一种新颖的低层视觉特征到高层语义的映射方法  

Novel Method of Mapping Semantic Gap

作　　者：莫浩澜[1] 石跃祥[1] 文立[1] 

机构地区：[1]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《计算机应用研究》2007年第6期173-175,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60234030);湖南省自然科学基金重点资助项目(02JJY2091)

摘　　要：通过在低层视觉特征上提取图像的局部颜色和纹理特征,再利用模糊神经网络建立低层视觉特征和高层语义特征之间的映射,实现了图像属性信息的有效传递和可靠检索。最后,为了提高检索效率,把相关反馈引进到图像检索系统中。实验证明,该方法取得了较好的检索查全率和准确率。Firstly color and the texture features were extracted. Secondly used the fuzzy nerve network to map low-features and the high-level semantic features. So this method had realized information of image＇ s attribution effective transmission and reliable retrieval. Finally, in the image retrieval system, introduced correlation feedback to enhance the retrieval efficiency. The experiment proves that it has obtained the high accuracy.

关 键 词：语义图像检索 局部颜色和纹理特征 模糊神经网络 相关反馈 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]