Facial Image Detection Using Grayscale Intensity and Adaptive Boosting

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing facial image detection methods face challenges in achieving high precision and robustness against varying face poses, especially when relying on color information and illumination, which affects detection rates and accuracy, particularly in single-image searches.

Innovation Solution

The method involves pre-processing images to minimize illumination effects, creating facial records with n×n pixel-sized windows, using strong classifiers learned through adaptive boosting for face and eye detection, and applying clustering techniques to determine accurate center coordinates of faces and eyes, ensuring robust detection across different face poses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If color information and segmentation methods are used for facial image detection, then the detection process is simplified, but the detection rate and precision deteriorate due to dependence on illumination conditions

Engineering Contradiction:

Improvedetection process complexityVSAvoidface detection rate

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent transforms the detection approach by changing parameters from color-based detection to grayscale intensity-based detection. Specifically, it uses the relationship between eyebrow and cheek pixel intensities (parameter changes in intensity ratios) to identify facial regions, which remains effective across varying illumination conditions and resolves the contradiction between simplified processing and reliable detection

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical/optical approach of color-based detection with a computational approach using grayscale intensity analysis and mathematical relationships between pixel values. This substitution allows the system to achieve both simplified processing and illumination-invariant detection by relying on structural intensity patterns rather than color information

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Speed

If adaptive boosting technology is used for real-time facial image detection, then detection speed is improved, but detection precision deteriorates in single-image searches requiring high accuracy

Engineering Contradiction:

Improvedetection speedVSAvoidface detection precision

Core Design Contradiction:

SpeedVSMeasurement precision

Solution Approach 1:

The patent divides the facial detection task into multiple segmentation stages: initial face region detection using adaptive boosting, followed by subdivision into sub-regions (eyebrow, cheek, mouth areas), and further segmentation into individual facial features (eyes, eyebrows, mouth). This multi-level segmentation enables both rapid initial detection and precise final localization, resolving the speed-precision contradiction

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent adds dimensional complexity by introducing vertical position analysis and multi-scale detection. It detects facial features not only in the horizontal dimension but also analyzes vertical relationships between features (e.g., eyebrows above cheeks, mouth below cheeks). This dimensional expansion allows the system to maintain high speed through efficient 2D scanning while achieving high precision through multi-dimensional feature verification

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

3Reliability

If learning database is changed for offline face learning, then face detection rate varies, but system adaptability to different face poses deteriorates

Engineering Contradiction:

Improveface detection rateVSAvoidrobustness to face pose changes

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal detection framework that functions across multiple scenarios and face poses. By using illumination-invariant grayscale intensity relationships and multi-scale detection, the system achieves universality that works for front-facing, profile, and intermediate poses without requiring pose-specific training databases, thus maintaining both high detection rate and adaptability

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent performs preliminary detection of the overall face region before analyzing specific facial features. This preliminary action establishes a search space that adapts to different poses, and subsequent feature detection (eyebrows, eyes, mouth) is performed within this adapted context. The system also performs preliminary normalization of intensity relationships to account for illumination variations before pose-specific analysis, enabling robust multi-pose detection

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS8326000B2Apparatus and method for detecting facial image

Publication Date: 2012.12.04 ELECTRONICS & TELECOMM RES INST

AI summary

A method of detecting a facial image includes pre-processing an image; and detecting a face region from the pre-processed image to create facial records of the detected face region. Further, the method of detecting the facial image includes detecting the facial image by creating coordinates of the face and eyes in the input image by using the facial records.