the degree of difference between the hairstyle parameters of the curled hair image simulated after curling according to the user's three-dimensional hair image and the standard hairstyle parameters of the target hairstyle.

[0077] Specifically, by obtaining the user's hair image and determining the corresponding hair quality information according to the hairstyle image, the hair quality information is transmitted to the client. The client simulates a three-dimensional hair image corresponding to the hair quality information, and performs a preliminary curling simulation on the three-dimensional hair image according to the curling parameters determined by the target hairstyle. However, since the curling effect of curling with the same curling parameters may be different due to different hair quality information, the curling fitting offset index is determined by comparing the hairstyle parameters of the curled hair image after curling simulation with the standard parameters of the target hairstyle. Whether it is necessary to correct the curling parameters according to the user's hair quality information to obtain corrected curling parameters suitable for the user's hair quality information is determined according to the size of the curling fitting offset index.

[0078] Specifically, through the intelligent analysis of the original hair image by the deep learning model, the hair quality information of the user's hair can be accurately obtained, and a realistic three-dimensional hair image can be generated based on this information. According to the original hair images of different users, personalized hairstyles can be accurately simulated. By comparing the target hairstyle with the simulated image, fine matching of multiple angles and multiple parameters can be achieved. Through an adaptive parameter correction mechanism, key parameters such as curling temperature, time, and gap can be dynamically adjusted according to the fitting offset index, approaching the target hairstyle to the greatest extent. The client can preview the simulation results in real time, helping the user more intuitively understand the effects achieved by curling with different curling parameters, determining the best curling parameters for forming the target hairstyle, and improving the curling efficiency of the curling iron.

[0079] Please continue to refer to Figure 2 , such as Figure 2 shown, which is the flowchart for determining whether to correct the curling parameters in the embodiment of the present invention;

[0080] Specifically, determining whether the curly hair parameters need to be corrected based on the curly hair fitting offset index includes:

[0081] Comparing the first curly hair fitting offset index with the first standard curly hair fitting offset index to obtain a first comparison result;

[0082] Comparing the second curly hair fitting offset index with the second standard curly hair fitting offset index to obtain a second comparison result;

[0083] Determining whether the curly hair parameters need to be corrected based on the first comparison result and the second comparison result.

[0084] Specifically, the first standard curly hair fitting offset index and the second standard curly hair fitting offset index are the maximum allowable difference degrees of curly hair effects affected by hair quality information.

[0085] Specifically, by introducing a comparison method of the first curly hair fitting offset index and the second curly hair fitting offset index, a multi-dimensional accurate evaluation of hairstyle simulation is realized, which can analyze the differences between the simulated hairstyle and the target hairstyle more comprehensively and meticulously, greatly improving the accuracy and reliability of hairstyle matching. By strictly comparing the actual curly hair fitting offset index with the preset standard curly hair fitting offset index, the parameters that need to be adjusted can be quickly and accurately identified, greatly improving the efficiency of hairstyle simulation, reducing the time and cost of manual intervention, and significantly enhancing the scientificity and reliability of hairstyle simulation.

[0086] Specifically, determining whether the curly hair parameters need to be corrected based on the first comparison result and the second comparison result includes:

[0087] If the first comparison result is that the first curly hair fitting offset index is greater than the first standard curly hair fitting offset index and the second comparison result is that the second curly hair fitting offset index is greater than the second standard curly hair fitting offset index, it is determined that the curly hair parameters need to be corrected when either condition is met.

[0088] Specifically, in the specific implementation process, the threshold of the first standard curly hair fitting offset index is 10%, the threshold of the second standard curly hair fitting offset index is 15%, the first curly hair fitting offset index of the curly hair image is 10%, and the second curly hair fitting offset index is 20%. Since the second curly hair fitting offset index is greater than the second standard curly hair fitting offset index, it indicates that the hair quality information of the user has a greater impact on the curly hair effect, and the curly hair parameters need to be corrected.

