摘要
利用四叉树算法将图像分割成若干子块,根据子块的颜色直方图,将提取图像的空间特征与提取图像的颜色特征结合起来,然后运用粒计算理论对提取的图像空间特征进行粒化,得到特征向量,并对其进行归一化。在此基础上,结合相关反馈机制设计基于粒计算的融合多特征的人机交互式图像检索的综合算法,该算法的复杂度低,能够在很大程度上提高图像检索的效率和准确性。
Abstract： The image is divided into several sub-block through the quadtree algorithm. According to the color histogram of sub-block, combined the extraction of image＇s space characteristics and color characteristics. Then use the granular computing theory to granulate the spatial characteristic extracted from image. This way, we can get a feature vector and normalize it. Based on that, associating the relevance feedback mechanism with that to design the man-machine interactive synthesis algorithm of image retrieval that integrate image＇s multiple features based on granular computing. This algorithm has low complexity. It can greatly improve the efficiency and accuracy of image retrieval.
出处
《微计算机信息》
2012年第3期171-172,183,共3页
Control & Automation
基金
基金申请人:宋俊雅
项目名称:基于粒计算的人机交互式图像检索系统的研究
基金颁发部门:教育部(091047616)
关键词
粒计算
空间特征
四叉树
图像分割
图像检索
granular computing
spatial characteristics
quadtree
image segmentation
image retrieval
