Light Field Image Processing Apparatus Data Reduction via Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Light Field (LF) data contains large amounts of information about angle and space distribution of light intensity, resulting in significant data volume.
Innovation Solution
An image processing apparatus that sets areas within the image and applies different levels of addition processing to subpixel signals based on in-focus information to generate first and second image data, reducing the data amount while maintaining necessary information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If LF data is constructed by a plurality of parallax images containing angle distribution information and space distribution information, then refocusing capability is improved, but data amount increases significantly
Solution Approach 1:
The image area is divided into multiple regions, and different processing methods are applied to each region. Specifically, the image is segmented into a first region where subpixel signals are added and a second region where they are not added, allowing selective data reduction while preserving refocusing capability where needed
Solution Approach 2:
Different quality levels of data are assigned to different regions of the image. The first region maintains full subpixel information for refocusing operations, while the second region uses reduced data, creating local variations in data quality based on regional requirements
2Quantity of substance
If all subpixel signals are added together to generate a single image, then data amount is reduced, but refocusing information is lost
Solution Approach 1:
The processing is segmented by region rather than applying a uniform approach to the entire image. This allows preservation of refocusing information in the first region while reducing data in the second region, avoiding complete information loss
Solution Approach 2:
Instead of adding all subpixel signals everywhere (excessive action), the patent applies addition processing only partially in specific regions where it is most beneficial, maintaining necessary information in other regions
Data Source
AI summary
An image processing apparatus for processing image pickup signals of subpixels of an image pickup element including a plurality of unit pixels each of which has a plurality of subpixels for receiving light passing through different partial pupil areas of a focusing optical system and generating image data of a photographed image, obtains in-focus information of the photographed image, sets a first area and a second area different from the first area onto the photographed image on the basis of the obtained in-focus information, and adds a part of the image pickup signals of the subpixels obtained from the image pickup element so that the number of image pickup signals of the subpixels per unit pixel in the image data of the photographed image corresponding to the first area is larger than that of the image data of the photographed image corresponding to the second area.