3D Face Grid Fusion for Authentic Image Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image fusion methods based on 2D models often result in poor authenticity of the final target face image due to mismatched face angles and incomplete user face information, leading to unrealistic fusion effects.
Innovation Solution
The method involves obtaining source and resource face image data, performing image recognition to extract feature points, generating three-dimensional grids, and fusing these grids to create a target face grid, followed by complexion fusion using source and resource complexion data, all based on a three-dimensional model to enhance authenticity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If image fusion is performed using 2D models, then the processing complexity is reduced, but the authenticity and realism of the target face image deteriorates
Solution Approach 1:
The patent transitions from 2D image processing to 3D model-based processing. By constructing three-dimensional face models from source and resource images, the system performs fusion operations in 3D space, which preserves spatial relationships and facial geometry more accurately, thereby improving authenticity while managing complexity through structured 3D representation.
Solution Approach 2:
The patent employs curved surface modeling to represent facial geometry. By using 3D mesh models with curved surfaces instead of flat 2D planes, the system captures the natural curvature and depth of facial features, resulting in more realistic fusion outcomes that maintain anatomical accuracy.
2Manufacturing precision
If 3D model-based image fusion is performed, then the authenticity of the target face image is improved, but the device complexity increases
Solution Approach 1:
The patent divides the face into multiple regions (skin, hair, eyes, lips, etc.) and processes each region separately using region-specific fusion parameters. This segmentation allows complex 3D fusion to be broken down into manageable sub-tasks, reducing overall processing complexity while maintaining high authenticity through specialized handling of each facial component.
Solution Approach 2:
The patent applies different fusion strategies and parameters to different regions of the face. By tailoring the fusion approach to local characteristics (e.g., different blending modes for skin versus hair, different geometric transformations for eyes versus lips), the system achieves high authenticity without uniformly applying complex processing across the entire face.
3Loss of information
If face angle alignment is performed in 3D space, then the completeness of user face information is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary 3D model construction and face angle normalization before the actual fusion operation. By pre-aligning the source and resource 3D models to a standard coordinate system and pre-computing transformation matrices, the system reduces the computational burden during real-time fusion, thereby minimizing processing time while ensuring complete face information alignment.
Data Source
AI summary
Embodiments of this application disclose an image fusion method performed by a computing device. The method includes the following steps: obtaining source face image data of a current to-be-fused image and resource configuration information of a current to-be-fused resource, performing image recognition processing on the source face image data, to obtain source face feature points corresponding to the source face image data, and generating a source face three-dimensional grid of the source face image data according to the source face feature points, performing grid fusion by using a resource face three-dimensional grid and the source face three-dimensional grid to generate a target face three-dimensional grid, and performing face complexion fusion by using source complexion data of the source face image data and resource complexion data of resource face image data on the target face three-dimensional grid, to generate fused target face image data.