Camera-Based Road Surface Modeling for Reflection-Robust Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vehicle environment modeling systems face challenges in accurately detecting road surfaces under varying weather and lighting conditions, with noise from shadows and reflections causing interference, and require robust real-time hazard detection for autonomous driving.

Innovation Solution

A system utilizing a convolutional neural network (CNN) trained with unsupervised techniques to compute gamma directly from image sequences, incorporating sensor motion and epipole information, effectively filters noise and accurately models road surfaces by aligning images based on height and distance, enabling precise detection of reflective areas and moving objects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional road surface detection methods are used, then the system can operate in varying weather conditions, but noise from shadows and reflections interferes with detection accuracy

Engineering Contradiction:
Improvedetection accuracyVSAvoidnoise from shadows and reflections
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent segments the road surface detection problem by separating reflective areas from non-reflective areas through gamma value analysis. By computing gamma values for different image regions and comparing them against thresholds, the system isolates reflective interference (shadows, wet roads) from actual road surface features, enabling accurate detection despite varying weather conditions

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces gamma values as an intermediary metric to bridge the gap between raw image data and road surface detection. Instead of directly analyzing pixel intensities that are affected by shadows and reflections, the system computes gamma values that represent the relationship between height and distance, providing a noise-resistant intermediate representation for accurate road surface modeling

Inventive Principle:
Principle #24Intermediary (Mediator)

2Speed

If real-time hazard detection is implemented, then fast detection of hazards is achieved, but hardware burden increases due to processing difficulties

Engineering Contradiction:
Improvehazard detection speedVSAvoidhardware burden
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent extracts only the essential information needed for road surface detection by computing gamma values from image sequences. Instead of processing entire high-resolution images or using complex sensor arrays, the system extracts gamma values that capture the critical height-distance relationship, significantly reducing computational load while maintaining real-time detection capability

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the detection problem by changing from direct pixel analysis to gamma value analysis. This parameter transformation converts complex image processing into simpler mathematical computations based on the gamma relationship between height and distance, enabling real-time processing with reduced hardware requirements

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3673233B1Vehicle environment modeling with a camera
Publication Date: 2025.09.10 MOBILEYE VISION TECH LTD
  • EP3673233B1 patent drawingFigure 1
  • EP3673233B1 patent drawingFigure 2
  • EP3673233B1 patent drawingFigure 3

AI summary

System and techniques for vehicle environment modeling with a camera are described herein, A time-ordered sequence of images representative of a road surface may be obtained. An image from this sequence is a current image. A data set may then be provided to an artificial neural network (ANN) to produce a three-dimensional structure of a scene. Here, the data set includes a portion of the sequence of images that includes the current image, motion of the sensor from which the images were obtained, and an epipole. The road surface is then modeled using the three-dimensional structure of the scene.