Mobile Body Tracking Using Predicted Destination Areas
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile body tracking systems struggle to accurately track vehicles between time-series images, especially when the frame rate is low and the amount of vehicle movement is significant, leading to reduced tracking accuracy.
Innovation Solution
A mobile body tracking apparatus that includes detection, prediction, and tracking means to predict a destination area using past information on mobile body positions and track the same mobile body across time-series images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the frame rate of the camera is low, then the amount of movement of the vehicle between time-series images is large, but it becomes difficult to track the vehicle accurately
Solution Approach 1:
The system performs preliminary extraction of destination candidates in the t+1-th image before tracking. By pre-identifying potential destination areas based on predicted moving direction and speed, the system prepares candidate regions in advance, reducing the search space and improving tracking accuracy even when vehicles move significantly between frames.
Solution Approach 2:
The system introduces pair features (pixel pairs from the same position) as an intermediary to bridge the gap between the t-th image and t+1-th image. By calculating similarity based on these pixel pairs rather than direct image comparison, the system can accurately track vehicles that have moved significantly between frames.
2Measurement precision
If a large number of destination candidates are extracted from the t+1-th image, then the system can search for the vehicle destination, but the processing complexity increases
Solution Approach 1:
The system pre-predicts the moving direction and speed of the vehicle from previous tracking results to determine the search range for destination candidates. This preliminary action restricts the candidate extraction to relevant regions only, maintaining detection accuracy while reducing the overall number of candidates and processing complexity.
Solution Approach 2:
The system applies different processing strategies to different regions of the image. By focusing computational resources on predicted destination areas rather than the entire image, the system achieves high tracking accuracy while minimizing processing complexity in regions where tracking is less critical.
Data Source
AI summary
Detection means detects a mobile body from each of time-series images obtained by capturing images of a road. Prediction means predicts a destination area of the mobile body by using past information indicating positions where mobile bodies have been detected on the road in the past. In a case where the mobile body is detected from a second image in the destination area that is predicted for the mobile body detected from the first image, tracking means tracks the mobile body detected from the first image and the mobile body detected from the second image as the same mobile body.