Facial Expression Determination Using Angle Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing expression determination devices struggle to accurately identify facial expressions when the subject's face is not oriented at a front view, as they rely on normalization techniques that fail to effectively account for varying angles, leading to reduced identification rates and accuracy.
Innovation Solution
A processor-based system that acquires image data, detects faces, estimates the angle of the face, and determines facial expressions by using a combination of face detection, angle estimation, and expression identification functions, which includes a smile identifier and a dialogue processor to stabilize expression determination values regardless of face orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If normalization to front view face using three-dimensional face model is used, then face authentication capability is improved regardless of orientation, but expression determination accuracy deteriorates when face is not in front view
Solution Approach 1:
The system segments the face image into multiple regions (eye region, nose region, mouth region) and processes each region separately to extract orientation information. This segmentation allows accurate determination of face orientation angles without requiring full-face normalization, thereby maintaining expression determination accuracy for non-frontal views.
Solution Approach 2:
Instead of applying uniform normalization to the entire face, the system applies local processing to specific facial regions. By focusing on key landmarks in different regions and their relative positions, the system can accurately estimate face orientation and expressions without distorting the original image geometry through aggressive normalization.
2Ease of operation
If face image normalization is applied, then expression determination is simplified, but accuracy deteriorates when the original face orientation information is needed
Solution Approach 1:
The system performs preliminary detection of face orientation and key facial landmarks before expression determination. By pre-identifying the orientation angle and adjusting processing parameters accordingly, the system simplifies subsequent expression analysis while maintaining accuracy for various face orientations without requiring full normalization.
3Measurement precision
If the system only processes front view faces, then expression determination accuracy is improved, but adaptability to various face orientations deteriorates
Solution Approach 1:
The system changes processing parameters based on the detected face orientation angle. For different orientation ranges (front view, slightly turned, significantly turned), the system adjusts landmark detection thresholds, region of interest definitions, and expression evaluation criteria to maintain high accuracy across all orientations without requiring normalization.
Data Source
AI summary
A determination result is easily obtained even in expression determination on a face image that is not a front view. A Robot includes a camera, a face detector, a face angle estimator, and an expression determiner. The camera acquires image data. The face detector detects a face of a person from the image data acquired by the camera. The face angle estimator estimates an angle of the face detected by the face detector. The expression determiner determines an expression of the face based on the angle estimated by the face angle estimator.