3D Image Processing With Region-Adaptive Radiance Field Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional technologies for estimating three-dimensional information face accuracy issues when the number of learning parameters is small for the resolution of captured images, leading to reduced image quality of virtual viewpoint images.
Innovation Solution
An image processing apparatus that sets a learning model with higher pixel resolution for specific partial regions, adjusting the number of learning parameters per volume based on the pixel resolution, to improve the accuracy of three-dimensional information estimation while reducing computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of learning parameters is increased to improve the accuracy of three-dimensional information estimation, then the image quality of virtual viewpoint images is improved, but the computational requirements and processing time increase
Solution Approach 1:
The patent applies local quality by differentiating the learning parameter count based on spatial location. High-resolution regions (where objects are located) use a larger number of learning parameters to capture fine details, while low-resolution regions (background areas) use fewer learning parameters. This spatially adaptive approach allocates computational resources efficiently, improving accuracy where needed while maintaining overall computational efficiency.
2Manufacturing precision
If the number of learning parameters is increased to improve image quality, then the accuracy of radiance field estimation is improved, but the amount of computation required increases
Solution Approach 1:
The system implements local quality by setting different learning parameter counts for different spatial regions. Regions containing objects use higher learning parameter counts to achieve accurate radiance field estimation, while empty or background regions use lower learning parameter counts. This reduces the overall computational complexity while maintaining high accuracy in critical areas.
3Measurement precision
If uniform high learning parameter count is used across all regions, then the accuracy is improved, but the computational overhead increases significantly
Solution Approach 1:
The patent employs local quality by adapting the learning parameter count to the local importance of each region. High-resolution regions with objects receive higher learning parameter counts for accurate estimation, while low-resolution background regions receive fewer parameters. This selective approach minimizes computational resource consumption while maintaining high accuracy where it matters most.
Data Source
AI summary
An object is to estimate three-dimensional information capable of generating virtual viewpoint images with high image quality while reducing the number of learning parameters for the estimation of the three-dimensional information. An image processing apparatus: obtains a plurality of captured images obtained by image capturing of an image capturing region from a plurality of directions; sets at least one partial region in the image capturing region; sets a learning model corresponding to the partial region such that the higher a pixel resolution for the partial region in each of the plurality of captured images, the larger the number of learning parameters per volume; and trains the learning model by using the plurality of captured images.