Facial Feature Detection Using Brightness Distribution Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting facial characteristic points, such as those representing the outlines of eyes and mouths, face challenges in accuracy and robustness, particularly under varying lighting conditions and require user intervention for correction, leading to inconsistent results.
Innovation Solution
A method that detects faces, eyes, and facial characteristic points using machine learning techniques with reference data and brightness distribution analysis, setting reference lines and calculating characteristic amounts to accurately identify inner and outer corners of eyes and mouth corners, enabling efficient and precise detection without user burden.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If template matching is employed to detect facial parts, then initial detection can be achieved, but detection accuracy deteriorates under varying lighting conditions
Solution Approach 1:
The detection process is divided into multiple stages: first detecting facial parts using template matching, then detecting characteristic points on those parts using brightness distribution analysis. This segmentation allows each stage to use the most appropriate method for its specific task, maintaining simplicity while improving accuracy.
Solution Approach 2:
The patent introduces brightness distribution analysis as an intermediary step between template matching and final characteristic point detection. This intermediary method analyzes the brightness characteristics of pixels along search lines to precisely identify characteristic points, overcoming the limitations of template matching under varying lighting conditions.
2Productivity
If existing methods are used to detect facial characteristic points, then detection can be performed, but user intervention is required for correction leading to inconsistent results
Solution Approach 1:
The system performs self-correction by automatically detecting characteristic points through brightness distribution analysis without requiring user intervention. The method autonomously identifies inner and outer corners of eyes, corners of mouth, and other characteristic points, ensuring consistent results across different users and situations.
Solution Approach 2:
The patent implements a feedback mechanism where the detection system continuously refines its results by analyzing brightness distributions and adjusting characteristic point positions accordingly. This automated feedback loop eliminates the need for manual correction and ensures consistent, reliable detection results.
3Measurement precision
If comprehensive facial characteristic point detection is implemented, then detection accuracy improves, but system complexity increases
Solution Approach 1:
The complex task of detecting all facial characteristic points is segmented into manageable sub-tasks: detecting eyes and their corners, detecting mouth corners, and detecting other characteristic points. Each sub-task uses brightness distribution analysis on specific search lines, making the overall complex system tractable and implementable.
Solution Approach 2:
Different detection strategies are applied to different facial regions based on their specific characteristics. For example, eye corners are detected by analyzing brightness distributions along vertical search lines through the eye region, while mouth corners use horizontal search lines. This local customization achieves high accuracy without requiring a single overly complex universal method.
Data Source
AI summary
First, a face within an image, which is a target of detection, is detected. Detection data of the face is employed to detect eyes which are included in the face. Detection data of the eyes are employed to detect the inner and outer corners of the eyes. Detection data of the inner and outer corners of the eyes is employed to detect characteristic points of the upper and lower eyelids that represent the outline of the eyes.