AR Virtual Object Stabilization via Image Feature Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current position and motion sensors in augmented reality systems suffer from measurement inaccuracies and 'jitter', causing virtual objects to visibly 'jump' or shake, leading to user discomfort, especially at larger distances.
Innovation Solution
Incorporating image processing methods to stabilize the position of virtual objects by recording camera images and comparing them with reference images, using displacement vectors to determine when to rely on sensor data for accurate positioning, thereby eliminating jitter.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If position and motion sensors are used to determine current position and direction of view, then the system can track user movement and display virtual objects accordingly, but measurement inaccuracies and drift cause the virtual objects to jitter and jump around
Solution Approach 1:
The patent combines sensor-based position determination with camera-based image feature tracking to create a hybrid positioning system. The camera records sequences of images and determines displacement vectors based on image feature changes, which are then merged with sensor data to calculate a corrected current position that compensates for sensor drift and inaccuracies.
Solution Approach 2:
The system continuously monitors the displacement vector from image feature tracking and uses this feedback to correct the position determined by sensors. When the displacement vector indicates movement, the system adjusts the current position calculation to account for sensor drift, creating a closed-loop feedback mechanism that stabilizes virtual object positioning.
2Adaptability or versatility
If the virtual object is displayed at large distances from the observer, then the application scope is expanded, but the jitter becomes more visible and causes user discomfort
Solution Approach 1:
The patent replaces reliance on purely mechanical/sensor-based position tracking with an optical measurement system using a camera. By capturing images and tracking image features, the system obtains more stable position data that is less susceptible to the drift and inaccuracies of conventional motion sensors, thereby reducing jitter visibility at large distances.
3Reliability
If camera images are processed to determine displacement vectors, then position stability is improved, but the device complexity increases due to additional image processing requirements
Solution Approach 1:
The camera serves multiple functions: it captures images for displacement vector calculation, provides visual feedback for user orientation, and can be used for environmental mapping. This multi-functionality justifies the added complexity by consolidating several system functions into a single component.
Solution Approach 2:
The system uses the camera to capture images of the environment, which automatically provides reference features for displacement calculation without requiring external calibration objects or additional sensors. The environment itself serves as the reference frame, eliminating the need for separate calibration systems.
Data Source
Figure 1~2
AI summary
The invention relates to a method for inserting a computer-generated virtual object (Oi) into a person's perceived field of vision of real surroundings, particularly into the field of view of transparent augmented reality glasses, wherein the insertion of the virtual object (Oi) is dependent on the current position (Pi) and line of vision (Bi) of said person. According to the invention, a sequence of camera images of the real surroundings perceived by the person is captured using a camera at predetermined temporal intervals and, with the aid of known methods for tracking image features between a captured camera image and a reference image (Ri) associated with a known position (Pi) of the person, a displacement vector is determined on the basis of said image features, wherein, when the displacement vector falls below a predetermined displacement value, the virtual object (Oi) is inserted on the basis of the person's known position (Pi), associated with the corresponding reference image (Ri), as the current position (Pi) of the person.