摘要
针对基于区域的图像检索技术中在构造显著区域时忽略人眼视觉的影响而造成的检索率偏低问题,提出了一种新的图像检索算法。首先,引入人眼的视觉特性,提取图像的一些有意义的点特征,构造显著区域,在此基础上,定义图像的特征描述符,以描述符为线索,将图像的形状特征和空间颜色分布特征有机进行结合,不仅克服了基于目标区域检索时的缺点,而且降低了传统目标区域提取算法的复杂度。实验结果表明,该算法具有较高的检索效率。
An image retrieval based on object regions is proposed in this paper to solve the ques%ion of lacking of human visual information in constructing the object regions. Firstly, a robust and self- adaptive extraction algorithm of salient points is introduced. Based on which, According to the distribution of salient points, the object regions are achieved. Then, the feature of object is proposed for image retrieval connecting the shape with color information. The algorithm avoids the defects of object regions when using in the image retrieval and reduces the compute complex of traditional extraction algorithm. The experimental results demonstrate the proposed method has better performance than other algorithms and has better accuracy.
出处
《中国科技信息》
2013年第18期100-101,共2页
China Science and Technology Information
基金
河南理工大学博士基金(B2009-91)
关键词
基于内容的图像检索
显著区域
视觉特性
content-based image retrieval(CBR)
object reRion: human visual