摘要
相似性测量是图像后期特征匹配的关键步骤,本文深入研究了特征向量顺序比对和交叉比对方法的差异,并以形状上下文提取的图像二维直方图特征为基础,分别结合x2距离、推土机距离和扩散距离三种不同的相似性测量方法对多种图像标准库进行了识别验证.
Similarity measurement is a key step in image matching, this paper studies the differences between feature vector sequence alignment and cross comparison method, and based on two-dimensional histogram features which are extracted by shape context, respectively combined with x2 distance, earth mover's distance and the diffusion distance these three different similar measurements, it carries out the identification and verification of a variety of image standard library.
出处
《计算机系统应用》
2016年第8期241-245,共5页
Computer Systems & Applications
基金
广西高校科学技术研究项目(k Y15YB267)
广西高等学校高水平创新团队及卓越学者计划
关键词
推土机距离
扩散距离
相似性测量
形状上下文
web information extraction
knowledge presentation
data intensive web pages
ontology-based keyword library