基于图像敏感区域纹理特征的图像检索  

Image retrieval based on texture features at sensitive area

作　　者：朱梅[1] 李章维[2] 

机构地区：[1]浙江科技学院信息与电子工程学院,杭州310023 [2]浙江工业大学信息工程学院,杭州310023

出　　处：《浙江科技学院学报》2010年第1期31-35,共5页Journal of Zhejiang University of Science and Technology

基　　金：浙江省科技计划项目(2008C21080)

摘　　要：提出一种基于敏感区域动态特征的彩色图像识别方法,模拟人眼视觉特性,在小波变换域内利用K均值聚类提取敏感区域图像,提取敏感区域的动态纹理特征来表示图像内容,并以此计算图像间的相似度、对图像进行分类和检索。仿真实验结果表明,该方法能够较准确地查找出目标图像,并且明显地提高了检索精度。The paper proposes a method of the color image retrieval based on the dynamic characteristics of the sensitive areas. In conjunction with the human visual system characteristics, the scheme uses k-means clustering in the wavelet transform domain to extract the image from the sensitive area. The dynamic texture features being obtained from sensitive area represent the image content, which are employed to calculate the similarities between images and to perform image categories and retrieval. Simulation results show that the images can be precisely found out by this method, thus the retrieval accuracy will be significantly improved.

关 键 词：图像检索 敏感区域 离散小波变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]