摘要
传统的颜色直方图只关注颜色的统计信息,丢失了颜色的空间分布信息,检索效果往往不尽人意。该文提出一种融合粒计算的颜色直方图图像检索方法。它融合粒计算多粒度、多层次、多视角的思想,避开了颜色直方图空间位置信息丢失的弊端,同时对颜色直方图的量化区间大小进行分层粒化,并采用直方图相交距离的方法进行图像的相似性度量,实现了图像检索效率和精度的平衡,在不失精度的前提下,保证了检索的效率。实验结果表明,该文算法具有较好的查准率和查全率。
The retrieval result of traditional color histogram image is not satisfactory,they only consider the statistical information about color and forget its spatial distribution information.In this paper,a color histogram image retrieval method based on granular computing was put porward,which use more size,multi-level and perspectives thought of granular computing and keep the spa tial distribution information of color histogram image,and then granulating the quantizing interval of color histogram image,pro ceeding similarity matching according to histogram intersection distance method.To keep the retrieval efficiency in the premise of precision,the experiment result shows that the algorithm has better precision and recall rate.
出处
《电脑知识与技术（过刊）》
2013年第3X期1903-1905,共3页
Computer Knowledge and Technology
关键词
图像检索
颜色直方图
粒计算
image retrieval
color histogram image
granular computing
