3D Multi-Camera Target Determination for Performer Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video analysis systems struggle to accurately identify and track individuals performing specific actions, particularly in crowded environments like gymnastics competitions, due to issues such as erroneous detections and tracking failures.
Innovation Solution
A target determination method that utilizes multi-viewpoint image analysis to track the two-dimensional and three-dimensional positions of individuals, employing triangulation and specific region definitions to distinguish between performers and non-performers, using characteristics like height, distance from the bar, and speed to accurately identify the first performer in a specific action.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-viewpoint tracking is used to improve identification accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent transitions from two-dimensional image plane tracking to three-dimensional space tracking by introducing depth information through triangulation of multi-viewpoint camera data. This dimensional elevation enables precise identification of performers in crowded environments by utilizing spatial coordinates (x, y, z) rather than just two-dimensional positions, thereby resolving the contradiction between improved measurement precision and increased device complexity through effective use of additional spatial dimension.
Solution Approach 2:
The patent introduces a three-dimensional position specification unit as an intermediary component that processes two-dimensional position information from multiple cameras and converts it into accurate three-dimensional coordinates. This intermediary mechanism facilitates the transformation from complex multi-viewpoint data to simplified spatial representation, enabling precise performer identification while managing system complexity through modular processing architecture.
2Reliability
If three-dimensional position tracking is implemented to reduce erroneous detections, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent performs preliminary three-dimensional position specification for all detected persons before entering the main action recognition process. By pre-calculating and storing three-dimensional coordinates and movement trajectories, the system eliminates erroneous detections during action recognition, as the spatial context is already established. This preliminary action ensures reliable tracking while managing processing complexity through staged computation.
Solution Approach 2:
The patent changes the parameter representation from two-dimensional image coordinates to three-dimensional spatial coordinates. This parameter transformation fundamentally improves tracking reliability by adding depth information that disambiguates overlapping persons in crowded scenes. The system maintains manageable complexity by consistently using these transformed parameters throughout the processing pipeline, from detection to action recognition.
3Measurement precision
If start region definition is used to identify the first performer, then measurement precision is improved, but difficulty of detecting and measuring increases
Solution Approach 1:
The patent applies local quality by defining a specific three-dimensional start region with particular spatial characteristics around the gymnastics apparatus. This localized region has distinct properties (spatial boundaries, height thresholds) that differentiate it from other areas in the scene. By focusing detection efforts on this specific local region rather than the entire field of view, the system achieves precise performer identification while simplifying the overall detection process through spatial segmentation.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method enables precise identification and tracking of individuals performing specific actions, reducing erroneous detections and improving the accuracy of action recognition in crowded environments.
Implementation Method 1
specifying a three-dimensional position of the person based on the acquired two-dimensional position of the person and camera parameters of each of the plurality of cameras
Data Source
AI summary
A target determination device includes a processor that executes a procedure. The procedure includes: acquiring information obtained by tracking, between frames, a two-dimensional position of a person estimated to be the same person, the two-dimensional position being a two-dimensional position of the person in each frame of a video captured by each of a plurality of cameras that capture a predetermined capturing range from a plurality of different viewpoints; specifying a three-dimensional position of the person based on the acquired two-dimensional position of the person and camera parameters of each of the plurality of cameras; and determining, as a person that is a recognition target, a person who first enters a start region, determined in advance as a three-dimensional region in which a person who performs a specific action is present at a start of the specific action, based on the specified three-dimensional position of the person.