織物イメージにおける多様解画像の検討 : ニューラルネットワークモデルによる画像生成

概要

論文の詳細を見る
The purpose of this study was to generate various fabric illustrations from one of the visual impressions of fabrics by using a multi-layer neural network model that employed the deviation values of the impressions as input units. Firstly, sensory evaluation was carried out to investigate the visual impressions of fabrics illustrations. Secondly, the relations among the visual impressions, the deviation values and the parameters generating the fabric illustrations were formulated using a multi-layer neural network model. Then more fabric illustrations to one of the visual impressions were predicted through the neural network model considering the deviation of the visual impressions of fabrics. As a resul it was suggested that the deviation of visual impressions could be reflected in the predicted fabric illustrations through the neural network model.
日本デザイン学会の論文
2000-10-16