An intelligent hairstyle matching and realistic image synthesis system and method based on multi-dimensional portrait feature perception

Abstract

The application discloses a kind of intelligent hairstyle matching and real image synthesis system and method based on multidimensional portrait feature perception, it is related to artificial intelligence image recognition and intelligent hairdressing technical field.System includes standardization image acquisition hardware unit, face contour and hair volume multi-feature identification module, hairstyle data model analysis and recommendation module, hairstyle image adaptive fusion synthesis module and hairdresser terminal synchronous module.The present application is through high-precision portrait acquisition, multidimensional hairdressing feature quantitative identification, structured hairstyle data model matching, head shape adaptive image fusion synthesis, realizes scientific hairstyle recommendation and high real effect preview, and the final effect is synchronized to hairdresser terminal for guiding pruning.The present application is accurate, synthesis real, soft and hard integration, and has strong landing, can significantly reduce hairdressing communication error, improve service standardization level, and is suitable for the intelligent upgrading of various hairdressing stores.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence image recognition, facial feature quantitative analysis, hairstyle data modeling, realistic image fusion, and intelligent hairdressing hardware interaction technology. Specifically, it relates to a closed-loop intelligent hairdressing assistance system and method that integrates multi-dimensional facial feature perception, intelligent hairstyle matching, and image synthesis for hairdressing service scenarios. Background Technology

[0002] Currently, in the hairdressing industry, customer hairstyle selection heavily relies on the hairdresser's subjective experience, leading to high communication costs, significant expectation discrepancies, and a high risk of service disputes. Existing hairstyle try-on software or applications only achieve simple 2D texture overlay without adaptive fitting based on real head structure, resulting in distorted composite effects. They lack quantitative detection of key hairdressing features such as hair density, hairline shape, and head-shoulder ratio; they haven't established structured, computable hairstyle data models and intelligent matching algorithms; they lack standardized data collection hardware support, resulting in poor recognition stability and scenario applicability; and they cannot form a complete technical closed loop from data collection, recognition, recommendation, synthesis to hairdresser execution, thus lacking professional hairdressing guidance value. Therefore, the industry urgently needs a high-precision, integrated hardware and software intelligent hairstyle analysis and effect synthesis system that can be directly implemented. Summary of the Invention

[0003] 1. Achieve automated, high-precision, and quantitative recognition of facial contours, face shape classification, hair density, hairline shape, and head-shoulder ratio.

[0004] 2. Build a standardized, tagged, and scalable hairstyle data model and intelligent matching engine to achieve scientific and accurate matching between hairstyles and user characteristics.

[0005] 3. Based on the key points of the head shape contour, the system achieves adaptive fitting of hairstyle, unified lighting and shadow, and fusion of realistic images to generate a preview image that can be directly used for hairdressing guidance.

[0006] 4. Provide dedicated standardized data acquisition hardware and interactive terminals to improve system stability, robustness, and scenario applicability.

[0007] 5. Construct a closed-loop technology system covering the entire process of data collection, identification, recommendation, synthesis, and execution to reduce communication errors and improve the standardization of hairdressing services. Technical solution

[0008] A smart hairstyle matching and realistic image synthesis system based on multi-dimensional facial feature perception includes: a standardized image acquisition hardware unit, a facial contour and hair volume multi-feature recognition module, a hairstyle data model analysis and recommendation module, a hairstyle image adaptive fusion and synthesis module, and a hairdresser terminal synchronization module. The system forms a complete intelligent hairdressing assistance process through hardware acquisition, feature recognition, model matching, image synthesis, and terminal execution. The standardized image acquisition hardware unit includes a high-definition fixed-focus camera, an adjustable brightness ring lighting module, a face positioning and fixing bracket, a touch-interactive terminal, and a data transmission unit, achieving standardized facial image acquisition with uniform posture, lighting, and distance. The facial contour and hair volume multi-feature recognition module includes a facial key point detection unit, a facial contour calculation unit, a hair semantic segmentation unit, a hair volume texture density analysis unit, a hairline shape recognition unit, and a head-shoulder ratio calculation unit, which can output structured feature data such as face shape classification, hair volume level, hairline type, and head-shoulder ratio. The hairstyle data model analysis and recommendation module includes a structured hairstyle tag library. Tag dimensions include gender, length, layering, curl, bangs type, volume, suitable face shape, suitable hair volume, and suitable hairline. A feature vector weight matching algorithm is used to calculate the similarity between user features and the hairstyle library and output the results sorted by suitability. The hairstyle image adaptive fusion and synthesis module includes head shape adaptive mesh deformation units, affine transformation units, illumination consistency correction units, hair color matching units, and edge feathering fusion units. This allows the hairstyle model to accurately fit the head shape, generating a highly realistic haircut preview image. The hairdresser terminal synchronization module pushes the user-confirmed preview image, hairstyle parameters, and trimming points to the hairdresser's terminal as the trimming execution standard.

