Pupil Hidden State Detection via Eye Region Brightness Gradient

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing driver state detection systems face difficulties in accurately determining if a pupil is hidden behind glasses, masks, hair, or other obstructions, as they require detection of the hiding object to assess the pupil's state, which is challenging due to the variety of forms and types of obstructions.

Innovation Solution

A drive assist display apparatus and vehicle-mounted camera system that directly detects the hidden state of a pupil by analyzing the vertical distribution of brightness gradients in the eye region, without needing to detect the obstructing object, using an imaging section, eye region detection, visual line direction detection, brightness gradient calculation, and a pupil hidden judging section to determine if the pupil is hidden based on intensity differences and evaluation values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the frame of the glasses is detected to determine pupil hidden state, then the pupil hidden state can be detected, but the detection accuracy deteriorates because there have so far been a wide variety of forms of the frames of the glasses

Engineering Contradiction:
Improvepupil hidden state detection reliabilityVSAvoidframe detection precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The invention extracts only the necessary information (pupil hidden state) directly from the eye region image without attempting to detect or identify the obscuring object (glasses frame, mask, hair). By taking out only the essential detection target (pupil visibility) and ignoring the complex varying obscuring objects, the system achieves reliable detection across diverse scenarios without being affected by frame variety

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The invention introduces brightness gradient analysis as an intermediary method to detect pupil hidden state. Instead of directly detecting the pupil or the obscuring object, the system uses brightness gradient characteristics in the vertical direction of the eye region as a mediator to infer whether the pupil is hidden. This intermediary approach enables accurate detection without needing to identify the specific type of obscuring object

Inventive Principle:
Principle #24Intermediary (Mediator)

2Device complexity

If a direct detection method for pupil hidden state is used without detecting the hiding object, then the detection process is simplified, but the detection accuracy may worsen without proper analysis methods

Engineering Contradiction:
Improvedetection process complexityVSAvoidpupil hidden state detection precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The invention changes the detection parameter from direct pupil visibility or frame identification to brightness gradient intensity in the vertical direction of the eye region. By transforming the detection task into analyzing brightness gradient parameters, the system achieves both simplified processing (no need to detect obscuring objects) and high precision (gradient intensity reliably indicates pupil hidden state)

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS8405714B2Pupil hidden state detecting apparatus and vehicle-mounted camera employing the same
Publication Date: 2013.03.26 PANASONIC AUTOMOTIVE SYST CO LTD
  • US8405714B2 patent drawing
  • US8405714B2 patent drawing
  • US8405714B2 patent drawing

AI summary

Provided is a pupil hidden state detecting apparatus which can detect directly a hidden state of a pupil without detecting a hiding object. The pupil hidden state detecting apparatus 20 is comprises: a visual line direction detecting section 22 operable to detect a direction of a visual line of a driver; a brightness gradient calculating section 23 operable to calculate a brightness gradient of an image of an eye area; a gradient intensity calculating section 24 operable to calculate a gradient intensity vertical distribution; a gradient intensity storing section 25 adapted to store therein the data of a reference gradient intensity vertical distribution under the state that the direction of the visual line of the driver is toward an instrument panel; a difference calculating section 26 operable to calculate a difference between the gradient intensity vertical distribution of the current frame and the reference gradient intensity vertical distribution; an evaluation value calculating section 27 operable to calculate an evaluation value indicating a degree of possibility that the pupil of the driver is hidden; and a pupil hidden judging section 29 operable to judge that the pupil is hidden, when the evaluation values indicative of the pupil being hidden is continued.