Real-Time Object Tracking via Localized Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking methods are inefficient in real-time processing, particularly when dealing with sudden changes in lighting and partial occlusion, and they lack accuracy in estimating object poses and patterns.
Innovation Solution
A method and apparatus that detect object feature points in one frame, estimate candidates in the next frame using pixel value differences, and search a surrounding region for a template matching process, incorporating the Lucas-Kanade-Tomasi algorithm for optical flow calculation and template matching to reduce processing time and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing object tracking methods are used, then object tracking can be performed, but processing time is excessive and real-time tracking is difficult
Solution Approach 1:
The patent performs preliminary action by pre-calculating and storing gradient values (Ix, Iy) and template gradients in the first frame before tracking begins. This preprocessing step allows the tracking algorithm to use these pre-computed values in subsequent frames, avoiding redundant calculations and significantly reducing processing time for real-time tracking.
Solution Approach 2:
The patent segments the tracking process into distinct phases: candidate feature point generation using pixel differences, template matching in local regions, and feature point refinement. This segmentation allows each phase to be optimized independently, with template matching only performed in small local regions around candidate points rather than across the entire image, thus reducing overall processing time.
2Reliability
If existing optical flow methods are used for object tracking, then object motion can be tracked, but the methods are weak in handling partial occlusion and lighting changes
Solution Approach 1:
The patent introduces template matching as an intermediary step between pixel difference-based candidate generation and final feature point determination. The template matching process uses correlation coefficients to evaluate similarity between templates and image regions, providing robustness against lighting changes and occlusion by relying on overall pattern similarity rather than pixel-by-pixel comparison, thereby maintaining tracking accuracy in challenging conditions.
Solution Approach 2:
The patent changes parameters dynamically by adjusting the correlation coefficient threshold for template matching acceptance. When lighting conditions or occlusion levels change, the system can adapt the threshold parameter to maintain reliable tracking, allowing the same algorithm to handle varying environmental conditions effectively while preserving measurement precision.
3Measurement precision
If template matching is performed over the entire image, then accurate feature point detection can be achieved, but processing time increases significantly
Solution Approach 1:
The patent applies local quality by performing template matching only in small local regions surrounding candidate feature points generated from pixel differences, rather than across the entire image. This localized approach maintains detection accuracy by focusing computational resources on relevant areas while dramatically improving processing efficiency by avoiding unnecessary comparisons in regions where feature points are unlikely to exist.
Solution Approach 2:
The patent uses partial action by generating candidate feature points using pixel differences as a preliminary filter, then performing exhaustive template matching only for these candidates. This two-stage approach performs partial template matching (only on candidates) rather than complete image-wide matching, achieving sufficient accuracy for real-time tracking while reducing processing time by focusing computational effort only where needed.
Data Source
AI summary
A method and apparatus for tracking an object, and a method and apparatus for calculating object pose information are provided. The method of tracking the object obtains object feature point candidates by using a difference between pixel values of neighboring frames. A template matching process is performed in a predetermined region having the object feature point candidates as the center. Accordingly, it is possible to reduce a processing time needed for the template matching process. The method of tracking the object is robust in terms of sudden changes in lighting and partial occlusion. In addition, it is possible to track the object in real time. In addition, since the pose of the object, the pattern of the object, and the occlusion of the object are determined, detailed information on action patterns of the object can be obtained in real time.