摘要
如何从傅立叶变换的频谱特征获取附加识别信息,构造相应的特征向量,并选择适宜的相似性度量方法,是基于频率域的图像检索的关键.比较分析了基于傅立叶频谱的两组特征,选取楔环特征应用于西北大学可视化研究所开发的医学影像数据库系统中,在医学图像的检索中取得了较好的效果.
How to extract feature from Fourier power spectrum to get additional information and how to select feasible similarity measurement are the key problems in frequency domain based image retrieval. Evaluated and analyzed two groups of features, which are based on the fourier frequency spectrum. The method is used in a medical image database developed by institute of visualization technology in Northwest University, it proves that using the approach opposed in the paper can get better results.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第5期807-809,共3页
Journal of Chinese Computer Systems
基金
国家自然科学基金 (60 2 710 3 2 )资助
关键词
纹理特征
傅立叶变换
楔环特性
基于内容的图像检索
texture feature
fourier transformation
wedge and ring feature
content-based image retrieval
