Scholarlink

<p>– Synthetic aperture radar (SAR) images are used for oil spill detection in ocean. In SAR image, Black spots resembles like oil spills, and differentiation of black spots and oil spills is the major problem. Black spots arises due to different reasons such as low winds, biogenic films, upwelling zones, rain cells and ship wakes. Black spots are look alikes in oil spill detection. Traditional methods such as multisensory fusions, time series monitoring and wind and weather data integration fails to distinguish look alikes more accurately. Low accuracy is due to data limitations and over simplified deep learning (DL) model. To solve above problem MFDCNN OSDC is proposed, which is a Multifusion DL model. The proposed MFDCNN –OSDC model consists of UNet++ architecture, multimodel fusion of googlenet, xception net, sequence net and PKO based hyperparameter tuned ABIGRU algorithm for classification of oil spill. Look alikes texture various due to different reasons as mentioned above and needs multiple algorithms to efficiently differentiate the blackspot. In this paper, different architecture and multimodel eliminates each blackspot, which arises due to different reasons. The proposed MFDCNN – OSDC algorithm classification accuracy is about 97.64% when compared to traditional algorithm.</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

Multimodal Fusion of Deep Convolutional Neural Networks with Optimization Algorithm for Oil Spill Detection and Lookalikes Classification using SAR Images

  • 期刊论文

  • 发表日期：

    2026年1月4日

查看全文
  • S. Sathya,
  • J. Senthil Murugan,
  • K. Selvam,
  • S. Bhuvana

摘要

– Synthetic aperture radar (SAR) images are used for oil spill detection in ocean. In SAR image, Black spots resembles like oil spills, and differentiation of black spots and oil spills is the major problem. Black spots arises due to different reasons such as low winds, biogenic films, upwelling zones, rain cells and ship wakes. Black spots are look alikes in oil spill detection. Traditional methods such as multisensory fusions, time series monitoring and wind and weather data integration fails to distinguish look alikes more accurately. Low accuracy is due to data limitations and over simplified deep learning (DL) model. To solve above problem MFDCNN OSDC is proposed, which is a Multifusion DL model. The proposed MFDCNN –OSDC model consists of UNet++ architecture, multimodel fusion of googlenet, xception net, sequence net and PKO based hyperparameter tuned ABIGRU algorithm for classification of oil spill. Look alikes texture various due to different reasons as mentioned above and needs multiple algorithms to efficiently differentiate the blackspot. In this paper, different architecture and multimodel eliminates each blackspot, which arises due to different reasons. The proposed MFDCNN – OSDC algorithm classification accuracy is about 97.64% when compared to traditional algorithm.