AR Image Tracking Points Using Corner Feature Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Augmented Reality (AR) technologies face challenges in maintaining accurate image tracking of objects when users move quickly or shake, due to significant differences between continuous video frames, leading to failed tracking.
Innovation Solution
A method and device for obtaining image tracking points by determining feature points in one video frame and comparing them to corresponding points in subsequent frames, using image similarity metrics to ensure accurate tracking, even in scenarios with rapid movement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image recognition technology is used for AR tracking, then virtual content can be overlaid at correct positions, but tracking fails when users move fast or shake due to large differences between continuous video frames
Solution Approach 1:
The patent changes the parameter of feature point selection from arbitrary feature points to corner points with specific geometric properties. Corner points have distinct local structures that remain recognizable under transformation, allowing reliable matching even when video frames differ significantly due to fast movement or shaking. This parameter change resolves the contradiction by making feature point matching robust to large frame differences.
Solution Approach 2:
The patent transforms the tracking problem from 2D image space to a multi-dimensional feature space by extracting corner point coordinates, curvature radii, and curvature directions. This dimensional transformation creates a more robust representation that can withstand variations in video frames, thereby improving tracking reliability without sacrificing matching precision.
2Ease of operation
If feature points are selected from arbitrary locations on the target object, then tracking can be implemented, but tracking accuracy decreases when there are significant differences between video frames
Solution Approach 1:
The patent applies local quality by selecting feature points based on their local geometric properties rather than uniform distribution. Corner points are identified by their distinctive local curvature characteristics, making them more suitable for tracking. This local quality approach ensures that selected feature points maintain high recognition accuracy even when video frames undergo significant transformations.
3Productivity
If traditional image recognition methods are used, then the system can identify target objects, but the tracking stability deteriorates under rapid movement conditions
Solution Approach 1:
The patent performs preliminary action by pre-selecting and marking corner points on the target object before tracking begins. These corner points are identified based on their geometric properties and stored as reference features. This preliminary preparation ensures that when rapid movement occurs, the system can quickly match these pre-identified stable features without losing tracking stability, thus resolving the contradiction between identification efficiency and tracking stability.
Data Source
AI summary
The present disclosure provides a method for obtaining image tracking points. The method can be applied to an image tracking point obtaining device, the method includes obtaining, when a current video frame comprises a first image of a target object, a second image of the target object and determining a position of a second feature point on the second image; obtaining, on the first image, a first feature point corresponding to the second feature point; obtaining a first area to which the first feature point belongs in the first image, and obtaining a second area to which the second feature point belongs in the second image; and determining the first feature point as a tracking point of the current video frame when an image similarity between the first area and the second area meets a screening condition.