Parking Line Tracking for Reliable Surround-View Frame Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing parking assist devices face challenges in accurately detecting parking area lines and frames due to factors like shadows, obstacles, and varying detection timing, leading to false detections and missed frames.

Innovation Solution

An image processor with a line extraction module, tracking determination module, comparison module, and parking area line determination module that extracts lines from images, tracks and compares them to past data to ensure similarity and coincidence, thereby accurately identifying parking area lines and setting frames.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If parking area lines are detected dynamically based on vehicle movement, then the detection adapts to changing conditions, but detection accuracy deteriorates due to shadows, obstacles, and varying detection timing

Engineering Contradiction:
Improvedetection adaptabilityVSAvoidline detection accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system performs preliminary line extraction and tracking before final parking frame determination. By extracting lines in advance and maintaining tracking information across multiple frames, the system prepares detection data proactively, allowing it to compensate for temporary detection failures caused by shadows or obstacles without sacrificing accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback by comparing currently extracted lines with previously tracked lines to determine similarity and coincidence. This feedback mechanism allows the system to verify detection accuracy continuously and correct for temporary detection errors caused by environmental factors like shadows and obstacles

Inventive Principle:
Principle #23Feedback

2Reliability

If angle parking frame detection is based on continuity of detected frames, then consistent angle parking areas can be identified, but false detections occur when isolated parking frames are mistakenly identified as angle parking frames

Engineering Contradiction:
Improveangle parking frame detection reliabilityVSAvoidparking frame type identification accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The system applies feedback by comparing current line detections with historical tracking data to verify the consistency of detected parking frames. By checking whether detected lines coincide with previously tracked lines, the system can reliably distinguish true angle parking frames from false detections, ensuring accurate parking frame type identification

Inventive Principle:
Principle #23Feedback

3Productivity

If detection is performed at predetermined timing, then processing efficiency is maintained, but lines may be missed due to shadows, obstacles, or unfavorable detection conditions

Engineering Contradiction:
Improvedetection processing efficiencyVSAvoidline detection completeness
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary line extraction at predetermined timing intervals while maintaining tracking information. By extracting lines in advance and storing tracking data, the system ensures efficient processing while preparing compensation data for cases where detection may be incomplete due to shadows or obstacles

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system maintains continuous tracking of extracted lines across multiple frames. This continuity ensures that even if lines are temporarily missed due to shadows or obstacles at specific timing moments, the tracking information persists and allows for accurate detection when conditions improve

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP3872697B1Image processor and image processing method
Publication Date: 2023.08.30 FAURECIA CLARION ELECTRONICS CO LTD
  • EP3872697B1 patent drawingFigure 1
  • EP3872697B1 patent drawingFigure 2
  • EP3872697B1 patent drawingFigure 3

AI summary

An image processor (100) includes a line extraction portion (111) configured to extract a line from an image (G) generated from a surrounding image of a vehicle (V) captured by an imaging device (20) at a predetermined timing, a tracking determination portion (113) configured to determine whether the extracted line fulfills a predetermined condition or not, a comparing portion (114) configured to compare the line that is determined as fulfilling the predetermined condition with a line that is extracted by the line extraction portion (111) from an image (G) generated from a surrounding image of the vehicle (V) captured by the imaging device (20) at a predetermined timing in past to obtain degrees of similarity and coincidence thereof, a parking area line determination portion (115) configured to determine the line as a parking area line when the obtained degrees of similarity and coincidence are equal to or more than a given value, and a parking frame setting portion (116) configured to set a parking frame by using the line that is determined as the parking area line.