Subject Tracking via Template and Histogram Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subject tracking techniques face challenges in accurately tracking subjects with changes in appearance, as template matching excels in classification between similar subjects but struggles with changes, while histogram matching is robust against appearance changes but weak in classification between similar subjects, and combining these methods is difficult due to differing properties.
Innovation Solution
A subject tracking apparatus that combines template matching and histogram matching by registering partial image areas as templates and generating histograms based on pixel values, with frequent updates for template registration and less frequent updates for histogram registration, allowing for precise estimation and tracking of subject areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template matching is used for subject tracking, then classification between similar subjects is improved, but tracking robustness against appearance changes deteriorates
Solution Approach 1:
The patent combines template matching and histogram matching into a unified tracking system. Template matching provides precise classification between similar subjects by comparing image patterns, while histogram matching provides robustness against appearance changes by comparing color distribution histograms. The integration allows the system to leverage the strengths of both methods simultaneously, resolving the contradiction between classification precision and tracking robustness.
Solution Approach 2:
The patent dynamically adjusts the weighting between template matching results and histogram matching results based on tracking conditions. When the subject appearance is stable, template matching is given higher weight for precise classification. When appearance changes are detected, histogram matching is given higher weight for robust tracking. This dynamic parameter adjustment resolves the contradiction by adapting the matching strategy to current conditions.
2Reliability
If histogram matching is used for subject tracking, then robustness against appearance changes is improved, but classification between similar subjects deteriorates
Solution Approach 1:
The patent integrates histogram matching with template matching to compensate for histogram matching's weakness in classifying similar subjects. The template matching component provides discriminative power for distinguishing similar subjects through detailed image pattern comparison, while histogram matching ensures robustness. The combined approach allows the system to achieve both high classification precision and tracking robustness simultaneously.
Solution Approach 2:
The patent uses template matching as an intermediary to enhance the classification capability of histogram matching. When histogram matching identifies potential subject locations, template matching acts as a verification step to distinguish between similar subjects. This intermediary mechanism resolves the contradiction by adding discriminative power to the otherwise appearance-change-robust but classification-weak histogram matching method.
3Reliability
If template matching and histogram matching are combined, then tracking performance is improved, but system complexity increases
Solution Approach 1:
The patent divides the tracking system into distinct modules: a template matching unit, a histogram matching unit, and a result integration unit. Each module performs a specific function independently. The template matching unit handles pattern-based tracking, the histogram matching unit handles color-distribution-based tracking, and the integration unit combines their results. This segmentation manages complexity by organizing the combined system into manageable, functionally-separated components.
Solution Approach 2:
The patent implements periodic updates of the template and histogram based on tracking confidence levels. When tracking is stable and confident, updates occur less frequently. When tracking uncertainty increases or appearance changes are detected, updates occur more frequently. This periodic action mechanism manages computational complexity by adjusting the frequency of expensive template and histogram regeneration operations based on actual tracking needs.
Data Source
AI summary
The subject tracking apparatus comprises: a first registering unit configured to register a partial area as a template indicative of a subject in one image of supplied images; a first matching unit configured to estimate a subject area by collating a partial area in newly supplied images with the template registered by the first registering unit; a second registering unit configured to register a histogram generated based on a pixel value of a partial area indicative of the subject in one image of supplied images; a second matching unit configured to estimate a subject area by collating a histogram of a partial area in newly supplied images with the histogram registered by the second registering unit; and a tracking area determination unit configured to determine a tracking area based on estimation results by the first matching unit and the second matching unit. The first registering unit and the second registering unit allow at least one of the template and the histogram previously registered to be updated, and the update for the registration of the template by the first registering unit is more frequently than that for the registration of the histogram by the second registering unit.