見張り支援ニューラルネットワークに関する基礎研究 : 夜間航海時の相手船針路の推定

概要

論文の詳細を見る
The authors had developed two prototype hierarchical neural networks which had to be used for a target ship's course estimation in daytime navigation. This paper is concerned with a picture recognition neural network system for a lookout support in nighttime navigation. The neural network topology is three-layered network, which consists of 182 units for input layer, 273 units for hidden layer and 21 units for output layer. The neural network input data is navigation lamps location's binary address which is calculated a result of lapped navigational environment picture over a recognition map. Characteristics of the network output is analyzed and it is investigated that a possibility of building the lookout support picture recognition neural network system in nighttime.
社団法人日本航海学会の論文
1995-03-25