摘要
针对单一的可见光或毫米波辐射图像在隐匿违禁物品探测中的局限性,该文将基于Contourlet变换的图像融合算法引入到可见光与毫米波辐射图像的融合应用中。该算法首先利用Contourlet变换得到待融合源图像的多分辨率结构,然后对低频子带和高频子带分别采用基于区域方差和区域能量的融合规则,并通过接近度函数度量低频子带系数区域特性间的关联程度,最后经过Contourlet逆变换重构融合图像。实验结果表明,该算法可有效综合可见光图像中丰富的细节信息与毫米波辐射图像中显著的目标特征,其融合效果在主观视觉及客观评价指标上均优于传统的融合算法。
In view of the problem that both visible images and millimeter-wave（MMW）radiometric images have limitations in concealed contraband detection,an image fusion algorithm based on Contourlet transform is proposed for the fusion of visible images and MMW radiometric images.The multi-resolution structures of the source images are obtained in Contourlet transform domain.The corresponding fusion rules based on region variance and region energy are established for the fusion of low-pass and high-pass subbands,respectively.In addition,the approach degree function is employed to evaluate the correlation degree of region characteristics of low-pass subbands.Finally,the fused images are acquired through the inverse Contourlet transform.Experimental results show that the proposed algorithm can effectively integrate the abundant details of visible images and the significant target features of MMW radiometric images into the fused results and has better performance than traditional fusion algorithms both in subjective visual effect and objective evaluation criteria.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2012年第1期106-111,共6页
Journal of Nanjing University of Science and Technology
基金
高等学校博士学科点专项科研基金(20093219120018)
南京理工大学科技发展基金资助项目(XKF09071)
南京理工大学自主科研专项计划资助项目(2011YBXM73)
