Stereo Vision–PDAF Depth Detection for Occlusion-Prone Close Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer stereo vision struggles with determining depth data for objects in close proximity and certain shapes due to occlusion and varying vantage points, leading to increased computing resource usage and latency.
Innovation Solution
A depth determination system combines stereo vision and Phase Detection Auto Focus (PDAF) to switch between methods based on a threshold value and object proximity, using PDAF for close objects and stereo vision for others, ensuring accurate depth estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If stereo vision is used to determine depth for all objects, then depth determination can be performed uniformly, but computing resource usage increases and latency increases for close objects
Solution Approach 1:
The patent segments the depth determination task by dividing objects into two categories: close objects (within a threshold distance) and distant objects. Different methods are applied to each segment - PDAF for close objects and stereo vision for distant objects - which resolves the contradiction by avoiding computationally intensive stereo vision for close objects while maintaining efficient processing for all objects.
Solution Approach 2:
The patent applies different depth determination methods to different spatial regions. PDAF is applied locally to close objects where it provides superior accuracy and lower latency, while stereo vision is applied to distant objects where it is computationally more efficient. This local differentiation optimizes both accuracy and computing resource usage.
2Measurement precision
If stereo vision is used for close objects, then depth can be determined, but measurement precision deteriorates due to occlusion and varying vantage points
Solution Approach 1:
The patent segments objects based on their distance from the optical sensor, applying PDAF to close objects where stereo vision fails due to occlusion and vantage point variations. This segmentation ensures that the most appropriate method is used for each spatial region, maximizing measurement precision.
Solution Approach 2:
The patent introduces PDAF as an intermediary method that mediates between the optical sensor and close objects, providing accurate depth measurement without being affected by occlusion or vantage point variations that plague stereo vision. PDAF acts as a specialized tool for close-range depth measurement.
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
This approach enhances depth determination accuracy and reduces computing resources by leveraging PDAF for close objects and stereo vision for distant ones, improving performance in challenging scenarios.
Implementation Method 1
Phase Detection Auto Focus (PDAF)
Implementation Method 2
Computer stereo vision allows for the extraction of three-dimensional information from digital images. Two cameras displaced horizontally from one another are used to capture images depicting two differing views of a scene from two different vantage points. The relative depth of objects captured in the images is determined by comparing the relative positions of the objects in the two images.
Data Source
AI summary
Disclosed are systems, methods, and non-transitory computer-readable media for varied depth determination using stereo vision and phase detection auto focus (PDAF). Computer stereo vision (stereo vision) is used to extract three-dimensional information from digital images. To utilize stereo vison, two optical sensors are displaced horizontally from one another and used to capture images depicting two differing views of a real-world environment from two different vantage points. The relative depth of the objects captured in the images is determined using triangulation by comparing the relative positions of the objects in the two images. For example, the relative positions of matching objects (e.g., features) identified in the captured images are used along with the known orientation of the optical sensors (e.g., distance between the optical sensors, vantage points the optical sensors) to estimate the depth of the objects.