Vehicular Vision Object Tracking With Adaptive Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional vehicle imaging systems suffer from temporal inconsistencies and missed detections due to image noise and frame-to-frame variations in object location, leading to inconsistent object tracking and detection.
Innovation Solution
A vehicular vision system employs an object tracking-and-detection system with modules for object identification, merging, and smoothing, utilizing a Kalman filter with adaptive thresholding to stabilize object tracking by correlating object positions across frames and resetting when drift exceeds a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional image processing is used without adaptive thresholding, then the system is simpler to implement, but temporal inconsistencies and missed detections occur due to image noise and frame-to-frame variations
Solution Approach 1:
The patent implements adaptive thresholding where the threshold value dynamically adjusts based on the standard deviation of object positions across multiple frames. This dynamic adaptation allows the system to automatically respond to varying noise levels and frame-to-frame variations, improving detection consistency without requiring manual intervention or complex fixed-threshold configurations
Solution Approach 2:
The system calculates the standard deviation of detected object positions across frames and uses this statistical feedback to adjust the threshold for position matching. This feedback mechanism enables the system to learn from temporal variations and maintain reliable detection despite noise, resolving the contradiction between simplicity and reliability
2Stability of the object's composition
If a fixed threshold is used for object position matching, then the processing is faster and simpler, but location variations and tracking instability occur under varying noise conditions
Solution Approach 1:
The patent changes the threshold parameter from a fixed value to an adaptive value calculated as a function of the standard deviation of object positions. This parameter transformation allows the system to maintain tracking stability by adjusting the threshold according to actual noise levels, while avoiding the computational overhead of more complex adaptive algorithms
3Measurement precision
If adaptive thresholding with standard deviation calculation is implemented, then detection accuracy and tracking consistency improve, but computational complexity increases
Solution Approach 1:
The patent applies adaptive thresholding only to the critical dimension of object position matching, rather than implementing full adaptive processing across all image parameters. This partial application achieves improved measurement precision for object tracking while limiting computational complexity to only the necessary calculations for position standard deviation and threshold adjustment
Data Source
AI summary
A vehicular vision system includes a camera disposed at a vehicle. Responsive to processing at an image processor of a first frame of image data captured by the camera, the system detects a first object exterior of the vehicle, and responsive to processing at the image processor of a second frame of image data captured by the camera, the system detects a second object exterior or the vehicle. The system determines, based on a similarity measure, that the second object is the same as the first object and, responsive to determining that the second object is the same as the first object, using a smoothing filter, modifies the determined second position of the second detected object. The system, based on the modified determined second position of the second detected object relative to the equipped vehicle, updates parameters of the smoothing filter. The system tracks the second detected object.