Scholarlink

<p>Detecting fractures in landslide-prone areas is important, as they often serve as early indicators of slope instability. This study reports on automatic fracture detection applied to the Vaches Noires cliffs in Normandy, France. While artificial intelligence and high-performance processors have enabled widespread use of automatic fracture detection in fields such as road engineering, their application in geological studies remains limited. To address this gap, the study demonstrates that training a convolutional neural network on aerial imagery is an effective and promising strategy for improving the mapping of fractured zones. In addition, incorporating temperature data from thermal infrared imagery—alongside visible images—through an early data fusion approach proves to be a key factor in reducing detection errors caused by complex image textures and sensitivity to weather variations. Comparison of results with and without data fusion reveals that this combined approach improves detection accuracy by 5<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> and raises the F1-score by 2<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation>, notably reducing false positives by leveraging the additional thermal information. These findings highlight the potential of machine learning techniques and the added value of bimodal imaging in fracture detection, while also pointing to broader implications for the development of early warning systems and the strengthening of hazard management strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A new automatic approach based on visible-thermal infrared data fusion to detect geological fracturation: the case of the Vaches Noires badland, Normandy, France

  • 期刊论文

  • 发表日期：

    2025年11月15日

查看全文
  • Zoé Lambert,
  • Raphael Antoine,
  • Clément Mauget,
  • Vincent Guilbert,
  • Bruno Beaucamp,
  • Laetitia Aubin,
  • Robert Davidson,
  • Stéphane Costa,
  • Olivier Maquaire,
  • Cyrille Fauchard

摘要

Detecting fractures in landslide-prone areas is important, as they often serve as early indicators of slope instability. This study reports on automatic fracture detection applied to the Vaches Noires cliffs in Normandy, France. While artificial intelligence and high-performance processors have enabled widespread use of automatic fracture detection in fields such as road engineering, their application in geological studies remains limited. To address this gap, the study demonstrates that training a convolutional neural network on aerial imagery is an effective and promising strategy for improving the mapping of fractured zones. In addition, incorporating temperature data from thermal infrared imagery—alongside visible images—through an early data fusion approach proves to be a key factor in reducing detection errors caused by complex image textures and sensitivity to weather variations. Comparison of results with and without data fusion reveals that this combined approach improves detection accuracy by 5 \(\%\) % and raises the F1-score by 2 \(\%\) % , notably reducing false positives by leveraging the additional thermal information. These findings highlight the potential of machine learning techniques and the added value of bimodal imaging in fracture detection, while also pointing to broader implications for the development of early warning systems and the strengthening of hazard management strategies.