Light Field Depth Map Generation via Epipolar Line Slope Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for obtaining depth maps from plenoptic cameras are computationally intensive, making it difficult to achieve real-time processing on mobile devices, such as smartphones and tablets, which require efficient algorithms to handle 60 frames per second video images.
Innovation Solution
A computationally efficient method that processes light field images by detecting edges and calculating depth only for identified edges in epipolar images, using slope analysis to generate depth maps, reducing computational requirements by focusing on edge detection and slope calculation rather than processing all image points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If correspondence algorithms or depth-from-defocus techniques are used to extract depth information from light field, then depth map accuracy is improved, but computational complexity increases making real-time processing impossible
Solution Approach 1:
The patent extracts only the essential depth information from light field by analyzing epipolar images and identifying epipolar lines, rather than processing all image points. This selective extraction of critical features (epipolar line positions and slopes) reduces computational complexity while maintaining depth measurement accuracy.
Solution Approach 2:
The patent segments the light field processing into distinct steps: generating epipolar images, detecting edges, identifying epipolar lines, and calculating depths. By dividing the complex processing task into manageable segments focused on specific regions (epipolar lines), the computational burden is reduced while preserving measurement precision.
2Loss of information
If all image points are processed to generate depth maps, then comprehensive depth information is obtained, but processing time increases to hundreds of milliseconds or minutes per frame
Solution Approach 1:
The patent extracts depth information only from epipolar lines in epipolar images, which contain the essential depth cues. By focusing processing on these specific line structures rather than all image points, the method achieves real-time processing speeds (60 fps) while maintaining complete depth information coverage.
Solution Approach 2:
The patent performs partial processing by analyzing only the necessary epipolar line features rather than exhaustive processing of all pixels. This partial action approach processes only the critical elements needed for depth calculation, achieving real-time performance without sacrificing depth information quality.
3Measurement precision
If computationally intensive depth extraction algorithms are used, then depth map quality is improved, but energy consumption increases draining mobile device batteries quickly
Solution Approach 1:
The patent extracts depth information efficiently by focusing computational resources on epipolar line analysis in epipolar images. This targeted extraction approach maintains high depth map quality while significantly reducing energy consumption, enabling sustainable operation on mobile devices with battery power.
Solution Approach 2:
The patent uses partial processing of only essential epipolar line features rather than exhaustive analysis of all image data. This reduces the computational workload and energy consumption while preserving depth map quality, making the system viable for battery-powered mobile devices.
4Productivity
If simple light field gradient methods are used, then computational efficiency is improved, but depth map reliability deteriorates with poor depth estimations
Solution Approach 1:
The patent introduces epipolar images as an intermediary representation between the raw light field and the final depth map. By transforming the light field into epipolar images first, the method maintains computational efficiency while improving depth estimation reliability through structured line-based analysis of epipolar features.
Data Source
AI summary
A device and method for obtaining distance information from views is provided. The method generating epipolar images from a light field captured by a light field acquisition device; an edge detection step for detecting, in the epipolar images, edges of objects in the scene captured by the light field acquisition device; in each epipolar image, detecting valid epipolar lines formed by a set of edges; determining the slopes of the valid epipolar lines. The edge detection step may calculate a second spatial derivative for each pixel of the epipolar images and detect the zero-crossings of the second spatial derivatives, to detect object edges with subpixel precision. The method may be performed by low cost mobile devices to calculate real-time depth-maps from depth-camera recordings.