Single Camera Driver State Detection Using Facial Feature Extraction
Find Innovative SolutionsGenerate 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
Engineering Contradiction Analysis
1Measurement precision
If multiple cameras are used to determine driver state, then measurement precision is improved, but device complexity and cost increase
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
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
2Measurement precision
If two cameras are used for eye gaze measurement, then measurement precision is improved, but device complexity and cost increase
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
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
3Device complexity
If single camera system is used, then device complexity is reduced, but measurement precision of three-dimensional head pose deteriorates
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
4Device complexity
If single camera system is used, then device complexity is reduced, but driver state detection capability deteriorates
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
Data Source
Figure 1~2
Figure 3
Figure 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.