摘要
针对传统颜色直方图无法反映像素的空间信息和图像的局部细节,提出一种整合整体和局部颜色特征的图像检索技术,先将图像分割为环形区域,获得图像的局部颜色特征。把赋以不同权值的整体颜色直方图和局部颜色之间的相对距离累加和定义为图像之间的相似度,以反映人们对图像的判定过程。实验表明,该方法反映了图像颜色的空间分布及局部细节,在查准率和查全率上都要优于基于传统的颜色直方图的检索方法。
For solving these problems that the conventional color histogram can＇t reflect the position of the point and the local detail of the image, a new retrieval method which integrates the local character containing color and position into the whole character is presented in this paper. Firstly, the image is divided into annular region by annular division, and then calculates the local color character. And then the similarity of the image is defined as the combination of the similarity of the integer and the similarity of the annular blocks to reflect the recognition process by people. The experimental result on real-world images indicates that the proposed approach reflecting the spatial information of colors and the local detail of the image is better than the conventional retrieval method based on color histogram in an accurate and complete survey.
出处
《微型机与应用》
2009年第19期22-24,共3页
Microcomputer & Its Applications
关键词
颜色直方图
环形分割
图像检索
局部颜色特征
color histogram
annular division
image retrieval
local color character
