Camera Subject Tracking Using Multi-Template Barycenter Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing photographic subject tracking devices experience decreased accuracy due to abrupt changes between frames, leading to false matching and reduced tracking precision.
Innovation Solution
A computer program product that performs template matching, classification, and tracking of photographic subjects across frames by using multiple template images of different sizes, updating these images based on movement, and specifying the subject position using the barycenter of representative points to maintain accurate tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If template matching is performed using a single template image, then the matching process is simple and fast, but tracking accuracy decreases when abrupt changes occur between frames
Solution Approach 1:
The patent divides the template matching process into multiple segments by using multiple template images instead of a single template. Each template image represents a different state or variation of the photographic subject, allowing the system to handle abrupt changes between frames more effectively. This segmentation of the template into multiple versions resolves the contradiction by improving tracking reliability while maintaining manageable process complexity through structured classification of matching regions.
Solution Approach 2:
The patent changes the parameter of template image variation by incorporating multiple template images with different characteristics into the matching process. By varying the template parameters (such as appearance variations, pose changes, or environmental conditions), the system can adapt to abrupt changes between frames. This parameter change approach improves tracking accuracy without significantly increasing complexity, as the variation is systematic rather than random.
2Measurement precision
If multiple template images are used to improve tracking accuracy, then false matching is reduced, but the classification and processing complexity increases
Solution Approach 1:
The patent applies local quality by classifying matching regions based on their specific characteristics and mutual distances rather than treating all regions uniformly. Each matching region is evaluated locally for its proximity to other regions and its relevance to the photographic subject. This local quality assessment improves measurement precision by selecting the most accurate regions while managing complexity through localized rather than global processing.
Solution Approach 2:
The patent uses partial action by selecting only the necessary number of matching regions (half or more of the extracted regions) for specifying the photographic subject position. Instead of processing all possible matching regions, the system selectively processes a sufficient subset that achieves accurate tracking. This partial action approach improves measurement precision while reducing the effective complexity of the classification step by focusing computation on the most critical regions.
3Reliability
If template images are updated to adapt to shape changes, then tracking accuracy is maintained, but processing time and computational load increase
Solution Approach 1:
The patent applies preliminary action by preparing multiple template images in advance that represent different states of the photographic subject. These pre-prepared templates are used during the actual tracking process, eliminating the need for real-time template generation. This preliminary preparation improves tracking reliability under shape changes while minimizing processing time, as the templates are ready for immediate use rather than being generated on-demand during tracking.
Solution Approach 2:
The patent incorporates feedback mechanisms that monitor tracking results and determine when template images need to be updated. By using feedback from the tracking process to trigger template updates only when necessary (when abrupt changes are detected), the system maintains high tracking accuracy while reducing unnecessary processing time. The feedback loop ensures that template updating occurs only when it improves tracking reliability, avoiding constant processing overhead.
Data Source
AI summary
A computer executes performing template matching upon each frame of an input image with each of a plurality of template images, extracting a plurality of regions whose similarities with each of the template images are the highest from within the input image, classifying the extracted regions into regions that are to be used for specifying a photographic subject position within the input image and another region, on the basis of the mutual distances between the extracted regions within the input image, specifying the photographic subject position, on the basis of the positions of the regions that have been classified as the regions that are to be used for specifying the photographic subject position within the input image, and tracking the movement of the photographic subject over the image consisting of a plurality of frames, by tracking the photographic subject position between frames.