Irradiance Volume Accuracy Filtering for Medical Imaging Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current global illumination algorithms in medical imaging introduce grainy noise due to stochastic lighting simulation, which can obscure image details and are computationally expensive to reduce, especially when filtering three-dimensional volumetric data in real-time.
Innovation Solution
An image processing method that calculates an accuracy measure for each point in the irradiance volume, allowing for targeted filtering based on lighting accuracy, where lower-accuracy regions are filtered more aggressively to reduce noise while preserving details in higher-accuracy areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If stochastic lighting simulation is used in global illumination algorithms, then lighting realism is improved, but grainy noise is introduced into the image
Solution Approach 1:
The patent applies local quality by differentiating between high-accuracy and low-accuracy regions in the irradiance volume. High-accuracy regions (where more light paths contribute) retain stochastic simulation details for realism, while low-accuracy regions (where fewer light paths contribute) are smoothed to reduce grainy noise. This selective application of filtering based on local accuracy metrics resolves the contradiction between maintaining lighting realism and reducing noise.
2Object-affected harmful factors
If the number of iterations is increased to reduce grainy noise, then noise reduction is improved, but computational cost increases
Solution Approach 1:
The patent changes the parameter being optimized from the number of iterations to a per-region accuracy metric. Instead of uniformly increasing iterations across the entire volume (which is computationally expensive), the system calculates an accuracy metric for each region and applies selective filtering. This parameter change enables efficient noise reduction by focusing computational effort only where needed, rather than uniformly across all regions.
Solution Approach 2:
The patent applies partial action by selectively filtering only the low-accuracy regions of the irradiance volume rather than processing the entire volume uniformly. This approach reduces grainy noise in problematic areas while avoiding the computational overhead of processing high-accuracy regions that already have sufficient light path sampling. The selective application of filtering represents partial action that achieves noise reduction with minimal additional computational cost.
3Object-affected harmful factors
If filtering is applied to reduce grainy noise, then noise reduction is improved, but image detail may be lost
Solution Approach 1:
The patent preserves image detail by applying filtering selectively based on local accuracy metrics. High-accuracy regions, which contain important image details and sufficient light path information, are excluded from filtering to prevent detail loss. Low-accuracy regions, which typically correspond to areas with fewer light paths and less critical detail, are filtered to reduce noise. This local differentiation ensures that filtering reduces noise without sacrificing important image details.
4Object-affected harmful factors
If three-dimensional volumetric filtering is performed, then noise reduction is improved, but processing complexity increases
Solution Approach 1:
The patent extracts the filtering operation from the complex three-dimensional volumetric domain and applies it selectively to two-dimensional slices or regions of the irradiance volume. By identifying low-accuracy regions and applying filtering only to those specific areas rather than the entire 3D volume, the system reduces processing complexity while maintaining noise reduction effectiveness. This extraction approach avoids the computational burden of full volumetric filtering.
Data Source
AI summary
An image processing apparatus comprises processing circuitry configured to: obtain an irradiance volume representative of virtual light cast into a volumetric imaging data set, the irradiance volume comprising a respective irradiance value for each of a plurality of points in the irradiance volume; determine, for each of a plurality of reference points in the irradiance volume, a respective value for an accuracy measure, wherein the value for the accuracy measure at each reference point is representative of an accuracy with which irradiance has been determined at or near that reference point; and perform a rendering process using the irradiance volume, wherein the rendering process is performed in dependence on the determined values for the accuracy measure.