Forward Collision Warning Using Optical Flow Model Fitting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current driver assistance systems for forward collision warning rely on object recognition, which can be inefficient and prone to false alarms, especially in complex scenes with changing lighting conditions and multiple objects, and do not effectively utilize optical flow to determine time-to-collision without prior object detection.

Innovation Solution

A method and system that use a camera-mounted in a vehicle to track optical flow of image points between frames, fit them to road surface and vertical object models, and determine time-to-collision without prior object recognition, issuing a warning based on relative scale and time intervals, thereby avoiding object recognition and reducing false alarms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If object recognition is used to determine collision risk, then the system can identify specific objects (pedestrians, vehicles), but the system becomes prone to false alarms and inefficiency in complex scenes

Engineering Contradiction:
Improvecollision warning reliabilityVSAvoidobject recognition complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the essential collision risk information (time-to-collision, relative scale) from the visual scene without performing full object recognition. Instead of identifying what objects are present, the system extracts geometric and motion parameters directly from image frames to determine collision risk, eliminating the complexity and false alarms associated with object classification.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameters used for collision assessment from object-based features to optical flow-based parameters (time-to-collision, relative scale). By monitoring changes in these parameters over time rather than relying on object recognition confidence, the system achieves more reliable collision warnings in complex scenes.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If optical flow tracking is performed without prior object detection, then the system avoids object recognition complexity, but the system must process all image points increasing computational load

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the image processing task by focusing optical flow tracking only on relevant regions (image patches) rather than processing the entire image. This segmentation approach maintains the benefits of avoiding object recognition while reducing computational load by limiting processing to specific areas of interest in the visual field.

Inventive Principle:
Principle #1Segmentation

3Loss of information

If the system monitors all image points for optical flow, then comprehensive motion information is obtained, but the measurement and processing becomes more difficult

Engineering Contradiction:
Improvemotion information completenessVSAvoidoptical flow measurement difficulty
Core Design Contradiction:
Loss of informationVSDifficulty of detecting and measuring

Solution Approach 1:

The patent divides the image into multiple patches and processes optical flow independently for each patch. This segmentation reduces the difficulty of measuring optical flow by limiting the number of points processed simultaneously while maintaining comprehensive motion information coverage across the entire visual scene through aggregated patch results.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2993654B1Method and system for forward collision warning
Publication Date: 2017.05.03 MOBILEYE VISION TECH LTD
  • EP2993654B1 patent drawingFigure 1a~1b
  • EP2993654B1 patent drawingFigure 2a~2b
  • EP2993654B1 patent drawingFigure 3a~3d

AI summary

A method for providing a forward collision warning using a camera mountable in a motor vehicle acquires multiple image frames at known time intervals. A patch may be selected in at least one of the image frames. Optical flow may be tracked between the image frames of multiple image points of the patch. The image points may be fit to at least one model. Based on the fit of the image points to the at least one model, a time-to-collision (TTC) may be determined if a collision is expected. The image points may be fit to a road surface model and a portion of the image points is modeled to be imaged from a road surface. A collision is not determined expected based on the fit of the image points to the road surface model. The at least one model may also include a mixed model where a first portion of the image points may be modeled to be imaged from a road surface and a second portion of the image points may be modeled to be imaged from a substantially vertical object. The image points may be fit to a vertical surface model, where a portion of the image points may be modeled to be imaged from a vertical object. The TTC may be determined based on the fit of the image points to the vertical surface model.