改进交叉视觉皮质模型的医学图像融合方法

Medical image fusion algorithm based on improved intersecting cortical model

传统的交叉视觉皮质层模型(intersecting cortical model,ICM)在图像边缘检测和图像的分割等领域得到了广泛的应用,但模型中的一些参数需要人工去选取,从而降低了模型应用结果的准确度。为了使ICM中的参数能够自适应选取,对传统的ICM进行改进,提出改进的ICM与非下采样Contourlet变换(non-subsampled Contourlet transform,NSCT)相结合的方法应用于医学图像的融合。实验结果表明,该算法无论从主观性评价还是从六个客观性评价指标均优于其他融合算法,不仅提高了图像的清晰度,而且较大程度地保留了图像的细节信息,具有边缘信息突出、亮度对比度高的优点,取得了满意的效果。

The traditional ICM has been widely used in image edge detection and image segmentation,etc. However,some parameters of the model need to artificial selection,thereby reduces the model's accuracy. Therefore,to improve the traditional ICM that parameters could be adaptively selected,this paper improved ICM combined with NSCT to apply to medical image fusion. The experiment results show that the proposed algorithm in terms of subjective evaluation or six objective evaluation indicators are better than other fusion algorithm. It not only improves the clarity of the image,but also a greater degree preserves the image details,and it has a protruding edge information,high contrast and brightness advantages,and achieves satisfactory results.