Multi-Camera Tracking Prediction Using Motion Trajectory Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video surveillance systems fail to accurately predict which camera a moving person will appear in, especially when camera imaging positions are close, leading to potential missed tracking due to the person disappearing before being highlighted.
Innovation Solution
An information processing system and method that analyzes the traveling direction and speed of a person to predict their subsequent appearance in another camera, using a formula combining disappearance and appearance position probabilities to dynamically switch camera displays and highlight likely appearance positions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system waits for the person to disappear from the current camera view before predicting the next camera, then the prediction can be made after confirmation of disappearance, but the person may frame out and be missed before the prediction is displayed
Solution Approach 1:
The system performs prediction of the next camera in advance, before the person actually disappears from the current camera view. The prediction is calculated based on the person's current position and trajectory, and the result is prepared and displayed before the disappearance event occurs, ensuring no tracking is missed.
2Ease of operation
If the system highlights the next camera prediction, then the user can quickly identify where the person will appear, but when camera imaging positions are extremely close, the person may frame out before being highlighted
Solution Approach 1:
The prediction calculation is performed in advance based on the person's current position and movement trajectory, determining which camera will capture the person next before the person actually moves out of the current camera's view. This advance prediction allows the system to prepare and display the correct next camera view, preventing the person from framing out undetected.
3Adaptability or versatility
If the system uses topology information and histograms to estimate camera connections, then the system can track persons across cameras, but the prediction accuracy decreases when camera imaging positions are extremely close
Solution Approach 1:
The system changes from using static topology information and histograms to using dynamic real-time parameters including the person's current position coordinates, movement trajectory, and calculated velocity. By incorporating these time-varying parameters, the system achieves higher prediction accuracy even when camera positions are extremely close together, as the prediction is based on the person's actual motion state rather than predetermined camera relationships.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided are an information processing system, an information processing method and a program capable of suitably predicting, when tracking a person with a plurality of video cameras, an image of a video camera, in which a moving body that is appearing in a video camera will subsequently appear. The information processing system of the present invention includes an interior view angle person position acquisition unit 110 for identifying a traveling direction of a moving body appearing in an image of a first video camera among a plurality of video cameras, and an appearance probability calculation/sorting unit 150 for predicting, based on the traveling direction of the moving body in the image of the first video camera, one or more second video cameras in which a possibility of the moving body that is appearing in the image of the first video camera subsequently appearing is higher than other video cameras among the plurality of video cameras.