Light Field Imaging Device Compressive Sensing Data Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional light-field capture devices generate large amounts of data, leading to challenges in compression and increased computational load during processing, with existing methods either losing valuable information or requiring resource-intensive interpolation.

Innovation Solution

A light-field capture device equipped with a compressive sensing calculation unit that determines the homogeneity of input pixels and reduces output data to either a single pixel or multiple pixels based on a standard deviation threshold, while also correcting vignetting effects, thereby reducing the size of raw image data and processing load.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If conventional light-field capture devices capture light from different viewpoints using a micro-lens array, then additional optical information about directional distribution of light rays is captured, but the amount of data generated becomes considerable and compression becomes challenging

Engineering Contradiction:
Improveoptical informationVSAvoiddata size
Core Design Contradiction:
Loss of informationVSQuantity of substance

Solution Approach 1:

The patent divides the photo-sensor into multiple dedicated areas, each corresponding to a micro-lens image. A compressive sensing calculation unit is assigned to each area to independently process and reduce the data from that specific region, enabling selective compression while preserving important optical information.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different compression strategies to different regions of the photo-sensor based on their specific characteristics. By analyzing the content and importance of each dedicated area, the system reduces data size in regions where compression is safe while maintaining higher quality in regions where optical information is critical.

Inventive Principle:
Principle #3Local quality

2Quantity of substance

If existing compression methods are applied to reduce light-field data size, then data compression is achieved, but valuable information is lost or heavy resource-consuming interpolation is required

Engineering Contradiction:
Improvedata sizeVSAvoidvaluable information
Core Design Contradiction:
Quantity of substanceVSLoss of information

Solution Approach 1:

The patent performs preliminary analysis of each dedicated area to determine the appropriate compression level before actual data reduction. By pre-evaluating the characteristics of each region, the system can apply compression strategies that preserve valuable information while still reducing data size, avoiding the need for heavy interpolation later.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The compressive sensing calculation unit continuously monitors and adjusts the compression process based on feedback from the dedicated areas. This feedback mechanism ensures that compression is applied intelligently, maintaining information quality while reducing data size, and preventing the loss of valuable optical information.

Inventive Principle:
Principle #23Feedback

3Loss of information

If all captured pixels are stored and transferred to post-processing, then complete light-field data is preserved, but the computational load and processing time increase significantly

Engineering Contradiction:
Improvelight-field dataVSAvoidprocessing speed
Core Design Contradiction:
Loss of informationVSProductivity

Solution Approach 1:

The patent segments the photo-sensor into multiple dedicated areas, each handled by its own compressive sensing calculation unit. This segmentation enables parallel processing of different regions, reducing the overall computational load while preserving complete light-field data through systematic data reduction in each segment.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary data reduction at the dedicated area level before the data reaches post-processing stages. By pre-compressing and filtering data in each segment, the system significantly reduces the amount of data that requires intensive post-processing, thereby increasing overall processing speed while maintaining data completeness.

Inventive Principle:
Principle #10Preliminary action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The solution effectively reduces the size of raw image data while preserving valuable information, significantly lowering the computational load during processing and maintaining the quality of captured light-field data.

Implementation Method 1

a main lens configured to refract light from a scene towards an image focal field

Methodology Applied
Scientific EffectRefraction: Refraction

Implementation Method 2

a micro-lens array positioned in the image focal field between the main lens and the photo-sensor, each micro-lens being configured to project a micro-lens image onto a respective area of the photo-sensor

Methodology Applied
Scientific EffectRefraction: Refraction

Data Source

PatentEP3104596B1Light field imaging device
Publication Date: 2019.11.20 INTERDIGITAL CE PATENT HOLDINGS SAS
  • EP3104596B1 patent drawingFigure 1
  • EP3104596B1 patent drawingFigure 2
  • EP3104596B1 patent drawingFigure 3

AI summary

A light-field capture device 1 comprising: • a photo-sensor 5, • a main lens 2 configured to refract light from a scene towards an image focal field, • a micro-lens array 4 positioned in the image focal field between the main lens 2 and the photo-sensor 5, each micro-lens 7 being configured to project a micro-lens image 8 onto a respective area 9 of the photo-sensor 5; wherein the light-field capture device 1 comprises at least one calculation unit 6 connected to at least one of the respective areas 9 and configured to obtain homogeneity data of input pixels of said respective area 9, and to provide output data representative of the input pixels of said respective area 9; the output data comprising either a number of output pixels corresponding to the number of input pixels, or a single output pixel, depending on a comparison of the homogeneity data with a homogeneity threshold.