Face Identification Device Reliability Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face identification methods experience decreased accuracy due to unfavorable conditions such as wearing articles, extreme lighting, or sideways-facing subjects, requiring additional processing time and uncertain restoration accuracy.
Innovation Solution
A face identification device with a reliability determination unit that assesses and weights feature values based on conditions like face direction, lighting, and worn articles, preventing accuracy deterioration by adjusting processing accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional face identification methods are used without reliability assessment, then processing is simpler and faster, but identification accuracy deteriorates under unfavorable conditions such as worn articles, extreme lighting, or sideways-facing subjects
Solution Approach 1:
The patent segments the identification process into distinct functional units: a reliability determination unit that assesses quality metrics (lighting conditions, face orientation, presence of worn articles) and an identification unit that performs feature extraction and comparison. This segmentation allows the system to evaluate reliability separately before committing to identification, thereby improving accuracy under unfavorable conditions while maintaining modular processing efficiency.
Solution Approach 2:
The reliability determination unit performs preliminary assessment of image quality and extraction conditions before the main identification process. By evaluating metrics such as lighting uniformity, face angle, and occlusion levels in advance, the system can determine whether the extracted feature values are reliable enough for accurate identification, preventing erroneous results while avoiding unnecessary processing of clearly unsuitable images.
2Measurement precision
If multiple face identification devices for various face directions are prepared, then identification accuracy is improved, but device complexity and processing time increase
Solution Approach 1:
Instead of using fixed, direction-specific identification devices, the patent implements a dynamic reliability assessment mechanism that adapts to the actual extraction conditions. The reliability determination unit evaluates face orientation and other parameters in real-time, dynamically adjusting the weighting of different feature values based on their reliability under the current conditions. This dynamic approach eliminates the need for multiple pre-configured devices while maintaining high accuracy across various face directions.
3Measurement precision
If all feature values are used with similar weighting in identification processing, then processing is simpler, but identification accuracy deteriorates when specific feature values cannot be obtained with high accuracy
Solution Approach 1:
The patent applies local quality by assigning different weights to different feature values based on their individual reliability assessments. Instead of uniform weighting, the system evaluates each feature value's reliability separately (e.g., features from well-lit regions receive higher weights than those from shadowed regions) and adjusts their contribution to the final identification result accordingly. This localized quality assessment improves accuracy by emphasizing reliable features while minimizing the impact of unreliable ones.
Data Source
AI summary
A method for identifying a person includes the steps of detecting the face of a person in an input image, determining the reliability of each feature value from the input image, and obtaining a plurality of feature values from the detected face. Based on the feature values obtained from the detected face, the feature values stored on a storage unit, and the reliability of each feature value, an identification result according to the detected face is decided. A device includes the components for implementing said method.