KLT Object Tracking Robustness in ADAS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle tracking systems in ADAS applications face challenges in accuracy and robustness, particularly in tracking vehicles and bikes, due to issues like mis-assignments, missing detections, volatile box trajectories, and model drift, which affect Time To Contact (TTC) estimation.
Innovation Solution
The method employs the Kanade Lucas Tomasi (KLT) tracking technique to track objects of interest by determining an initial patch and its boundary box, forming a patch template of fixed size, and using it as a search template to track the object across successive images, with robustness checks to detect drift and track loss, allowing for partial tracking outside the image frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template tracking with KLT methodology is used, then measurement precision of object position is improved, but reliability of tracking is worsened due to model drift and track loss
Solution Approach 1:
The patent implements dynamic template adaptation by periodically updating the search template with newly detected object patches. This allows the template to adapt to appearance changes (resolving model drift) while maintaining subpixel precision tracking. The system dynamically adjusts between using historical templates and fresh detections based on confidence metrics.
Solution Approach 2:
The patent incorporates robustness checks that continuously evaluate tracking quality metrics (e.g., correlation strength, position consistency). When degradation is detected, the system triggers template updates or switches to detection-based re-initialization, creating a feedback loop that maintains reliability while preserving measurement precision.
2Reliability
If detection based approaches are used, then reliability of tracking is improved through robust detection, but measurement precision of bounding box is worsened due to volatile box trajectories
Solution Approach 1:
The patent merges template tracking (for precision) with detection-based methods (for reliability) into a hybrid system. The detection component provides robust object identification and initialization, while the template tracking component maintains precise continuous positioning. The system seamlessly switches between modes based on confidence metrics, achieving both precision and reliability.
3Adaptability or versatility
If highly adaptive template models are used, then adaptability to appearance changes is improved, but reliability is worsened due to drift away from original object
Solution Approach 1:
The patent applies partial adaptation by selectively updating only certain regions or aspects of the template based on confidence metrics. Instead of fully adapting the template every frame, the system performs incremental updates only when detection confidence is high and appearance changes are verified, preventing over-adaptation and drift while still handling legitimate appearance variations.
Data Source
AI summary
A method of tracking an object of interest between temporally successive images taken from a vehicle based camera system comprising:a) from an initial image, determining an initial patch with a respective boundary box encapsulating an identified object of interest;b) using a search template in Kanade Lucas Tomasi (KLT) methodology to track said object of interest in a temporally successive image from said camera system; so as to determine therein a new patch having a respective new boundary box or portion thereof, with respect to said object of interest, characterized in; andc) performing a check on the robustness of the tracking step in step b) by analyzing one or more parameters output from step b).