基于BEMD的疲劳断口图像分割

Fatigue Fracture Image Segmentation Based on BEMD

疲劳断口图像的纹理具有多样性,对多种纹理混合在一起的纹理图像分割是一大难点。基于二维经验模式分解提出了一种新的疲劳断口图像的分割方法。对图像进行BEMD分解后提取本征模式函数bimf中的纹理能量作为特征:首先,对疲劳断口图像进行BEMD分解,得到一系列的本征模式函数bimf和残差函数;然后,采用Laws纹理能量描述方法分别对bimf和残差提取纹理能量作为特征分量;最后,在最近邻准则下对所得特征分量进行分割。通过对比经典的傅立叶方法,断口图像的分割结果表明该方法对疲劳断口图像的分割具有较好的效果。

The texture of fatigue fracture is diverse,so it is difficult to segment an image with a variety of textures.A new image segmentation method for fatigue fracture based on BEMD was presented.The image was decomposed by BEMD and bimf texture energy was extracted as the feature.First,the image was decomposed by BEMD into a series of intrinsic mode function bimf and residual functions.Then,Laws texture energy method was used to extract the bimf and residual texture energy as the feature elements.Finally,the feature was segmented by the nearest neighboring criterion.Compared with the classical Fourier method,this method applies to the segmentation of fatigue fracture images and can obtain satisfactory results.