改进的二维经验模式分解方法

为了解决图像处理中应用到的传统二维经验模式分解算法存在边界效应和过度分解的问题,提出了一种改进的二维经验模式分解算法.该算法首先对原始图像的边界进行延拓处理,在图像信号的边界处增加一部分数据;然后对处理后的图像使用传统的二维经验模式分解方法进行图像筛分,筛分截止后对每个筛分过度的内在模式函数增加一个对应的补偿量.应用改进的二维经验模式分解算法对图像进行了处理,计算了处理后得到的重构图与原图的标准差.实验结果表明,改进的二维经验模式分解算法消除了边界效应,也解决了图像分解过度的问题.重构图与原图像的标准差很小,证明了重构图与原图的图像灰度波动很小即图像吻合得很好,并且由于处理边界问题时附加的图像信息并不多乃至计算量小,使处理简单易行,论证了改进的二维经验模式分解算法在图像处理中的可行性.

Decomposition method of improved two-dimensional empirical mode

To solve boundary effects and excessive decomposition of traditional two-dimensional empirical mode decomposition algorithm in image processing,an improved two-dimensional empirical mode decomposition algorithm was proposed.Firstly,the boundary of the original image was extended to increase the data in the boundary part of the image signal.Secondly,processed image was screened using decomposition algorithm of traditional two-dimensional empirical mode;Finally,a corresponding compensation was added for each intrinsic mode function which was excessively screened.The improved two-dimensional empirical mode decomposition algorithm was applied to processing image,and standard deviation between reconstruction image and original image was calculated.The results show that the improved two-dimensional empirical mode decomposition algorithm eliminates boundary effects,solves the problem of image＇s excessive decomposition.The standard deviation between reconstruction image and original image is small,so the conclusion is proved that the fluctuation of image gray-scale between reconstruction image and original image is small and two images are in agreement with each other.So that additional image information for dealing with boundary issue is not much,computation is small and treatment is simple.The improved two-dimensional empirical mode decomposition algorithm in image processing is feasible.