一种基于区域分割的红外可见光图像融合方法  

Fusion of Infrared and Visible Light Images Based on Region Segmentation

作　　者：邢永康[1] 葛小青[1] 

机构地区：[1]重庆大学计算机学院,重庆400030

出　　处：《重庆理工大学学报（自然科学）》2011年第1期59-63,106,共6页Journal of Chongqing University of Technology：Natural Science

摘　　要：提出了一种基于提升小波和区域分割的红外可见光图像融合方法。先对红外图像和可见光图像进行提升小波分解,得到各自的高频和低频子图像,利用红外图像的热效应特征显著的特点对红外图像的低频子图像进行检测分割,用分割得到的二值图像来指导低频子图像的融合决策,对于高频子图像,采用区域方差匹配度决策法,最后对融合后的小波系数进行重构得到融合图像。实验结果表明,这种算法能够合理有效地保留红外图像的热目标信息以及可见光图像中丰富的光谱信息,提高了图像的可理解性,客观评价准则与目视效果吻合良好。A new image fusion algorithm of infrared and visible light images based on region segmentation and lifting wavelet transformation was proposed.Firstly,the two source images were respectively decomposed into their high frequency sub-images and low frequency sub-images using the lifting wavelet transform.Secondly,the target region was extracted from the low frequency component of the infrared image,and then fused with background region of the low frequency components using different fusion rules.For the high frequency components,the region variance match measure was introduced to make the fusion decision.Finally,the inverse lifting transform was executed to get the fused image.Experimental results showed that with the proposed method,the fused image exhibited good infrared target features as well as clear visible background,and the intelligibility of the images was improved.The objective evaluation criteria coincided with the visual effects very well.

关 键 词：图像融合 提升小波变换 区域分割 区域方差 匹配度 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]