Camera Vehicle Width Detection Using Wheel Outer Edges

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing camera systems for motor vehicles inaccurately determine the width and position of target vehicles due to varying bounding box sizes and detection errors, particularly affecting collision warning and automatic brake systems.

Innovation Solution

A method using a camera system to detect the wheels of a target vehicle within a region of interest, determining the width by identifying the lateral outer edges of the wheels, which are invariant points, and refining the bounding box to provide a precise width measurement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If detection algorithms output a rectangular bounding box to specify target vehicle position and dimensions, then the target vehicle can be detected and tracked, but the width and height are specified inaccurately and the bounding box size varies over image sequences

Engineering Contradiction:
Improvetracking accuracyVSAvoidwidth and height measurement accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent extracts the wheels from the target vehicle as a separate feature for measurement. Instead of using the bounding box dimensions directly, the method identifies wheel regions within the bounding box and measures the distance between wheel outer edges to determine the true vehicle width, extracting this information from a specific sub-region rather than the entire bounding box

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies local quality by focusing measurement on specific critical features (wheels) rather than the entire vehicle轮廓. The wheel regions are identified and processed with specialized algorithms to detect outer edges accurately, while other parts of the vehicle are less critical for the measurement task

Inventive Principle:
Principle #3Local quality

2Measurement precision

If exterior points of the target vehicle are determined to measure width, then the width can be calculated, but the exterior points vary over image sequences due to detection errors and the measurement is associated with high noise

Engineering Contradiction:
Improvewidth measurement capabilityVSAvoidmeasurement consistency over time
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent performs preliminary action by pre-identifying the wheel regions within the bounding box before width measurement. The wheel detection and outer edge identification are performed as preparatory steps that establish stable reference points, filtering out detection errors that would affect other exterior points

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses the wheel outer edges as a copy or surrogate for the vehicle width measurement. Instead of measuring the vehicle exterior points directly (which are noisy), it measures the wheel boundaries which provide a more stable and consistent reference that can be reliably copied across image sequences

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3087532B1Method for determining a width of a target vehicle by means of a camera system of a motor vehicle, camera system and motor vehicle
Publication Date: 2023.09.06 VALEO SCHALTER & SENSOREN GMBH
  • EP3087532B1 patent drawingFigure 1~2
  • EP3087532B1 patent drawingFigure 3~5
  • EP3087532B1 patent drawingFigure 6~7

AI summary

The invention relates to a method for determining a width (w') of a target vehicle (6) by means of a camera system (2) of a motor vehicle (1), wherein an image (7) of an environmental region (4) of the motor vehicle (1) is provided by a camera (3) of the camera system (2) and the width (w') of the target vehicle (6') in the image (7) is determined by an image processing device of the camera system (2). The target vehicle (6') is detected in the image (7) by means of a detection algorithm, wherein a partial region (8) of the image (7) is determined by the detection algorithm, in which the target vehicle (6') is depicted. A region of interest (ROI) is determined within the partial region (8) by the image processing device such that wheels (9, 10) of a common axle of the target vehicle (6') are depicted in the region of interest (ROI). Respective outer edges (11, 12) of the depicted wheels (9, 10) are identified within the region of interest (ROI), and the width (w') of the depicted target vehicle (6') is determined as the distance between the outer edges (11, 12).