Eye Tracking Video Trajectory Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining motion trajectories in video data, such as object tracking algorithms, face challenges with occlusions and high computational requirements, leading to inaccurate path tracking and user experience issues, especially in complex scenes with multiple moving objects.
Innovation Solution
A method that captures a viewer's eye movements to generate trajectory data, using gaze tracking to determine object paths with high accuracy, reducing the need for computationally expensive image processing algorithms by leveraging human accuracy in following objects, even under occlusions, and combining this data with traditional image-based tracking for improved reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object tracking algorithms based on image processing are used, then trajectory data can be determined, but processing power requirements and processing time increase significantly
Solution Approach 1:
The patent introduces an intermediary element (such as a fiducial marker or known reference object) in the scene that emits or reflects light with known characteristics. This intermediary serves as a mediator between the camera system and the tracking algorithm, providing easily detectable reference points that simplify the trajectory determination process without requiring complex image processing of the entire scene.
Solution Approach 2:
The patent changes the parameters of the object being tracked by equipping it with active light sources or reflective elements that modify its optical properties. By making the tracked object emit or reflect light in specific patterns or wavelengths, the system can detect the object's position more easily and accurately, reducing the computational burden while maintaining or improving trajectory accuracy.
2Measurement precision
If object tracking algorithms are used, then trajectory data can be determined, but processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-equipping the tracked object with identifiable features (such as colored markers, reflective elements, or active light sources) before tracking begins. These pre-configured features enable the tracking algorithm to quickly identify and follow the object without requiring complex real-time analysis, significantly reducing processing time while maintaining accurate trajectory determination.
Solution Approach 2:
The patent uses simple, easily detectable visual markers or reference objects that can be processed quickly by the camera system. These markers serve as temporary, easily replaceable elements that provide sufficient tracking information without requiring sophisticated processing, allowing for fast trajectory determination with minimal computational overhead.
3Measurement precision
If traditional object tracking is used in complex scenes with occlusions, then trajectory data may be determined, but reliability decreases due to occlusions and path fragmentation
Solution Approach 1:
The patent transitions from two-dimensional image plane tracking to three-dimensional spatial tracking by incorporating depth information or using multiple camera angles. By adding this extra dimension, the system can maintain reliable trajectory determination even when objects are occluded in the 2D image plane, as the additional spatial information allows the algorithm to predict and continue tracking through occlusions.
Solution Approach 2:
The patent implements feedback mechanisms where the tracking system continuously monitors the object's expected position based on previous trajectory data and actively searches for the object in the predicted location during each frame. This predictive feedback approach allows the system to maintain tracking reliability during brief occlusions by anticipating where the object should appear and quickly reacquiring it when visible again.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
The present invention relates to a method for generating trajectory data for video data. According to the method, an eye movement of an eye (11) of a viewer (10) viewing moving visual images is captured with a capturing device (14). The trajectory data (32) is automatically determined based on the eye movement with a processing device (15) and the trajectory data (32) is automatically assigned to the video data comprising the moving visual images.