Parametric Reshaping of Portrait Images for Weight-change

Area of Science:

• Computer Vision
• Image Processing
• Computer Graphics

Background:

• Digital image manipulation is prevalent.
• Facial appearance editing requires sophisticated techniques to maintain realism.
• Existing methods may produce artifacts or lack global consistency.

Purpose of the Study:

• To develop an easy-to-use parametric method for thinning or fattening faces in single portrait images.
• To ensure the edited face maintains high similarity to the original image.
• To achieve globally consistent editing effects without noticeable artifacts.

Main Methods:

• 3D face reconstruction from a single image using a morphable model.
• Calculation of new feature point positions based on linear regression and user-set weight-change parameters.
• Application of Laplacian deformation for non-feature points in the 3D model.
• Image retargeting using mesh parametrization for seamless blending with the background.

Main Results:

• The proposed method enables parametric control over facial thinning and fattening.
• Globally consistent face reshaping is achieved with minimal artifacts.
• Seamless integration of the edited face into the original image background.
• Experimental validation and user studies confirm the algorithm's effectiveness.

Conclusions:

• The presented parametric image retouching method offers an effective solution for realistic facial appearance editing.
• The approach successfully balances user control with the preservation of facial identity and image integrity.
• This technique has potential applications in portrait editing and digital media.