摘要
颜色一致向量方法容易丢失图像内容的空间信息,针对该问题,提出一种新的图像检索方法。引入狭长度和标准差特征,设计改进的图像分块策略,给出离心距概念和距离比较公式。提取颜色连通区域的大小、狭长度和颜色特征,以及图像的像素个数、标准差和离心距特征,计算图像间内容的相似度。在Corel图像库上的实验结果表明,该方法能有效利用图像的空间分布信息,检索精度较高。
Against the problem that the method of color coherence vector is easy to lose the spatial information of image content,a new method for image retrieval is proposed in this paper.It introduces the feature of narrow degree and standard deviation,proposes a new method for image block-dividing,the concept of centrifugal distance and a new distance formula.Specific practices are as follows.It extracts the feature of size,narrow degree and color of color connected regions.It extracts the number of pixels,the standard deviation and centrifugal distance of the image.It calculates the content similarity between images and retrieval.Experimental results in Corel image database show that this method can effectively use the spatial distribution information of the image,and accuracy is increased.
出处
《计算机工程》
CAS
CSCD
2012年第15期211-214,共4页
Computer Engineering
基金
国家自然科学基金资助重点项目(40830530)
中国博士后科学基金资助项目(904-180793)
测绘遥感信息工程国家重点实验室专项基金资助项目
关键词
基于内容的图像检索
颜色-空间特征
颜色一致向量
颜色连通区域
直方图
狭长度
Content-based Image Retrieval（CBIR）
color-spatial feature
Color Coherence Vector（CCV）
color connected region
histogram
narrow degree
