Light-Field Panoramic Imaging via Micro-Lens Array Refocusing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional panoramic image generation from light-field data faces challenges in providing depth perception while optimizing memory utilization and minimizing parallax error, especially when stitching multiple fields of view.
Innovation Solution
A device and method that utilize a micro-lens array with light sensors to capture and process light-field data, allowing for dynamic refocusing and realignment of images based on user input, which enables the generation of panoramic images with depth perception by combining focused frames while minimizing parallax error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple fields of view are stitched together to generate panoramic images, then the coverage area is improved, but memory requirements increase substantially
Solution Approach 1:
The light-field data from multiple fields of view is segmented and processed individually. Each FOV's LF data is refocused and realigned separately before stitching, allowing memory-efficient processing of divided data sets rather than loading all data simultaneously
Solution Approach 2:
The patent transforms the high-dimensional light-field data into focused 2D or 3D images through computational refocusing. This dimensionality reduction converts the complex multi-dimensional LF data into manageable image formats that require less memory for storage and processing
2Area of stationary object
If light-field data from multiple fields of view is stitched together, then panoramic coverage is improved, but parallax error increases
Solution Approach 1:
The patent applies dynamic refocusing and realignment operations to each field of view's light-field data. By dynamically adjusting the focal plane and realigning images based on depth information, the system compensates for parallax effects that would otherwise cause misalignment in the stitched panoramic image
Solution Approach 2:
The system changes key parameters including focal length, focal plane position, and alignment transformations for each FOV based on depth analysis. These parameter adjustments are made to minimize parallax error while maintaining the panoramic coverage, allowing flexible optimization of the stitching process
3Manufacturing precision
If computational refocusing is applied to light-field data, then depth perception is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary analysis of the light-field data to identify subjects, objects, and depth information before executing the refocusing operations. This preliminary segmentation and depth mapping simplifies subsequent processing by pre-organizing the data structure and identifying regions that require specific refocusing treatments
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 generates panoramic images with depth perception, optimizing memory usage and reducing parallax errors, allowing for efficient storage and display of high-quality panoramic images with enhanced depth perception.
Implementation Method 1
an imaging device that comprises a micro-lens array which may further comprise a plurality of micro-lenses, in conjunction with a light sensor, may be configured to capture light-field (LF) data
Data Source
AI summary
Various aspects of a device, such as an imaging device, and method to generate a panoramic image are disclosed herein. The imaging device comprises a micro-lens array configured to capture light-field (LF) data associated with a field-of-view (FOV) of the imaging device. One or more circuits in the imaging device focus on a first subject to determine a first frame. The first frame is based on a first realignment of the stored LF data of the first subject in the FOV. Determination of one or more other frames occurs when the one or more circuits dynamically refocus on one or more other subjects. A panoramic image is generated from a combination of the determined first frame and the determined one or more other frames. The generated panoramic image corresponds to a pseudo three-dimensional image that contributes to the depth perception of the selected one or more subjects.