Single Camera Driver State Detection Using Facial Feature Extraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vehicle monitoring systems require multiple cameras to accurately determine driver state, which is costly and inefficient, while single-camera systems can only detect drowsiness and lack comprehensive measurement of three-dimensional head pose.

Innovation Solution

A low-cost vehicle operator state detection system using a single camera to calculate head pose by processing three facial features, focusing on yaw and pitch components, and generating a non-stereo video image to track and recognize driver states, with output signals triggering safety responses.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multiple cameras are used to determine driver state, then measurement precision is improved, but device complexity and cost increase

Engineering Contradiction:
Improvedriver state detection accuracyVSAvoidnumber of cameras
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and processes only three specific facial features (two eyes and nose) from a single camera image to calculate head pose. By focusing on these key features rather than using multiple cameras to capture the entire face, the system achieves accurate head pose measurement with reduced device complexity

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system segments the facial recognition task by identifying and processing only the critical three points (right eye, left eye, nose) rather than analyzing the entire face. This segmentation allows accurate head pose calculation from a single camera image, resolving the contradiction between precision and complexity

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If two cameras are used for eye gaze measurement, then measurement precision is improved, but device complexity and cost increase

Engineering Contradiction:
Improveeye gaze direction accuracyVSAvoidnumber of cameras
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical/stereo vision approach (requiring two cameras for 3D eye gaze calculation) with a computational geometry approach using trigonometric calculations on 2D image coordinates. This substitution allows accurate eye gaze measurement from a single camera by calculating angles based on the positions of three facial features

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system transforms the 3D eye gaze measurement problem into a 2D image processing problem. By using trigonometric calculations on the 2D coordinates of facial features in the camera image, the system achieves 3D head pose and eye gaze direction without requiring stereo vision from multiple cameras

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Device complexity

If single camera system is used, then device complexity is reduced, but measurement precision of three-dimensional head pose deteriorates

Engineering Contradiction:
Improvenumber of camerasVSAvoidthree-dimensional head pose accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces the mechanical stereo vision system with a computational geometry system. By using trigonometric calculations on the 2D image coordinates of three facial features, the system accurately computes three-dimensional head pose parameters (yaw, pitch, roll angles) from a single camera, eliminating the need for multiple cameras while maintaining measurement precision

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

4Device complexity

If single camera system is used, then device complexity is reduced, but driver state detection capability deteriorates

Engineering Contradiction:
Improvenumber of camerasVSAvoiddriver state measurement comprehensive
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent makes the single camera system universal by using it to detect multiple driver state indicators (head pose, eye gaze direction, eye closure duration) through different processing algorithms. The same single camera captures data for various driver state assessments, making the system versatile without requiring multiple specialized cameras

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP1801730B1Method of detecting vehicle-operator state
Publication Date: 2011.09.14 DELPHI TECHNOLOGIES INC
  • EP1801730B1 patent drawingFigure 1~2
  • EP1801730B1 patent drawingFigure 3
  • EP1801730B1 patent drawingFigure 4~5

AI summary

A method of detecting the state of a driver of a vehicle utilizes a low-cost operator state detection/monitoring system (24) having no more than one camera (26) located preferably in the vehicle (22) and directed toward a driver (30). A processor (48) of the detection system processes (52) preferably three points (36, 38, 40) of the facial feature (32) of the driver to calculate (70) head pose and thus determine driver state (i.e. distracted, drowsy, etc.). The head pose is generally a three dimensional vector that includes the three angular components of yaw, pitch and roll. Preferably, an output signal (76) of the processor is sent to a counter measure system (44) to alert the driver and/or accentuate vehicle safety response.