Image Capture Device Dual Area Shifting for Tracking Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image capture devices with face detection functions struggle to accurately track features in moving pictures, especially when the feature being tracked cannot be searched for accurately, leading to potential mislocation or failure in tracking.
Innovation Solution
The device employs a dual area shifting mechanism, where a first area associated with face information and a second area associated with color information are shifted independently, with the second area's destination determined based on statistical relations defined from past image data, ensuring higher accuracy even in uncertain tracking situations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single area is shifted based on face detection results, then the tracking process is simple, but the tracking accuracy deteriorates when the face cannot be searched accurately
Solution Approach 1:
The patent divides the tracking area into multiple independent areas (first area for face information, second area for color information) that can be shifted independently. This segmentation allows each area to be optimized for its specific feature type, improving overall tracking accuracy without significantly increasing system complexity.
Solution Approach 2:
The patent introduces environment information as an intermediary element that helps determine the destination of area shifts. When face detection accuracy is low, the environment information provides additional context to improve the accuracy of feature location estimation.
2Productivity
If the search range is limited by rough location estimation accuracy, then the search process is efficient, but the tracking reliability deteriorates when rough location accuracy is poor
Solution Approach 1:
The patent makes the search range dynamic by adjusting it based on the detected feature type and confidence level. For high-confidence detections, a limited search range maintains efficiency. For low-confidence detections, the search range automatically expands to improve reliability, allowing the system to adapt to varying tracking conditions.
Solution Approach 2:
The patent changes the search parameter (search range size) based on the detection confidence and feature type. When tracking faces with high accuracy, a smaller search range is used. When accuracy is poor or tracking color information, the search range is expanded to ensure reliable feature location.
3Measurement precision
If multiple areas are shifted independently based on different feature information, then the tracking accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent creates a universal area shifting mechanism that can handle multiple feature types (face information, color information) using the same basic framework. The first area shifting section and second area shifting section share common functionality but are specialized for different feature types, reducing overall complexity compared to separate dedicated systems.
Data Source
AI summary
An image capture device according to the present invention includes: an image capturing section for capturing a subject's image and generating image data, thereby outputting a moving picture; a first area shifting section for shifting a first area, which is associated with a first kind of feature information that represents a first feature in the image data, based on a result of search for the first kind of feature information through the image data; and a second area shifting section for shifting, independently of the first area, a second area, which is associated with a second kind of feature information that represents a second feature in the image data, based on a result of search for the second kind of feature information through the image data. If unable to detect the second kind of feature information, the second area shifting section shifts the second area with respect to the location of the first area.