Image Feature Calculation Using Regional Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image data processing devices struggle to accurately classify images from different events, such as sea bathing and ski trips, due to similar image feature information extracted from primary colors like blue and white, which leads to incorrect categorization.
Innovation Solution
An image data processing device that calculates image feature information by giving more weight to pixels in a region around the face, as photographers tend to capture event features in these areas, enhancing the representation of event-specific features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image feature information is calculated based on primary colors (blue, white) from the entire image, then the calculation is simple and fast, but the classification precision deteriorates because images from different events (sea bathing and ski trip) have similar color features
Solution Approach 1:
The patent divides the image into multiple regions (face region, body region, background region) and calculates image features separately for each region. This segmentation allows the system to focus on event-specific features in the background region while maintaining computational efficiency, resolving the contradiction between simple calculation and accurate classification.
Solution Approach 2:
The patent applies different weighting factors to different image regions, giving higher weight to the background region where event-specific features are located and lower weight to the face and body regions. This local quality approach enables the system to emphasize relevant features for classification while maintaining overall calculation efficiency.
2Loss of information
If image feature information is calculated based on the entire image including all pixels, then all image features are captured, but the classification precision deteriorates because features from irrelevant regions (face, body) dominate and mask event-specific features
Solution Approach 1:
The patent segments the image into face region, body region, and background region, then calculates features for each segment separately. This allows the system to preserve relevant event-specific information from the background while excluding dominant but irrelevant features from the face and body regions, improving classification precision.
Solution Approach 2:
The patent applies region-specific weighting where the background region receives higher weight for event classification purposes. This local quality approach ensures that event-specific features from the background are preserved and emphasized, while reducing the influence of face and body features that do not contribute to event classification.
3Device complexity
If uniform weighting is applied to all pixels in the image, then the calculation process is simple, but the classification precision deteriorates because pixels in event-relevant regions (background) are treated equally with pixels in event-irrelevant regions (face, body)
Solution Approach 1:
The patent implements local quality by assigning different weighting factors to different image regions. The background region is given higher weight because it contains event-specific features, while face and body regions receive lower weight. This approach maintains relatively simple calculation processes while significantly improving classification precision through intelligent regional weighting.
Data Source
AI summary
When photographing images including a person's face at an event, photographers tend to photograph images so that features of the event appear in a region around the person's face. An image data processing device of the present invention extracts image feature information so that an image feature calculated based on pixels in the region around the person's face, which tends to represent features of an event, is reflected more than that calculated based on pixels in a region remote from the person's face, which tends not to represent features of an event. This allows the image data processing device to calculate image feature information reflecting features of an event more than that calculated by a conventional image data processing device. The image data processing device therefore improves classification precision compared to the conventional device when classifying images using the image feature information calculated by the image data processing device.