Dynamic 3D Model Sequence Compression via 4D Fusion Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voxel capture systems generate a large amount of 3D model data for dynamic 3D model sequences, making it difficult to export and store, especially for mobile applications.
Innovation Solution
A method for efficiently compressing dynamic 3D model sequences using 4D fusion, which involves storing a reference model and determining initial and final fusion parameters through model alignment and iterative optimization, allowing the reference model to deform into the target model using a small number of parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voxel capture systems generate complete 3D content for dynamic 3D model sequences, then the completeness and quality of 3D content is improved, but the data volume and storage requirements increase significantly
Solution Approach 1:
The patent segments the 3D model into a reference model and multiple target models, processing them separately through model alignment and fusion. This segmentation allows the system to maintain complete 3D content quality while reducing overall data volume by reusing the reference model across multiple frames.
Solution Approach 2:
The patent performs preliminary model alignment to establish initial correspondence between reference and target models before final fusion. This preliminary action pre-computes transformation parameters, reducing the computational and storage burden for generating complete dynamic 3D sequences.
2Reliability
If the system processes and stores complete dynamic 3D model sequences, then the quality and detail of motion capture is improved, but the complexity of data management and processing increases
Solution Approach 1:
The patent extracts only the essential deformation information from target models through model alignment, separating the static reference model from the dynamic deformation parameters. This extraction reduces data management complexity by storing only necessary transformation data rather than complete model sequences.
Solution Approach 2:
The patent transforms the 3D model representation from complete geometric data to parameter-based fusion models, where target models are represented by transformation parameters relative to the reference model. This parameter change simplifies data management while preserving motion capture quality.
3Manufacturing precision
If the patent uses model alignment and iterative optimization to determine fusion parameters, then the accuracy of 3D model transformation is improved, but the computational time and processing resources increase
Solution Approach 1:
The patent performs preliminary model alignment to establish initial vertex correspondence and rough transformation parameters before iterative optimization. This preliminary action provides a good starting point for optimization, reducing the number of iterations needed and thus computational time while maintaining accuracy.
Solution Approach 2:
The patent segments the optimization process into model alignment phase and iterative fusion phase, processing vertices in blocks rather than all at once. This segmentation reduces memory requirements and allows parallel processing, improving computational efficiency without sacrificing transformation accuracy.
Data Source
AI summary
The present disclosure relates to a method for efficiently compressing a dynamic 3D model sequence based on 4D fusion. In some embodiments, comprising: storing a 3D model of one frame of the dynamic 3D model sequence as a reference model; determining an initial correspondence between vertices of the reference model and a target model to align the reference model and the target model by optimizing an energy function of the reference model so as to obtain initialized fusion parameters, wherein the target model is a 3D model of remaining frames of the dynamic 3D model sequence; and determining the final fusion parameters for deforming the reference model into the target model by iteratively optimizing the initialized fusion parameters.