Seat Belt Wear Detection Using Chest and Pelvis Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing seat belt wearing determination systems face inaccuracies in detecting correct belt usage, particularly with fake buckles and incorrect wearing methods, and may misunderstand the state of seat belt wear due to large exposed belt areas or diagonal wearing of 3-point belts.
Innovation Solution
An apparatus and method utilizing an image sensor unit to capture occupant images, a portion detection unit to identify chest and pelvis regions, and a control unit to determine seat belt wear by comparing area values of segmented regions against preset reference values, with a notification unit for abnormal wear alerts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a buckle sensor is used to determine seat belt wearing, then the system can detect when the buckle is engaged, but it cannot distinguish between correct wearing and fake buckle usage
Solution Approach 1:
The patent introduces an image sensor as an intermediary device to capture visual information of the occupant and seat belt. This intermediary provides additional data that mediates between the buckle sensor signal and the final determination, allowing the system to verify actual wearing through image analysis rather than relying solely on the buckle sensor.
Solution Approach 2:
The patent replaces the purely mechanical/electrical buckle sensor system with an optical imaging system that uses image processing algorithms. Instead of relying on electrical contacts from the buckle sensor, the system uses computer vision to detect the presence and position of the seat belt on the occupant's body, substituting mechanical detection with optical field-based detection.
2Measurement precision
If the exposed seat belt area is used to determine wearing status, then the system can detect belt presence, but it may misunderstand unbelted states as worn when large areas are exposed
Solution Approach 1:
The patent applies local quality analysis by examining specific regions of the image rather than the entire exposed area. It focuses on detecting the seat belt in specific locations such as across the chest and over the shoulder, rather than simply measuring total exposed belt area. This localized detection approach distinguishes between belt properly positioned on the body versus belt lying loosely on the seat.
3Ease of manufacture
If simple buckle sensor detection is used, then the system is easy to implement, but it cannot detect incorrect wearing methods such as wearing behind the back
Solution Approach 1:
The patent transitions from one-dimensional buckle sensor detection (simple on/off signal) to two-dimensional image space analysis. By capturing and analyzing images, the system gains spatial dimension information that allows it to detect the position and orientation of the seat belt, determining whether it is worn correctly across the body or incorrectly behind the back.
Data Source
AI summary
A seat belt wearing determination apparatus including an image sensor unit which captures a front image of a vehicle occupant, a portion detection unit which detects a chest portion and a pelvis portion of the occupant using the image, an image segmentation unit which obtains a first target region related to a seat belt by segmenting a partial image of the chest portion of the occupant into a plurality of regions and obtains a second target region related to the seat belt by segmenting a partial image of the pelvis portion into a plurality of regions, and a control unit which determines that the seat belt is worn normally when a first area value of the first target region is greater than a preset first reference value and a second area value of the second target region is greater than a preset second reference value.