Video-See-Through Headset Correction for Pupil-Aligned Views
Solution Overview
Problem
Conventional video-see-through head mounted display systems fail to accurately align the captured video images with the user's actual viewpoint due to camera placement discrepancies, leading to distortions in size and distance perception.
Innovation Solution
A head mounted display system with at least two video cameras mounted adjacent to the display, relative to the user's pupils, and a computational device that processes the captured images to generate a view corresponding to the user's actual viewpoint, using techniques like light field rendering to correct for these discrepancies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video cameras are mounted at different physical locations than the user's pupils, then the head mounted display can capture video images of the physical world, but the captured video images do not accurately correspond to the user's pupil position and distance
Solution Approach 1:
The patent introduces an intermediary computational process that captures images from cameras positioned away from the pupils and computationally transforms them to match the pupil viewpoint. This mediator (image processing algorithm) bridges the gap between the camera positions and the desired pupil-centered viewpoint, allowing accurate viewpoint representation without requiring cameras to be physically mounted at the pupil locations.
Solution Approach 2:
The patent changes the parameters of the captured images through computational processing, specifically adjusting viewpoint parameters, perspective, and geometric transformation to match the user's pupil position and orientation. By modifying these image parameters rather than the physical camera positions, the system achieves accurate viewpoint alignment while maintaining flexible camera mounting configurations.
2Ease of operation
If conventional video cameras are used to capture video images, then the system can provide video-see-through functionality, but the viewpoint of the captured images does not directly correspond to the actual viewpoint of the user
Solution Approach 1:
The computational image processing acts as an intermediary that transforms the raw camera images into corrected images that correspond to the user's actual viewpoint. This intermediary process maintains the simplicity of using conventional video cameras while correcting the viewpoint mismatch through algorithms that account for the relative positions of cameras and pupils.
Solution Approach 2:
The patent replaces the mechanical solution of mounting cameras directly at the pupil locations with a computational solution. Instead of mechanically positioning cameras at precise pupil locations, the system uses computational image processing to achieve the same effect, substituting mechanical precision requirements with algorithmic correction.
Data Source
AI summary
A head mounted display system with video-see-through (VST) is taught. The system and method process video images captured by at least two forward facing video cameras mounted to the headset to produce generated images whose viewpoints correspond to the viewpoint of the user if the user was not wearing the display system. By generating VST images which have viewpoints corresponding to the user's viewpoint, errors in sizing, distances and positions of objects in the VST images are prevented.