Subject Tracking Device Template Update Mechanism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subject tracking devices face issues with accurately tracking a photographic subject as the template images become less similar over time, leading to potential misidentification and tracking of incorrect subjects.
Innovation Solution
A subject tracking device that uses both an initial template image and an update template image, calculating similarity levels and adjusting weighting coefficients to generate a new update template image based on pixel information from both, ensuring accurate tracking even as the subject's shape changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the update template image is repeatedly updated to track subject shape changes, then the tracking adaptability improves, but the similarity to the initial template image decreases, causing erroneous tracking of wrong subjects
Solution Approach 1:
The patent segments the template image into two distinct components: an initial template image that remains fixed and provides reference stability, and an update template image that is dynamically updated to track shape changes. This segmentation allows the system to separately manage the conflicting requirements of adaptability (update template) and reliability (initial template), resolving the technical contradiction by having each template serve its specific function without interfering with the other.
Solution Approach 2:
The patent changes the parameter of template update frequency by introducing a decision-making mechanism that determines whether to update the update template image based on similarity comparisons. Instead of continuously updating (which would lose reliability), the system selectively updates only when necessary, controlled by the parameter of update frequency adjustment. This resolves the contradiction by dynamically adjusting the update parameter to balance adaptability and reliability.
2Measurement precision
If the template image is continuously updated to reflect current subject appearance, then the tracking accuracy for changing subjects improves, but the risk of tracking wrong subjects increases
Solution Approach 1:
The patent implements a feedback mechanism where the arithmetic operation unit calculates similarity levels between the initial template image and the update template image, and the decision-making unit uses this similarity feedback to determine whether to perform updates. This feedback loop ensures that updates are only executed when the subject has genuinely changed, preventing erroneous tracking of wrong subjects while maintaining accurate tracking of the intended subject throughout its transformations.
3Reliability
If multiple template images are used to handle subject changes, then the tracking robustness improves, but the device complexity increases
Solution Approach 1:
The patent merges the functions of multiple templates by combining the initial template image (for stability) and the update template image (for adaptability) into a unified tracking system. The arithmetic operation unit efficiently compares both templates against input images and integrates their similarity assessments, providing robust tracking without requiring separate independent systems. This merging approach maintains reliability while avoiding the complexity of fully independent parallel tracking systems.
Data Source
AI summary
A subject tracking device includes: an input unit that sequentially inputs input images; an arithmetic operation unit that calculates a first similarity level between an initial template image and a target image and a second similarity level between an update template image and the target image; a position determining unit that determines a subject position based upon at least one of the first and the second similarity level; a decision-making unit that decides whether or not to update the update template image based upon the first and the second similarity level; and an update unit that generates a new update template image based upon the initial template image multiplied by a first weighting coefficient and the target image multiplied by a second weighting coefficient, and updates the update template image with the newly generated update template image, if the update template image is decided to be updated.