峰值皮层模型耦合Weber局部描述符的图像融合算法

The Image Fusion Algorithm Based on Peak Cortical Model Coupled with WLD

为了解决当前医学图像融合(Medical Image Fusion, MIF)方法易产生伪影,丢失部分图像细节,以及对比度较低的问题,本文提出了一种新的峰值皮层模型(Spiking Cortical Model, SCM)耦合WLD (Weber Local Descriptor)的图像融合算法.首先,通过SCM对源图像进行分解,获得不同的二进制脉冲图像;其次,利用输出脉冲图像生成点火映射图像,并构建了SCM脉冲输出点火数量的融合准则;然后,结合SCM脉冲输出的信息熵与点火映射图像的Weber局部描述这二者的相似性来计算融合权重,完成图像融合.通过实验表明:与当前常用的MIF算法相比,本文所提算法具有更好的视觉效果,其融合图像质量与对比度更高,同时,在客观评价标准IE,MI,AG,SSIM方面也具有更大的优势,有效地保持了源图像的有效信息.

Current medical image fusion (MIF) methods are prone to produce artifacts, lose some image details and have low contrast. In order to overcome these defects, a new image fusion algorithm based on spiking cortical model and Weber local descriptor was designed in this paper. Firstly, different binary pulse images were obtained by using the SCM to decompose the source image. Secondly, a fusion rule of SCM pulse output ignition number was constructed by using the output ignition image to generate an ignition mapping image. Then the image fusion was completed based on the similarity between the entropy of the SCM pulse and Weber local descriptor of the ignition mapping image to calculate the fusion weights. Experiment results showed that compared with the MIF methods which are now widely used, this algorithm has better visual effects and higher fused image quality and contrast. In addition, it has bigger advantage in the objective evaluation criteria as IE, MI,AG and SSIM, for it can effectively preserve the valid information of the source image.