Asymmetric Pixel Distribution for Light Field Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Light field cameras capture view information uniformly in all directions, which does not meet the differential requirements for multiview image generation and 3D reconstruction, as human eyes are more sensitive to horizontal views, leading to inefficient use of pixel resources.
Innovation Solution
Adjusting the pixel distribution in the image sensing unit to change the ratio of pixels along orthogonal directions, allowing for differential capture of view information by increasing or decreasing pixel quantities in specific directions, thereby optimizing the capture of parallax information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If uniform pixel distribution is used in all directions, then the image sensing unit can capture view information uniformly, but the pixel resources are not efficiently utilized for human vision-sensitive directions
Solution Approach 1:
The patent applies asymmetry by changing the pixel distribution from uniform to non-uniform, with different pixel densities in different directions. Specifically, the effective region is divided into first and second regions with different pixel distribution densities, allowing more pixels to be allocated to directions where human eyes are more sensitive (horizontal views) and fewer pixels to directions where sensitivity is lower, thereby resolving the contradiction between uniformity and efficiency.
Solution Approach 2:
The patent implements local quality by assigning different pixel distribution characteristics to different regions within the effective region. The first region and second region have different pixel distribution densities tailored to their specific directional requirements, optimizing the capture quality for each direction according to human visual sensitivity rather than applying a single uniform distribution across the entire sensor.
2Device complexity
If pixel quantities along first direction and second direction are equal, then the pixel distribution is symmetric and simple, but the capture of differential view information required for 3D reconstruction is insufficient
Solution Approach 1:
The patent deliberately introduces asymmetry in pixel distribution to capture sufficient parallax information for 3D reconstruction. By making the pixel quantities along the first direction and second direction different, the system captures differential view information that is essential for accurate depth perception and 3D reconstruction, accepting increased complexity as a necessary trade-off for information completeness.
3Ease of manufacture
If pixels are distributed evenly in round effective region, then manufacturing is simple, but the ratio of pixels along orthogonal directions cannot be adjusted to meet differential capture requirements
Solution Approach 1:
The patent segments the effective region into multiple sub-regions (first region and second region) with different pixel distribution characteristics. This segmentation allows independent optimization of pixel density in each region according to directional requirements, providing the flexibility to adjust pixel ratios along orthogonal directions while maintaining a structured manufacturing approach.
Solution Approach 2:
The patent applies different pixel distribution strategies to different local regions within the effective region. By making the pixel distribution non-uniform and direction-dependent, the system achieves adaptability in pixel ratio adjustment to meet differential capture requirements, while the underlying segmented structure maintains manufacturing feasibility.
Data Source
AI summary
Image capture control methods and apparatuses are provided. A method comprises: adjusting pixel distribution of an image sensing unit, to change a ratio of pixels distributed along two directions in an effective region of the image sensing unit, wherein light through a lens that is in an imaging system and corresponds to an image sensing unit is imaged on the image sensing unit in the effective region, the two directions are parallel with the image sensing unit and orthogonal to each other, the image sensing unit comprises pixels adjustable in distribution, and the pixels are used for recording multidirectional view information of a same object of a scene; and capturing an image of the scene by using the imaging system. Proportions of parallax information recorded by the image sensing unit in the two directions can be changed, and differential capture of view information in different directions can be implemented.