基于小波的彩色图像融合技术

由于图像拍摄时的几何差异和色彩差异,传统融合技术往往出现模糊或假边界.该文介绍一种融合两幅具有重叠区域的彩色图像的实用算法.算法采用交互式选取控制点并指定几何变形类型进行图像配准;在YIQ色彩空间分别融合各分量;在小波变换多尺度分析的基础上,模拟时域加权函数,提出基于小波变换的归一化Mask算法进行图像融合,算法在不同尺度上对各分解系数进行不同程度的融合,体现了时域中不同区域的重要程度.算法实用性强,应用方便.实验表明,融合图像平滑自然,具有很好的视觉效果.

The Color Image Fusion Algorithm Using Wavelet Transform

Because of geometry and color differences,the traditional image fusion method usually obtains faint result or false boundary.This paper introduces a practical image fusion algorithm mosaicing two overlapped color images.First we select control points by manual operation and specify transforming type to implement image matching,then fuse three color components respectively in YIQ space.On the basis of multiscale wavelet analysis,a normalized Mask algorithm is proposed,which fuses the coefficients at each scale to reveal the different importance of image regions.The algorithm is practical and of high stability.Experiment demonstrates that the fusing image is smooth and with good visual effect.