基于能量、梯度与方差的多聚焦图像融合

Multi-focus image fusion based on energy, gradient and variance

摘要:

大多数图像融合算法只就图像的某一个特征进行融合，容易造成其他特征信息损失。针对此问题，提出了一种利用哈尔小波变换的特性，考虑多种区域特征进行融合的策略，将图像进行哈尔小波变换后，根据图像的低频部分集中图像大部份能量的特征，采用梯度和能量相结合，根据图像高频部分反映图像细节的特征，采用区域方差与变换系数相结合的方法进行融合，最后经哈尔小波逆变换得到融合结果。通过对多组多聚焦图像进行融合实验，采用均值、方差、熵和平均梯度4种客观评价指标来评价融合图像效果，结果表明该方法能很好地保留图像信息，融合效果好。

Abstract:

Most of methods for image fusion are based on only one characteristic of image, which is liable to cause some information loss about other characteristics. In order to solve this problem, a new method for image fusion based on Haar wavelet and multi-characteristics is proposed. After Haar wavelet transformation, the low-frequency part is fused by gradient and energy since it contains most energy of the image. And the high-frequency part is fused by variance and transforming coefficient because it reflects the details of image. Finally, the inverse transform is performed to obtain fused image. After multi-group images fusion by this method, the performance of image fusion is evaluated by four criteria including average value, variance , entropy and average gradient. It is indicated that this method can keep the image information very well, and the fusion effect is good.