[0089] Specifically, by setting flexible judgment conditions, subtle deviations in hairstyle simulation can be captured in a timely manner, parameters that need to be adjusted can be automatically identified, and parameter adjustment can be carried out in a timely manner, thereby minimizing the difference between the simulated hairstyle and the target hairstyle, ensuring that the simulation result can be as close as possible to the target hairstyle, greatly reducing the workload of manual intervention, and improving the efficiency and accuracy of hairstyle simulation.

[0090] Please continue to refer to Figure 3 , such as Figure 3 shown, which is a flowchart for obtaining corrected curly hair parameters in an embodiment of the present invention;

[0091] Specifically, the correcting the curly hair parameters to obtain corrected curly hair parameters includes:

[0092] Determining the curly hair parameters that need to be corrected based on the first comparison result and the second comparison result;

[0093] Correcting the curly hair parameters that need to be corrected using a preset correction amplitude;

[0094] During the process of correcting the curly hair parameters, correcting the curly hair image in real time according to the change of the curly hair parameters to obtain a corrected curly hair image;

[0095] Determining the corrected fitting offset index between the corrected curly hair image and the target hairstyle map during the process of correcting the curly hair parameters;

[0096] Determining the correction difference factor between the corrected fitting offset index and the corresponding first standard curly hair fitting offset index or the second standard curly hair fitting offset index during the process of correcting the curly hair parameters;

[0097] Determining the corrected curly hair parameters according to the correction difference factor.

[0098] Specifically, the correction amplitude is the amplitude size of each correction adjustment for the curly hair temperature, curly hair time, and curly hair gap. The correction amplitude of the curly hair temperature is generally set at 5 - 10 °C, the correction amplitude of the curly hair time is generally set at 1 - 5 seconds, and the correction amplitude of the curly hair gap is generally set at 0.5 - 2 mm. In this embodiment, the correction amplitude of the curly hair temperature is 8 °C, the correction amplitude of the curly hair time is 1.5 seconds, and the correction amplitude of the curly hair gap is 1 mm. Determine the corrected fitting offset index of the curly hair image that gradually changes during the correction process, and determine the correction difference factor of the corrected fitting offset index to determine the corrected curly hair parameters when the correction difference factor is the smallest.

[0099] Specifically, by accurately identifying the specific parameters that need to be corrected and updating the curly hair image in real time during the parameter correction process, users can observe in real time the impact of parameter adjustment on the hairstyle simulation effect. This interactive optimization method greatly improves the user experience of hairstyle design. Through preset correction amplitudes and dynamic difference factor analysis, parameter correction strategies can be flexibly adjusted according to the characteristics of different hair qualities and hairstyles to adapt to various complex hairstyle simulation scenarios. By judging the correction difference factor, it is ensured that the simulation results can continuously approach the target hairstyle, significantly improving the accuracy and real-time performance of hairstyle simulation.

[0100] Specifically, the curly hair parameters that need to be corrected determined based on the first comparison result and the second comparison result include:

[0101] If the first curly hair fitting offset index is greater than the first standard curly hair fitting offset index, adjust the curly hair temperature or the curly hair time;

[0102] If the second curly hair fitting offset index is greater than the second standard curly hair fitting offset index, adjust the curly hair gap.

[0103] Specifically, the curly hair temperature is the temperature at which the curling iron curls the hair. High temperatures are suitable for curling more obvious curls, while low temperatures are suitable for creating soft curls. The curly hair time parameter refers to the time the curling iron stays on the hair. The longer the time, the higher the curling degree. The curly hair gap refers to the curling gap at the curling part of the curling iron. The smaller the interval, the smaller the curling length.

[0104] Specifically, by establishing an accurate mapping relationship between the curly hair fitting offset index and specific curly hair parameters, precise control of hairstyle simulation parameters is achieved. The offset indices in different dimensions directly correspond to specific parameter adjustment strategies, improving the scientific nature and pertinence of parameter correction. Corresponding curly hair parameters can be automatically identified and precisely adjusted according to the offset indices in different dimensions, significantly reducing the cost of manual intervention, improving the efficiency of hairstyle simulation, and ensuring that the simulation results are more comprehensive and accurate.

