Compressive Sensing Light Field Estimation for Virtual Object Rendering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for rendering virtual objects in 3D spaces often result in unrealistic images due to assumptions about infinite light sources, and capturing accurate light fields in real environments is costly and inefficient, especially in AR/VR applications where processing power is limited.

Innovation Solution

A method using compressive sensing to estimate light field values from captured pixel values, allowing for realistic rendering of virtual objects by projecting pixel values onto a hemisphere and reconstructing light field values at discrete positions and orientations within the virtual space.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If compressive sensing is used to estimate light field values from captured pixel values, then rendering efficiency is improved and processing cost is reduced, but measurement precision of light field values may be compromised

Engineering Contradiction:
Improverendering efficiencyVSAvoidlight field value accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies compressive sensing to estimate light field values at selected discrete positions rather than all possible positions. By capturing pixel values at a subset of locations and using compressive sensing algorithms to reconstruct the full light field, the system achieves efficient rendering while maintaining acceptable accuracy. This partial measurement approach reduces processing cost while still providing sufficient information for realistic virtual object rendering.

Inventive Principle:
Principle #16Partial or excessive action

2Reliability

If light field values are captured at multiple positions in real space, then rendering realism is improved, but device complexity and processing requirements increase

Engineering Contradiction:
Improverendering realismVSAvoidprocessing requirements
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the continuous light field into discrete positions and orientations that can be represented by captured pixel values. By dividing the light field into manageable discrete components and using compressive sensing to reconstruct values at uncaptured positions, the system maintains rendering realism while reducing processing complexity. The segmentation allows the system to work with discrete data points rather than continuous measurements.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent uses compressive sensing to create copies of light field information from captured pixel values to estimated values at other positions. The reconstruction algorithm generates approximate light field values at positions where direct measurement was not performed, effectively copying the essential lighting characteristics from measured to unmeasured locations. This copying approach maintains realism while avoiding the need for complex measurements at every position.

Inventive Principle:
Principle #26Copying

3Ease of operation

If ambient colour capture is used to render virtual objects, then ease of operation is improved, but adaptability to different lighting conditions deteriorates

Engineering Contradiction:
Improverendering simplicityVSAvoidlighting condition adaptation
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic lighting adaptation by capturing pixel values that represent the actual light field at the camera's position and orientation. Instead of using static ambient colour assumptions, the system dynamically estimates light field values based on the current viewing position and orientation. This dynamic approach allows virtual objects to be rendered with lighting conditions that adapt to the user's perspective and the actual environment, improving versatility while maintaining ease of operation through automated estimation.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3599588B1Rendering an object
Publication Date: 2021.09.08 SIEMENS HEALTHCARE GMBH
  • EP3599588B1 patent drawingFigure 1
  • EP3599588B1 patent drawingFigure 2~3
  • EP3599588B1 patent drawingFigure 4~5

AI summary

Apparatus and a method of rendering an object in a virtual three-dimensional space is disclosed. The method comprises obtaining a first plurality of pixel values of a first image captured by a camera in real space, the first plurality of pixel values being associated with a first three-dimensional position, in the virtual three-dimensional space, of the camera when the first image was captured, each one of the first plurality of pixel values being associated with an orientation in the virtual three-dimensional space, of a portion of a light field represented by the pixel. The method comprises estimating, using compressive sensing, and based on the first plurality of pixel values, a light field value for at least one second position, different from the first position, in the virtual three-dimensional space; and rendering an object, in the virtual three-dimensional space, located at the second position, using the estimated light field value.