[0009] S1. Capture the user's frontal portrait using standardized hardware; S2, Image Preprocessing and Multi-Feature Recognition; S3. Match hairstyle data model and generate a recommendation list; S4. The user selects a hairstyle and performs image compositing; S5. After confirming the effect, push it to the hairdresser's terminal; S6. The hairdresser completes the trimming based on the rendering. Beneficial effects

[0010] 1. Multi-dimensional feature joint recognition, high accuracy, strong robustness, and adaptability to different usage environments; 2. Hairstyle recommendations are scientific, objective, quantifiable, and scalable, significantly improving the suitability of the match; 3. The composite effect is realistic and practical, and can be directly used as a standard for hairdressing, reducing communication errors; 4. The integrated hardware and software design makes deployment simple and operation intuitive, suitable for all types of hair salons; 5. Form a complete service loop, reduce reliance on hairdressers' experience, and improve the standardization level of the industry; 6. It can accumulate user characteristics and hairstyle preference data, supporting continuous iteration and optimization of the model. Attached Figure Description Figure 1 This is a system overall architecture block diagram according to an embodiment of the present invention; Figure 2 This is a flowchart of the face contour and hair volume recognition process according to an embodiment of the present invention; Figure 3 This is a flowchart of the hairstyle data model and recommendation algorithm according to an embodiment of the present invention; Figure 4 This is a flowchart of the adaptive fusion and synthesis process of hairstyle images according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the hardware composition structure of an embodiment of the present invention; Figure 6 This is a flowchart illustrating the overall system workflow of an embodiment of the present invention.

Claims

1. A smart hairstyle matching and realistic image synthesis system based on multi-dimensional facial feature perception, characterized in that, include: The system comprises a standardized image acquisition hardware unit, a facial contour and hair volume multi-feature recognition module, a hairstyle data model analysis and recommendation module, a hairstyle image adaptive fusion and synthesis module, and a hairdresser terminal synchronization module. The standardized image acquisition hardware unit captures standardized portraits of users in fixed poses and under fixed lighting. The facial contour and hair volume multi-feature recognition module performs facial key point detection, contour extraction, face shape classification, hair region segmentation, hair density estimation, and hairline shape recognition on the acquired images, outputting quantified user feature data. The hairstyle data model analysis and recommendation module performs weighted matching calculations based on a structured labeled hairstyle library and user feature data to generate suitable hairstyle recommendations. The hairstyle image adaptive fusion and synthesis module adaptively fits the user-selected hairstyle model according to facial contours and head shape structure, performs lighting correction, and image fusion to generate a realistic haircut preview. The hairdresser terminal synchronization module pushes the user-confirmed preview and hairstyle parameters to the hairdresser's terminal to guide the actual haircut. The system forms a complete intelligent haircutting assistance closed loop through hardware acquisition, feature recognition, model recommendation, image synthesis, and terminal execution.

2. The system according to claim 1, characterized in that, The standardized image acquisition hardware unit includes a high-definition fixed-focus camera, an adjustable brightness ring light module, a face positioning and fixing bracket, a touch interactive display terminal, and a data transmission unit.

3. The system according to claim 1, characterized in that, The facial contour and hair volume multi-feature recognition module includes a 68-point or 468-point facial key point detection unit, a facial contour calculation unit, a hair semantic segmentation unit, a hair volume texture density analysis unit, a hairline shape recognition unit, and a head-shoulder ratio calculation unit.

4. The system according to claim 3, characterized in that, The face shape classification includes oval face, round face, square face, oblong face, heart-shaped face and diamond face; the hair density includes at least three quantitative output levels: sparse, medium and dense.

5. The system according to claim 1, characterized in that, The hairstyle data model analysis and recommendation module includes a structured hairstyle tag library, with tag dimensions including gender, length, layering, curliness, bangs type, volume, suitable face shape, suitable hair volume, and suitable hairline.

6. The system according to claim 5, characterized in that, The recommendation module uses a feature vector weight matching algorithm to calculate the similarity between user feature data and hairstyle tags, and outputs the Top 3 to Top 6 recommended hairstyles according to their suitability.

7. The system according to claim 1, characterized in that, The hairstyle image adaptive fusion synthesis module includes a head shape adaptive mesh deformation unit, an affine transformation unit, an illumination consistency correction unit, a hair color matching unit, and an edge feathering fusion unit.

8. The system according to claim 1, characterized in that, The image synthesis module outputs high-definition before-and-after comparison images, which have standard reference value for directly guiding barbers in trimming.

9. The system according to claim 1, characterized in that, The system is compatible with both 2D monocular camera acquisition mode and 3D depth camera / structured light acquisition mode.

10. A method for intelligent hairstyle matching and realistic image synthesis based on multi-dimensional facial feature perception, characterized in that, Includes the following steps: S1. Acquire frontal portrait images of users using standardized hardware; S2. Image preprocessing, facial landmark detection, contour extraction, face shape classification, hair segmentation, hair volume estimation and hairline shape recognition; S3. Input the quantified features into the hairstyle data model, perform feature weight matching, and generate a recommended hairstyle list; S4. Based on the hairstyle selected by the user, adaptive head shape fitting and image fusion synthesis are performed to generate a highly realistic preview image; S5. After the user confirms the final result, the effect image and hairstyle parameters are pushed to the hairdresser's terminal to guide the haircutting process.

11. The method according to claim 10, characterized in that, In step S2, hair volume estimation is based on a comprehensive calculation of hair region texture features, grayscale distribution, and depth information.

12. The method according to claim 10, characterized in that, The image synthesis in step S4 includes mesh deformation, lighting unification, color matching, and edge feathering fusion processing.

Images