[0105] Specifically, the determination of the corrected curly hair parameters according to the correction difference factor includes:

[0106] Determine the change of the correction difference factor during the correction of curly hair. If the correction difference factor gradually increases during the correction of curly hair, it is determined that the correction is invalid, and the corrected curly hair parameters are the curly hair parameters;

[0107] If the correction difference factor gradually decreases during the correction process, it is determined that the correction is effective, and the process curly hair parameters before the correction difference factor increases during the correction process are determined as the corrected curly hair parameters.

[0108] In the specific implementation process, the second curly hair fitting offset index is 20%, which is greater than the second standard curly hair fitting offset index threshold of 15%. The curly hair gap is adjusted. During the adjustment process, the corrected fitting offset indexes for increasing and decreasing the curly hair gap according to the correction amplitude are 16% and 17% respectively. This indicates that the correction difference factor gradually increases during the process of correcting the curly hair gap, indicating that the correction is ineffective. Due to the user's hair quality information, the best effect that can be achieved by using this curling iron is the effect of curling with the curling parameters. The corrected curling parameters are still the curling parameters.

[0109] Specifically, by dynamically monitoring the change trend of the correction difference factor, the effect of parameter correction can be evaluated in real time according to the real-time change of the difference factor. This can effectively reduce meaningless parameter corrections and ensure that the optimal parameter combination closest to the target hairstyle can be found eventually. This not only saves computing resources but also can more accurately lock the parameter configuration closest to the target hairstyle, making the hairstyle simulation process more efficient.

[0110] Specifically, the determining of the hair quality information based on the original hair image includes:

[0111] Preprocess the original hair image, including image enhancement and denoising, to obtain the hair image to be processed;

[0112] Use a pre-trained deep learning model to analyze the hair image to be processed and determine the hair quality information of the hair image to be processed.

[0113] Specifically, the hair image is enhanced by methods such as brightness adjustment and color balance, which are not limited here. The hair image is denoised by methods such as Gaussian filtering and median filtering, which are not limited here, so that the pre-trained deep learning model can more accurately identify and analyze the hair quality information. The deep learning model can be a convolutional neural network or other learning models, which are not limited here. The deep learning model for identifying hair quality information is trained with hair images of several different hair quality information.

[0114] Specifically, through professional preprocessing of the original hair image, including image enhancement and denoising, the quality and analyzability of the image are significantly improved, laying a solid data foundation for subsequent deep learning analysis. By using a pre-trained deep learning model for hair quality information analysis, hair images of different skin colors and different hair qualities can be processed, and the subtle features of the hair can be accurately identified, making the hair quality information analysis more comprehensive and accurate, and improving the accuracy of the simulation.

[0115] Specifically, the obtaining of the curly hair fitting offset index by comparing the target hairstyle map with the curly hair image includes:

[0116] Obtain the hairstyle parameters of the curly hair image, where the hairstyle parameters include the curl rate and the curl length;

[0117] Determine the first curly hair fitting offset index and the second curly hair fitting offset index according to the hairstyle parameters and the standard hairstyle parameters of the target hairstyle.

[0118] Specifically, the curl rate is the curling angle of the curly hair, and the curl length is the curling length of a section of curly hair in the hairstyle.

[0119] Specifically, by obtaining two key hairstyle parameters, the curl rate and the curl length, and strictly comparing the actual hairstyle parameters with the standard hairstyle parameters, the deviation degree of the hairstyle simulation can be accurately quantified, providing a scientific and accurate data basis for subsequent parameter optimization and contributing to the precision of the hairstyle simulation.

[0120] Specifically, the determining of the first curly hair fitting offset index and the second curly hair fitting offset index according to the hairstyle parameters and the standard hairstyle parameters of the target hairstyle includes:

[0121] Determine the first curly hair fitting offset index according to the curl rate and the standard curl rate of the target hairstyle;

[0122] Determine the second curly hair fitting offset index according to the curl length and the standard curl length of the target hairstyle.

