【摘要】 针对双式红外(中红外和长红外)提出了一种基于小波的分维融合算法.通过小波变换,分别对中波红外和长波红外的两幅图像进行小波分解,在小波变换域低频部分对小波系数用能量融合,高频部分采用分形分维进行融合,得到变换域中各个频带的融合图像,然后反变换进行重构,获取融合后的图像.实验结果表明,根据用分维进行数据融合的方法来确定两幅不同原图像在融合图像中所占的信息比例,可以有效地保留两幅原图像的边缘和纹理特征,避免融合图像平均化而出现的模糊现象,融合后的图像综合了两幅原图像的不同特征,使得处理后的图像更容易识别.因此,提出的双波段红外图像的小波分维融合算法是有效的,并且可以取得较好的效果.更多还原

【Abstract】 The paper put forward a fractal dimension algorithm based on wavelet transformation according to property of two-band (mid-wave and long-wave) infrared image, decomposing mid-wave and long-wave infrared images based on wavelet transformation respectively. Computing the fractal dimension of wavelet coefficients at high frequency in wavelet transform domain and fuse horizontal, vertical and diagonal coefficients of the two images separately took fractal dimension as weight before reconstruction. Then a fusion image was given. The experiment results show that the algorithm of fractal dimension to fuse two images from different source can confirm their message proportion accounted for among merging image and keep the feature of edge and texture so as to avoid the fuzzy phenomenon caused by fusing them equally. It is easier to recognize the target in fused image because it synthesizes different characters of the two images. Thus, the algorithm of wavelet fractal dimension on two-band infrared fusion presented in the paper is effective and can make better result.更多还原