Object Tracking Anti-Jitter Filtering With Speed-Based Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing location tracking systems using moving average filters suffer from jitter and inaccuracies due to missed raw location points, especially in scenarios like football where players are stationary, crowded, or leaning, leading to increased noise and errors in location data.
Innovation Solution
An anti-jitter filter system comprising inter-related filtering algorithms that dynamically adjust to improve location data quality by modifying filtering based on speed, sporadic data, accuracy, and gaps, using a combination of averaging, projection, multi-tag correction, and physical limit filters to minimize artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fixed-time moving average filter is used to calculate location data, then the filter delay remains constant and is easily handled by the system, but missed raw location points introduce errors into the location data due to the time correction mechanism
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed-time moving average filter to a dynamically adaptive filter that adjusts its parameters based on real-time conditions. The filter monitors the quality and completeness of received location data, and dynamically modifies its averaging window and time correction mechanisms to account for missed data points, thereby maintaining accuracy without requiring complex error correction protocols
Solution Approach 2:
The patent implements parameter changes by modifying filter characteristics such as the averaging window size, sampling rate, and weighting factors based on the detected quality of location data. When data quality degrades or points are missed, the filter automatically adjusts these parameters to compensate, maintaining reliable location tracking without increasing system complexity
2Measurement precision
If aggressive filtering is applied to reduce noise in location data, then the quality of location data improves, but unwanted artifacts such as loss of responsiveness and sudden jumps are introduced
Solution Approach 1:
The patent uses dynamics by implementing a responsive filter that adapts its aggressiveness based on the motion state of the tracked object. When the object is stationary or moving slowly, more aggressive filtering is applied to reduce noise. When the object accelerates or changes direction rapidly, the filter becomes less aggressive to maintain responsiveness and avoid introducing artifacts like sudden jumps
Solution Approach 2:
The patent implements feedback mechanisms where the filter continuously monitors the characteristics of location data and the motion patterns of the tracked object. Based on this feedback, the filter dynamically adjusts its parameters to optimize the balance between noise reduction and responsiveness, preventing the introduction of unwanted artifacts while maintaining data quality
3Measurement precision
If the averaging window period is increased to reduce noise for stationary objects, then the filter delay increases which may cause visible artifacts in the location data
Solution Approach 1:
The patent applies dynamics by making the averaging window period adaptive rather than fixed. The filter dynamically adjusts the window size based on the detected state of the tracked object. For stationary objects, a larger averaging window is used to reduce noise. For moving objects, the window size is reduced to minimize delay and avoid visible artifacts in the location data
Data Source
AI summary
A speed-based projection filtering method includes generating, based on a sequence of raw location points that indicate locations of a tracking tag attached to a tracked object, a sequence of averaged location points. Based on at least two most-recent averaged location points of the sequence, a latest speed and direction of the tracked object are determined. Based on the latest speed and direction, one or more projected location points are determined when the latest speed of the tracked object is greater than a high-speed threshold and raw location points are missing after the sequence of raw location points. The projected location points are then outputted.