摘要
随着机器视觉的发展,信息检索作为机器学习中的重要课题,各个领域发挥了至关重要的作用,基于图像内容的信息检索技术,通过对图像进行特征提取,然后依靠度量的计算方式来实现最终结果。本文主要研究的是基于极值曲率和相关图图像信息检索方法。对图像的内在曲率特征进行了研究,并融合颜色相关图提高鲁棒性,最后提出了基于极值曲率和相关图检索算法。
With the development of machine vision,information retrieval,as an important subject in machine learning,has played a vital role in various fields.The information retrieval technology based on image content extracts the features of the image,and then relies on the calculation method of measurement to achieve the final result.This article mainly studies image information retrieval methods based on extreme curvatures and correlation graph.The inherent curvature characteristics of the image are studied,and the color correlation map is integrated to improve the robustness.Finally,a retrieval algorithm based on extreme curvatures and correlation map is proposed.
作者
于彬
张皓翔
徐明辉
张高青
王金鹏
YU Bin;ZHANG Haoxiang;XU Minghui;ZHANG Gaoqing;WANG Jinpeng(Xintian Coal Mine of Yonggui Energy Development Co.,Ltd.,Qianxi,Guizhou Province,551514 China;China University of Mining and Technology,Xuzhou,Jiangsu Province,221116 China)
出处
《科技创新导报》
2021年第26期47-49,共3页
Science and Technology Innovation Herald
基金
2021年贵州省科技支撑计划“煤矿井下5G应用研究及示范”重点项目(项目编号:黔科合支撑[2021]重点003号)。
关键词
极值曲率
信息检索
相关图
内在曲率
Extreme curvatures
Information retrieval
Correlation graph
Intrinsic curvature