Adaptive Facial Recognition Accuracy Control for Lighting Variations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Facial recognition systems face decreased detection and recognition accuracy due to varying lighting environments, leading to false detections and recognitions, particularly in indoor low-light and outdoor sunny conditions.
Innovation Solution
A facial recognition apparatus and method that includes a photographing parameter input unit, a lighting information estimation unit, and a recognition accuracy control unit, which receives and processes photographing parameters to estimate lighting conditions and adjust recognition accuracy parameters, such as feature point usage, weights, and threshold settings, to minimize false recognition based on illuminance levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If facial recognition is performed in varying lighting environments, then the system can operate in diverse conditions, but detection accuracy and recognition accuracy decrease leading to false detections
Solution Approach 1:
The patent changes the parameter of recognition accuracy dynamically based on lighting conditions. The recognition accuracy control unit adjusts recognition parameters (such as threshold values, feature point weights, or matching criteria) according to the estimated lighting information, allowing the system to maintain high detection accuracy across different lighting environments by adapting its operational parameters to current conditions
Solution Approach 2:
The patent implements a feedback mechanism where the lighting information estimation unit continuously estimates current lighting conditions and feeds this information to the recognition accuracy control unit. This closed-loop feedback allows the system to automatically adjust recognition parameters in response to changing lighting environments, resolving the contradiction between adaptability and measurement precision
2Reliability
If recognition accuracy parameters are adjusted based on lighting information, then false recognition rates decrease, but system complexity increases
Solution Approach 1:
The patent implements a universal lighting information estimation mechanism that serves multiple purposes: it estimates lighting conditions for accuracy control and can also inform other system components. The recognition accuracy control unit uses the same lighting information to adjust multiple recognition parameters simultaneously, reducing overall system complexity by sharing a common estimation foundation across different functions
Solution Approach 2:
The system performs self-adjustment of recognition parameters based on its own lighting estimates without requiring external calibration or manual intervention. The recognition accuracy control unit automatically modifies recognition parameters using lighting information generated by the system's own estimation unit, enabling the system to serve itself and maintain high reliability without adding complex external control mechanisms
Data Source
AI summary
Photographing parameter(s) include a parameter(s) relating to exposure time and diaphragm, and is used upon capturing an image of a target. A database stores a table indicating a relationship between a value(s) of the photographing parameter(s) and a recognition accuracy threshold that is used upon determining a result of recognition. The recognition accuracy threshold associated with received photographing parameter(s) is extracted from the database, and the image is received. Recognition of the target is performed from the received image in accordance with recognition accuracy threshold.