クラスター特性関数を用いた主観的評価差に対応する画像検索法

書誌事項

タイトル別名
  • An Image Retrieval Method Using Cluster Characteristic Function for the Subjective Evaluation Difference
  • クラスター トクセイ カンスウ オ モチイタ シュカンテキ ヒョウカサ ニ タイオウ スル メンゾウ ケンサクホウ

この論文をさがす

抄録

In present paper, a new retrieval technique using kansei information on image-database is proposed.<BR>The development of an image retrieval system with high image hit-rate using kansei information has been difficult. The main cause has been that user's evaluation score with a high subjectivity to his target image has differed from an objective image attribute score ( we name this difference ’subjective evaluation difference’) .<BR>In proposal technique the images first are classified into several image-clusters that are constructed with similar images. Next a cluster characteristic function, which shows the feature of each image-cluster, is calculated. The cluster characteristic function distinguishes user's evaluation score with the subjective evaluation difference from appropriate user's input data, and applies a proper retrieval method to it.<BR>The apparel pattern retrieval system to which our proposal technique was mounted was evaluated in order to confirm its utility. As a result, the utility of the technique was proven.

収録刊行物

  • 図学研究

    図学研究 33 (3), 13-19, 1999

    JAPAN SOCIETY FOR GRAPHIC SCIENCE

参考文献 (13)*注記

もっと見る

詳細情報 詳細情報について

問題の指摘

ページトップへ