Elastic Registration for Personalized Facial Wrinkle Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for simulating facial skin aging fail to accurately predict an individual's unique wrinkle patterns based on their specific characteristics, such as skin type and ethnicity, often relying on generic masks or population norms, which limits the realism and accuracy of the aging simulation.
Innovation Solution
A computer-implemented method that captures neutral and expression images of a face, performs elastic registration and mapping to transport wrinkles from an expression image to a neutral image, using image mosaicing techniques to eliminate border artifacts and generate a wrinkle-changed facial image, allowing for personalized wrinkle aging simulation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If generic masks or population norms are used for aging simulation, then the simulation process is simplified and faster, but the accuracy and realism of the individual's unique wrinkle patterns deteriorates
Solution Approach 1:
The patent segments the aging simulation process into distinct components: capturing neutral and expression images, performing elastic registration to map expression wrinkles to neutral state, and compositing the results. This segmentation allows the system to maintain individual uniqueness while streamlining the overall process through automated pipeline integration.
Solution Approach 2:
The system performs preliminary actions by capturing both neutral and expression images before the actual aging simulation. The elastic registration and wrinkle extraction are pre-computed and stored, enabling rapid generation of aged images without repeating the complex computation each time, thus improving productivity while maintaining accuracy.
2Measurement precision
If elastic registration and patch-by-patch mapping are performed to accurately capture individual wrinkle patterns, then the realism and precision of the simulation improves, but the computational complexity and processing time increases
Solution Approach 1:
The patent divides the facial image into multiple patches and performs elastic registration and mapping on each patch independently. This segmentation approach simplifies the complex global transformation problem into manageable local operations, improving computational efficiency while maintaining high precision in wrinkle pattern capture.
Solution Approach 2:
The system creates a copy of the expression image's wrinkle patterns and maps them onto the neutral image through elastic registration. This copying approach allows the system to preserve the original images while generating the aged simulation, avoiding the need for complex in-place transformations and reducing processing complexity.
3Manufacturing precision
If image mosaicing techniques are used to eliminate border artifacts, then the visual quality and realism of the simulated image improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies image mosaicing techniques selectively to specific regions where border artifacts are most likely to occur, rather than processing the entire image uniformly. This partial action approach maintains high visual quality in critical areas while reducing overall processing time and computational resource requirements.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods and systems are disclosed to realistically simulate facial wrinkle aging of a person using a neutral state (natural look) image and one or more expression images (e.g., smile, frown, pout, wink) that induce wrinkles. The neutral and expression images are processed to simulate wrinkle aging by registering the wrinkles that are visible in the expression image onto the neutral image, thereby generating a wrinkle-aged simulated image. Advantageously, a person's own wrinkle histological data is utilized, hence providing an accurate and realistic wrinkle aging simulation. Similarly, the neutral image is processed to eliminate all visible wrinkles thereby generating a wrinkle de-aged simulation image. Additionally, blending of a neutral image with an aged or de-aged simulation image is disclosed, where the degree of blending is based on statistical modeling of skin condition with age and/or expected outcome of a particular type of treatment. The methods and systems disclosed have wide applicability, including, for example, areas such as dermatology, cosmetics and computer animation, among others.