[0123] Specifically, the first curly hair fitting offset index is the ratio of the absolute value of the difference between the curl rate and the standard curl rate to the standard curl rate, and the second curly hair fitting offset index is the ratio of the absolute value of the difference between the curl length and the standard curl length to the standard curl length.

[0124] In the specific implementation process, the standard curl rate corresponding to the target hairstyle is 50 degrees, the standard curl length is 10 cm, the curl rate of the curly hair image is 45 degrees, and the curl length is 8 cm. Then the first curly hair fitting offset index is |45 - 50| / 50 = 10%, and the second curly hair fitting offset index is |8 - 10| / 10 = 20%.

[0125] Specifically, by considering two key parameters, the curl rate and the curl length, simultaneously and precisely comparing the actual hairstyle parameters with the standard hairstyle parameters, the deviation degree of the hairstyle simulation can be quantified more accurately, and the differences between the simulated hairstyle and the target hairstyle can be evaluated more comprehensively and meticulously, significantly improving the accuracy and reliability of the hairstyle simulation.

[0126] Specifically, it further includes:

[0127] When it is determined that the curly hair parameters do not need to be corrected, display the curly hair parameters on the client side;

[0128] When it is determined that the curly hair parameters need to be corrected, the corrected curly hair parameters are displayed on the client side.

[0129] Specifically, when it is determined that the curly hair parameters do not need to be corrected, it indicates that the influence of the user's own hair quality information on the curly hair effect is relatively small, and the hair can be curled according to the standard curly hair parameters of the target hairstyle. When it is determined that the curly hair parameters need to be corrected, it indicates that the influence of the user's own hair quality information on the curly hair effect is relatively large. After the curly hair parameters are corrected according to the user's hair quality information and the best corrected curly hair parameters for the user to obtain the target hairstyle effect are determined, the hair is curled according to the corrected curly hair parameters.

[0130] Specifically, by displaying the curly hair parameters or the corrected curly hair parameters in real time on the client side, an intuitive and transparent hairstyle simulation interaction experience is provided for the user. The user can understand the key parameters of the hairstyle simulation in real time. By showing detailed parameter information, more references and guidance for hairstyle customization are provided for the user. The user can adjust the best curly hair parameters of the curling iron based on the displayed parameters to ensure the curly hair effect.

[0131] Please continue to refer to Figure 4 as Figure 4 shown, which is a structural diagram of the curling iron applied in the embodiment of the present invention;

[0132] In a specific application, an image recognition-based hairstyle simulation method in an embodiment of the present invention is applied to a curling iron, which includes an upper housing 1, a curling component 2, and a lower housing 3. The lower part of the curling component 2 is connected to the lower housing 3. When not in use, the curling component 2 is inside the upper housing 1. It is powered on by pressing the power-on button 33, and the curling gap of the curling iron is adjusted by adjusting the dial 31. Specifically, the curling gap can be adjusted to seven gears, corresponding to seven different sizes of curling gaps in the range of 1.5 mm - 10.5 mm. Each time the curling gap is adjusted, the curling gap will change by 1.5 mm. The temperature is increased or decreased by pressing the temperature button 32, and each adjustment will adjust the curling temperature by 8°C.

