Multi-Camera Tracking Refinement for Cross-Device Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional tracking processing devices face challenges in maintaining accuracy due to occlusions and unsuitable tracking information, which complicates the association of moving objects across multiple cameras and reduces tracking precision.
Innovation Solution
A tracking processing device that refines intra-camera tracking information by removing unsuitable data based on similarity degrees and invalid regions, extracting inter-camera tracking information to improve accuracy and association across cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If tracking information from all cameras is used without refinement, then the quantity of tracking data is maximized, but the accuracy of tracking across cameras deteriorates due to inclusion of unsuitable information
Solution Approach 1:
The patent applies preliminary action by performing refinement processing on tracking information before it is used for cross-camera association. The refining unit pre-processes tracking information to extract only suitable data (removing occluded or low-quality tracks) before the association unit uses it for matching objects across cameras. This ensures that only high-quality tracking information is used in the association process, improving accuracy without requiring complex real-time filtering during association.
2Measurement precision
If tracking information is refined to remove unsuitable data, then the accuracy of tracking across cameras is improved, but the quantity of usable tracking information decreases
Solution Approach 1:
The patent applies local quality by selectively processing different portions of tracking information with different quality standards. The refining unit evaluates each tracking information individually based on local characteristics (occlusion status, detection quality) and retains only those that meet quality criteria. This ensures that high-quality tracking information is preserved for cross-camera association while filtering out only the unsuitable portions, maintaining optimal quantity of usable data.
3Area of stationary object
If occlusion handling is implemented to reduce blind spots, then the coverage of tracking is improved, but the complexity of determining effective imaging range increases
Solution Approach 1:
The patent applies preliminary action by pre-determining the effective imaging range for each camera before tracking begins. The system calculates and stores which regions in each camera's field of view are affected by occlusions or blind spots in advance. During actual tracking, this pre-computed information is used to quickly identify and exclude tracking information from ineffective regions, avoiding complex real-time calculations and reducing processing complexity while maintaining accurate coverage awareness.
Data Source
AI summary
To improve, when performing tracking of moving objects by using captured images taken by multiple cameras, the accuracy of the tracking process across the cameras, a tracking processing device includes: a storage unit that stores, for each camera, a plurality of pieces of intra-camera tracking information including image information of persons obtained from the captured images; a refining unit that performs refining of the plurality of pieces of intra-camera tracking information and thereby extracts pieces of inter-camera tracking information to be used in a process of tracking the persons across the multiple cameras; and an associating unit that, on the basis of the pieces of inter-camera tracking information, associates the persons in the captured images across the multiple cameras.