融合整体与局部特征信息的图像检索  

Image Retrieval Combining Global and Local Features

作　　者：陈天娥[1] 程载和[1] 

机构地区：[1]无锡职业技术学院计算机技术系,无锡214024

出　　处：《微计算机信息》2010年第10期145-147,共3页Control & Automation

摘　　要：本文提出一种融合整体与局部特征信息的图像检索方法,该方法对改进的HSV色彩空间量化后统计聚类颜色直方图,利用分层分块的思想进行视觉像素的颜色均值和二值化边缘特征的提取,并定义了归一化特征相似度计算公式。检索实验表明,融合后的特征检索,更符合人类视觉感受,检索效果更好。A novel image retrieval method combining global and local features was proposed,this method Statistics clustering color histogram in improved HSV color quantization space,and extracts image feature about color mean of vision pixel and binarization edge based on multi-level and multi-block,also defines Similarity computational formula of normalized feature.Experiments show that the retrieval results obtain from combined features fits more closely with human perception,and the proposed method yields better retrieval performance.

关 键 词：二值化边缘 视觉像素 特征提取 图像检索 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]