改进的小波域多聚焦图像融合算法

A Novel Improved Multifocus Image Fusion Algorithm in Wavelet Domain

小波变换和清晰度评价相结合是一种有效的多聚焦图像融合方法。首先,对源图像进行小波分解得到低频子带和高频子带,其次,引入多聚焦图像空域融合中的清晰度评价指标,用改进的梯度能量确定低频部分的清晰像素并进行融合;在高频子带上,先计算每个像素与其邻域像素灰度值之差的绝对值和,然后选取和值较大的像素系数作为高频子带的融合系数。两组实验仿真结果表明,该算法更有效,融合图像更清晰、细节更丰富,更好地继承了源图像的信息。

The combation of wavelet transform and definition evaluation is an effective method for multifocus image fusion.First,wavelet transform was applied to the source images to obtain the low frequency and high frequency.Second,for low frequency,energy of gradient,which is improved in this article and usually used in spatial domain to determine the clear pixel or region,is introduced as a fusion regulation.For high frequency,a fusion method based on the sum of the absolute value of the difference between the center pixel and its neighbors is proposed.Two pairs of multi-focus images are selected to simulate this scheme.Simulation experiment results demonstrate the effectiveness of the proposed approach,which enriches the detailed information of the fused image and makes it more clearly,what’s more,is more capable to retain and pick up source image information.