[0133] So far, the technical solution of the present invention has been described in combination with the preferred embodiments shown in the drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0134] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A hairstyle simulation method based on image recognition, characterized in that: include: Obtaining an original hair image of the user, and determining hair quality information based on the original hair image; The hair quality information is transmitted to a client, and the client performs hair simulation according to the hair quality information to obtain a three-dimensional hair image; Determine the curling parameters of the curling iron corresponding to the target hairstyle; Performing curly hair simulation on the three-dimensional hair image according to the determined curly hair parameters to obtain a curly hair image; Comparing the target hairstyle image with the curly hair image to obtain a curly hair fitting deviation index; determining whether the curl parameter needs to be corrected based on the curl fitting deviation index; When it is determined that the curly hair parameter needs to be corrected, correcting the curly hair parameter to obtain a corrected curly hair parameter; The curling parameters include curling temperature, curling time and curling gap; The curl fitting deviation index includes a first curl fitting deviation index and a second curl fitting deviation index; The determining whether the curly hair parameter needs to be corrected based on the curly hair fitting deviation index comprises: comparing the first curl fitting deviation index with a first standard curl fitting deviation index to obtain a first comparison result; comparing the second curl fitting deviation index with a second standard curl fitting deviation index to obtain a second comparison result; determining whether the curling parameter needs to be corrected based on the first comparison result and the second comparison result; The modifying the curly hair parameter to obtain the modified curly hair parameter comprises: Determining the curly hair parameter that needs to be corrected based on the first comparison result and the second comparison result; Use the preset correction range to correct the curly hair parameters that need to be corrected; In the process of correcting the curly hair parameters, the curly hair image is corrected in real time according to the change of the curly hair parameters to obtain a corrected curly hair image; determining a corrected fitting offset index between the corrected curly hair image and the target hairstyle image in the process of correcting the curly hair parameters; determining a correction difference factor between the corrected fitting deviation index and the corresponding first standard curly hair fitting deviation index or second standard curly hair fitting deviation index during the correction of curly hair parameters; determining a corrected curl parameter according to the corrected difference factor; Determining the corrected curly hair parameter according to the corrected difference factor comprises: determining a change in a correction difference factor during the curly hair correction process, and if the correction difference factor gradually increases during the curly hair correction process, determining that the correction is invalid, and the correction curly hair parameter is the curly hair parameter; If the correction difference factor gradually decreases during the correction process, it is determined that the correction is effective, and the process curling parameters before the correction difference factor increases during the correction process are determined as the correction curling parameters; The step of comparing the target hairstyle image with the curly hair image to obtain the curly hair fitting deviation index comprises: Acquire the hairstyle parameters of the curly hair image, where the hairstyle parameters include curl rate and curl length; A first curl fitting deviation index and a second curl fitting deviation index are determined according to the hairstyle parameters and the standard hairstyle parameters of the target hairstyle.

2. The method for simulating hairstyle based on image recognition according to claim 1, characterized in that: The determining whether the curly hair parameter needs to be corrected based on the first comparison result and the second comparison result includes: If the first comparison result is that the first curl fitting deviation index is greater than the first standard curl fitting deviation index and the second comparison result is that the second curl fitting deviation index is greater than the second standard curl fitting deviation index, it is determined that the curl parameter needs to be corrected.

3. The method for simulating hairstyle based on image recognition according to claim 1, characterized in that: The determining of the curly hair parameter that needs to be corrected based on the first comparison result and the second comparison result includes: If the first curl fitting deviation index is greater than the first standard curl fitting deviation index, adjusting the curling temperature or adjusting the curling time; If the second curl fitting deviation index is greater than the second standard curl fitting deviation index, the curl gap is adjusted.

4. The method for simulating hairstyle based on image recognition according to claim 1, characterized in that: Determining the hair quality information based on the original hair image includes: Preprocessing the original hair image to obtain a hair image to be processed; The pre-trained deep learning model is used to analyze the hair image to be processed to determine the hair quality information of the hair image to be processed.

5. The method for simulating hairstyle based on image recognition according to claim 1, characterized in that: The determining of the first curl fitting deviation index and the second curl fitting deviation index according to the hairstyle parameters and the standard hairstyle parameters of the target hairstyle comprises: determining a first curl fitting deviation index according to the curl ratio and a standard curl ratio of a target hairstyle; A second curl fitting deviation index is determined according to the curl length and a standard curl length of the target hairstyle.

6. The method for simulating hairstyle based on image recognition according to claim 5, characterized in that: Also includes: When it is determined that the curly hair parameters do not need to be corrected, displaying the curly hair parameters on the client; When it is determined that the curly hair parameters need to be corrected, the corrected curly hair parameters are displayed on the client.

Images