SketchAging: Face Photo-Sketch Synthesis and Aging With Multi-Scale Feature Extraction.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Image Processing

Background:

• Generative Adversarial Networks (GANs) have advanced facial sketch generation and age transformation.
• Current methods often lack integration for cross-domain synthesis (sketch and age).
• Preserving identity while generating high-quality images remains a challenge.

Purpose of the Study:

• To propose an integrated framework for continuous age transformation and facial sketch generation.
• To address limitations in current cross-domain synthesis and identity preservation.
• To enhance realism and accuracy in facial image manipulation.

Main Methods:

• Developed Multi-Scale Feature Extraction Networks (MSFE) for image-to-image translation.
• Utilized a Dual Conditional Normalization Attention (DCNA) architecture.
• Integrated sketch feature extraction and facial image encoding into a pre-trained StyleGAN latent space.

Main Results:

• The proposed MSFE approach demonstrated superior performance over existing methods.
• Achieved enhanced realism and identity preservation in facial photo-sketch synthesis.
• Showcased improved age accuracy in generated images.

Conclusions:

• The MSFE framework effectively enables continuous age transformation with sketch synthesis.
• The method successfully maintains identity and stylistic characteristics across domains.
• Offers a promising solution for forensic and digital media applications requiring accurate facial reconstruction.