Multi-Camera Object Tracking via Spatial and Temporal Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle tracking methods using multiple cameras are inaccurate due to the influence of image shooting angle and object posture, and rely solely on image features, which are not robust enough for reliable tracking.
Innovation Solution
An object tracking method that involves obtaining multiple frames of images from multiple cameras, determining the distance between cameras, and using a combination of global and attribute feature similarities, along with moving speed and probability calculations, to accurately identify and track vehicles across different camera views.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If two frames of images are used to judge whether two vehicles are the same vehicle, then the tracking process can be simplified, but the tracking accuracy deteriorates due to the influence of image shooting angle and object posture
Solution Approach 1:
The patent combines multiple features including global features, attribute features, spatial information, and temporal information to form a comprehensive judgment criterion. This merging of multiple feature types resolves the contradiction by maintaining tracking accuracy while managing system complexity through integrated feature fusion rather than simple two-frame comparison
Solution Approach 2:
The patent introduces spatial dimension (distance between cameras, positions of objects in images) and temporal dimension (shooting moments, moving speed) to the traditional two-dimensional image feature comparison. This multi-dimensional approach improves tracking accuracy by providing additional判别 criteria beyond simple image similarity
2Device complexity
If only image features are used for tracking, then the system complexity is reduced, but the reliability of tracking results deteriorates
Solution Approach 1:
The patent changes the parameters used for tracking from purely image features to include spatial parameters (distance between cameras, object positions) and temporal parameters (shooting moments, moving speed). This parameter expansion improves reliability by providing multiple independent criteria for judgment, reducing dependence on any single feature type
3Measurement precision
If multiple features and spatial information are integrated for tracking, then tracking accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent segments the feature extraction and comparison process into distinct modules: global feature extraction, attribute feature extraction, spatial information acquisition, and temporal information acquisition. This segmentation manages complexity by organizing the multi-feature integration into manageable, independent components that can be processed separately and then combined
Data Source
AI summary
Embodiments of the present disclosure provide an object tracking method and an apparatus. The method includes: obtaining multiple frames of first images shot by a first camera apparatus and a first shooting moment of each frame of the first images, where the first images include a first object; obtaining multiple frames of second images shot by a second camera apparatus and a second shooting moment of each frame of the second images, where the second images include a second object; obtaining a distance between the first camera apparatus and the second camera apparatus; and judging whether the first object and the second object are the same object according to the multiple frames of the first images, the first shooting moment of each frame of the first images, the multiple frames of the second images, the second shooting moment of each frame of the second images and the distance.