Gaze Detection Correction Using Reference Distribution Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze direction detection technologies face challenges in accurately detecting a driver's gaze direction without calibration, particularly due to individual differences, and are affected by movements unrelated to the primary gaze direction.
Innovation Solution
A gaze direction detection device comprising a detector, determiner, generator, calculator, and corrector that calculates a correction parameter based on the difference between a reference distribution and the detected gaze distribution, allowing for accurate correction of the gaze information without pre-calibration, using a corneal reflection method and near-infrared imaging.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If calibration is performed to account for individual differences, then measurement precision is improved, but loss of time increases due to the calibration process
Solution Approach 1:
The system performs preliminary calibration during the period when the driver is not operating the vehicle (e.g., during parking or before driving starts). The gaze distribution is measured and stored in advance, so that when the vehicle is in operation, the pre-calibrated data can be directly used without requiring additional calibration time. This resolves the contradiction by performing the time-consuming calibration action beforehand, when it does not interfere with the primary driving function.
2Measurement precision
If calibration is performed for each individual, then measurement precision is improved, but device complexity increases due to storing and managing individual calibration data
Solution Approach 1:
The system uses a universal coordinate system that can represent gaze distributions for multiple different drivers. By transforming individual gaze distributions into this common coordinate system, the system can handle multiple users without requiring separate complex calibration data structures for each person. This universal framework simplifies the overall system architecture while maintaining the ability to accurately represent individual differences through the transformation process.
3Measurement precision
If the detection system accounts for all gaze variations including movements, then measurement precision is improved, but ease of operation worsens due to increased system complexity
Solution Approach 1:
The system extracts and separates the essential gaze direction information from irrelevant movements and distractions. By focusing only on the meaningful gaze patterns that indicate actual driver attention and filtering out unnecessary variations, the system maintains high measurement precision while keeping the operation simple and intuitive for the driver.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables precise and calibration-free detection of a driver's gaze direction, reducing errors caused by individual differences and movements, and effectively tracks the driver's focus even when not using navigation or audio systems.
Implementation Method 1
radiating near-infrared light of an LED or the like on the cornea
Implementation Method 2
a reflected image (Purkinje image) that is formed by radiating near-infrared light
Data Source
AI summary
A gaze direction detection device according to the present technology includes a detector for detecting a gaze of a driver over a predetermined period of time, a determiner for outputting second gaze information indicating that the driver is gazing, from first gaze information detected by the detector, a generator for generating a gaze distribution from the second gaze information output by the determiner, and a corrector for correcting the first gaze information detected by the detector, where the corrector calculates a center of a reference distribution that is set in advance and a center of the gaze distribution generated by the generator, and causes the center of the reference distribution and the center of the gaze distribution to overlap each other, and then calculates a correction parameter based on a difference between the reference distribution and the gaze distribution, and corrects the first gaze information with the correction parameter.