Facial beauty auxiliary system based on artificial intelligence

Abstract

The invention relates to the technical field of image processing, in particular to a facial beautification auxiliary system based on artificial intelligence, and the system comprises an acquisition module which carries out the preprocessing of a facial image, and obtains a preprocessed image; the facial mark alignment module is used for extracting identification points from the preprocessed image and aligning the preprocessed image with the shaping template according to the identification points to form a combined image; the feature extraction module is used for extracting facial features from the combined image; and the senescence prediction module is used for performing senescence prediction according to the facial features so as to obtain a senescence prediction image and generate senescence evaluation suggestions. In order to solve the problems that in the prior art, the evaluation process of beauty treatment is too subjective, and the population characteristics of Asian population are not considered, in the scheme, a specifically-designed plastic template is introduced, and the facial image of a patient is measured based on the plastic template in the planning process, so that the evaluation accuracy of the beauty treatment is improved. And the facial features needing to be corrected are determined and related suggestions are given, so that the problem that subjective evaluation is not accurate enough is solved.

Claims

1. A facial beauty assistance system based on artificial intelligence, characterized in that: include: An acquisition module, wherein the acquisition module uses a 3D camera to acquire a facial image of the patient, and preprocesses the facial image to obtain a preprocessed image; A facial landmark alignment module, the facial landmark alignment module is connected to the acquisition module, the facial landmark alignment module extracts identification points from the pre-processed image, and aligns the pre-processed image with a plastic surgery template according to the identification points to form a combined image; A feature extraction module, the feature extraction module is connected to the facial landmark alignment module, and the feature extraction module extracts facial features from the combined image; An aging prediction module, the aging prediction module is connected to the feature extraction module, and the aging prediction module performs aging prediction according to the facial features to obtain an aging prediction image; The aging prediction module generates an aging assessment suggestion according to the aging prediction image.

2. The facial beauty assistance system according to claim 1, characterized in that: The acquisition module comprises: A camera calibration module, which collects camera position information of the 3D camera after the 3D camera is set up; The camera calibration module calibrates the extrinsic parameter matrix of the 3D camera according to the camera pose information; An image capturing module, wherein the image capturing module is connected to the camera calibration module, and the image capturing module controls the 3D camera to capture the facial image and adds the external parameter matrix to form three-dimensional image information; An image quality assessment module, the image quality assessment module is connected to the image capture module, and the image quality assessment module evaluates the three-dimensional image information according to quality assessment dimensions to form an image quality score; An image correction module is connected to the image quality assessment module, and processes the three-dimensional image information according to the image quality score to obtain the pre-processed image.

3. The facial beauty assistance system according to claim 1, characterized in that: The facial landmark alignment module comprises: A feature pre-extraction module, wherein the feature pre-extraction module extracts features from each image in the pre-processed image through a series of convolutional networks to obtain a feature map; A feature classification module, the feature classification module is connected to the feature pre-extraction module, and the feature classification module classifies the feature map to determine a plurality of identification points; An alignment module is connected to the feature classification module, and the alignment module matches the plastic template according to the identification points, and scales and rotates the plastic template during the matching process to obtain the combined image.

4. The facial beauty assistance system according to claim 3, characterized in that: The alignment module comprises: A marker point classification module, wherein the marker point classification module classifies the marker points according to their positions on the face to form marker point groups; A vector assembly module, the vector assembly module is connected to the identification point classification module, and the vector assembly module forms identification point vectors according to identification point grouping; A template search module, the template search module is connected to the vector assembly module, and the template search module searches for a corresponding template vector for the identification point vector; A template scaling module, the template scaling module is connected to the template searching module, and the template scaling module calculates a module length ratio between the identification point vector and the corresponding template vector; The template scaling module scales the shaping template according to the template length ratio to obtain a scaled template; A template adjustment module is connected to the template scaling module, and the template adjustment module rotates and moves the scaling template so that all the identification point vectors match the corresponding template vectors to form the combined image.

5. The facial beauty assistance system according to claim 1, characterized in that: The feature extraction module comprises: A visual transformer, wherein the visual transformer extracts facial features in the matching image, thereby extracting facial features to be processed; An attention module, the attention module is connected to the visual transformer, and the attention module assigns a corresponding attention weight to each group of facial features to be processed; A feature output module, wherein the feature output module is connected to the attention module, and the feature output module generates the facial feature according to the attention weight and the facial feature to be processed.

6. The facial beauty assistance system according to claim 1, characterized in that: The aging prediction module comprises: A generator network, in a first prediction process, the generator network predicts an aging image according to the facial features to form an intermediate prediction image; a discriminator, the discriminator being connected to the generator, and the discriminator judging whether the aging simulation is correct according to the intermediate prediction image and generating an aging prediction judgment result; The generator network regenerates the intermediate prediction image according to the aging prediction discrimination result; The discriminator outputs the intermediate prediction image as the aging prediction image when the intermediate prediction image satisfies an output condition.

7. The facial beauty assistance system according to claim 1, characterized in that: The aging prediction module is also provided with a suggestion generation module, and the suggestion generation module includes: A part determination module, wherein the part determination module determines an adjustment part that needs to be adjusted according to the facial features and the plastic surgery template; a parameter determination module, the parameter determination module being connected to the part determination module, and the parameter determination module classifying the adjustment part and the facial features to determine the adjustment parameters; An aging correction module is connected to the parameter determination module, and the aging correction module corrects the adjustment parameters according to the aging prediction image to form a processing suggestion output.

8. The facial beauty assistance system according to claim 1, characterized in that: Also included is a feedback loop module; The feedback loop module includes a plurality of memory modules; Each of the memory modules is respectively provided with a memory gate and a memory classification module; The first input end of the memory gate is connected to the output end of the memory module of the previous stage; The first input end of the memory gate collects the shaping judgment information of the previous moment; The second input end of the memory gate receives current patient feedback information; The memory gate extracts features from the plastic surgery judgment information and the patient feedback information to form memory features; The memory classification module is connected to the memory gate; The memory classification module classifies the memory features to form adjustment suggestions; The memory classification module discards the adjustment suggestions according to a preset ratio and outputs them as new plastic surgery judgment information.

Images