非均匀纹理复杂表面缺陷多光谱图像显著性融合  

Saliency Fusion of Multispectral Images With Non-uniform Texture and Complex Surface Defects

作　　者：陈海永 杨佳博 陈鹏 刘坤 CHEN Haiyong;YANG Jiabo;CHEN Peng;LIU Kun(School of Artificial Intelligence and Data Science,Hebei University of Technology,Tianjin 300130,China)

机构地区：[1]河北工业大学人工智能与数据科学学院

出　　处：《北京工业大学学报》2020年第1期10-16,共7页Journal of Beijing University of Technology

基　　金：国家自然科学基金资助项目(61403119);河北省自然科学基金资助项目(F2018202078);河北省科技计划资助项目(17211804D);河北省青年拔尖人才资助项目(210003)

摘　　要：为了抑制多晶硅太阳能电池片复杂的背景对表面缺陷可靠提取的影响,并凸显缺陷特征,提出了一种基于经验小波的多光谱显著性融合算法.该算法利用结构纹理分解方法抑制了复杂的背景,通过具有经验信息的二维张量经验小波进行多尺度分解,得到包含缺陷信息的细节层,并对其进行显著性分析,增强权重较大部分图像的对比度,最终得到强缺陷特征图像.对比了平均梯度、边缘强度、清晰度及标准差4个质量评价指标,实验结果表明,所提算法的这些指标比经典融合算法依次提升了0. 01、11. 11、0. 48、2. 33以上,呈现出较好的性能.To suppress the influence of the complex background of polysilicon solar cells on the reliable extraction of surface defects and enhance the defect characteristics,a multi-spectral saliency fusion algorithm based on empirical wavelet was proposed.A structure-texture image decomposition method was used to suppress complex background.Through multi-scale decomposition of two-dimensional tensor empirical wavelet with empirical information,the detail layer containing defect information was obtained.The saliency analysis was carried out to enhance the contrast of the larger weight parts,and the strong defect feature image was finally obtained.The four quality evaluation indexes of average gradient,edge intensity,definition and standard deviation were compared.Results show that these indexes of the proposed algorithm are improved by 0.01,11.11,0.48 and 2.33 than those of the classic fusion algorithm,showing better performance.

关 键 词：结构纹理分解 经验小波变换 多光谱图像融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]