Subject Detection Tracking Using Adaptive Search Areas
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to accurately detect a local area of a specific subject unless the entire area of the subject has been previously detected, limiting the effectiveness of subject detection.
Innovation Solution
An image processing apparatus with a processor and memory that includes a first detection unit for detecting a first area, a second detection unit for detecting a broader area if the first area is not detected in subsequent images, and a tracking unit for subject tracking using detection results from both units.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only a first detection unit detecting a first area is used, then the detection precision for local areas is improved, but the reliability of continuous detection deteriorates when the subject is not detected in subsequent images
Solution Approach 1:
The detection system is divided into two independent detection units: a first detection unit that detects a first area (entire subject) and a second detection unit that detects a second area (local area). Each unit operates independently to detect different spatial scopes, allowing the system to maintain both precision for local areas and reliability for continuous detection by having the second unit serve as a backup when the first unit fails to detect the subject in subsequent images.
Solution Approach 2:
The system performs preliminary detection using the first detection unit to identify the entire subject area. Based on this preliminary detection result, the system prepares and positions the second detection unit to detect the local area. This preliminary action ensures that when the subject is not detected in subsequent images, the second detection unit can immediately take over with pre-positioned detection parameters, maintaining continuous detection reliability.
2Reliability
If a second detection unit detecting a broader area is added, then the reliability of continuous detection is improved, but the device complexity increases
Solution Approach 1:
The second detection unit is designed with multi-functionality: it can detect the local area when the first detection unit successfully identifies the subject, and it can also detect the entire subject when the first detection unit fails. This universal design allows a single additional unit to serve multiple purposes, improving reliability without requiring separate backup systems for each detection failure scenario.
Solution Approach 2:
The detection architecture is structured in a nested manner where the second detection unit encompasses a broader detection area that includes both the first area (entire subject) and the second area (local area). The detection parameters and processing logic of the first unit are nested within the overall detection framework, allowing the second unit to inherit and build upon the first unit's detection capabilities while adding broader coverage.
3Measurement precision
If tracking is performed using only the first detection unit, then the detection precision for local areas is maintained, but the loss of information occurs when the subject is not detected in subsequent images
Solution Approach 1:
The system implements a feedback mechanism where the tracking unit continuously monitors detection results from both detection units. When the first detection unit fails to detect the subject in subsequent images, the feedback loop triggers the second detection unit to perform detection. The detection results from either unit are fed back to update the tracking information, ensuring that no subject information is lost and tracking continuity is maintained based on real-time detection feedback.
Solution Approach 2:
The tracking unit performs preliminary tracking using detection results from the first detection unit. When the first unit fails to detect the subject, the tracking unit has already established baseline tracking information that can be combined with detection results from the second unit. This preliminary action ensures that tracking information is not lost during detection failures, as the system has pre-established tracking context that can be recovered or supplemented by the second detection unit.
Data Source
AI summary
An image processing apparatus acquires image data, detects a first area of a subject from image data, detects a second area of a subject from image data, and tracks the subject, wherein, when the first area is being detected from first image data, detection of the second area is performed in an area corresponding to the first area, and, in a case where the first area is detected in the first image data and the first area is not detected in second image data captured after the first image data, detection of the second area is performed in an area broader than the first area detected in the first image data, and wherein tracking of the subject is performed with use of at least one of detection results obtained in the first area and the second area.