基于高光谱成像技术的彩色图像标签优化算法研究  

Research on optimization algorithm of color image label based on hyperspectral imaging technology

作　　者：蔡爱平[1] 王涛涛[1] CAI Aiping;WANG Taotao(Jiangxi University of Technology,Nanchang 330098,China)

机构地区：[1]江西科技学院,南昌330098

出　　处：《激光杂志》2020年第12期70-72,共3页Laser Journal

基　　金：江西省教育厅科技项目项目(No.GJJ190993);江西省教育厅科学技术研究项目(No.GJJ180970)。

摘　　要：针对常规算法对于图像标签的分类精度低下的问题,提出基于高光谱成像技术的彩色图像标签优化算法研究。利用相关函数,确定像素点与其相邻点之间的相关程度,去除冗余信息,采用插值表示图像所包含的信息量差异,选择合适的高光谱波段,分析各标签间的语义相关性,通过聚类算法,构建相似度矩阵,实现标签优化,完成基于高光谱成像技术的彩色图像标签优化算法研究。设计仿真实验,分析所提出算法与其他3种常规算法间的差异,结果表明,将高光谱成像技术应用到彩色图像标签优化算法当中,能够将图像标签的分类精度提高15%左右,有效改善了标签分类效果。In order to solve the problem of low classification accuracy of traditional algorithms for image lables,an optimization algorithm for color image lables based on hyperspectral imaging technology is proposed.The correlation function is used to determine the correlation degree between pixels and the neighboring points,and to remove redundant information.The interpolation is used to represent the difference of information contained in the image.The appropriate hyperspectral band is selected to analyze the semantic correlation among lables.The similarity matrix is constructed through clustering algorithm to achieve lables optimization,and the lables optimization algorithm of color image based on hyperspectral imaging technology is completed.The simulation experiment is designed to analyze the difference between the proposed algorithm and the other three conventional algorithms.The results show that the application of hyperspectral imaging technology to the optimization algorithm of color image lables can improve the classification accuracy of image lables by about 15%,and effectively improve the efficiency of lables classification.

关 键 词：高光谱成像技术 图像标签优化 波段选择 相似度矩阵 

分 类 号：TN751[电子电信—电路与系统]