テクスチャ解析による海面波浪状態のニューラルネット学習

機関リポジトリ 日本農学文献記事索引 Web Site オープンアクセス

書誌事項

タイトル別名
  • Neural Network Learning of Ocean Wave Condition by Texture Analysis

この論文をさがす

抄録

In this study, littoral wave conditions were transformed into image data and used to assess the applicability of the method to constructing a system for automatically digitizing and monitoring wave conditions. An image of ocean wave conditions was treated as a texture and its characteristics were examined as texture feature quantities representing the surface conditions in response to wind. These feature quantities were input to a hierarchical neural network for learning. The network, which had a multilayer structure adapted for the back-propagation algorithm, facilitated the study of the influence of learning conditions on the network structure. In addition, digital sensitivity analysis was performed to identify optimal calculation conditions for presenting an optimal image of the sea surface. Analyses were also performed using spatial color concentration dependence, with texture feature quantities consisting of energy, entropy, correlation, local uniformity, and inertia.

収録刊行物

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