结合颜色空间变换与GrabCut的超声相控阵图像分割

Image segmentation of ultrasonic phased array combined color space transformation with GrabCut

为实现工业故障诊断的自动化,应先解决超声相控阵无损检测图像的目标分割问题.为此提出一种结合颜色空间变换与GrabCut算法的超声相控阵图像分割方法.该方法改进了传统的GrabCut算法,通过自适应直方图均衡化对超声相控阵图像进行增强,然后结合颜色空间变换和GrabCut算法对目标进行交互式图像分割得到图像目标分割结果.实验表明,与传统GrabCut算法相比,本文所提方法能够更加精确分割出图像中目标,并能克服背景噪声,保留目标图像细节.

In order to realize the automation of industrial fault diagnosis,the target segmentation problem of ultrasonic phased array nondestructive detection images should be solved first.To this end,an ultrasonic phased array image segmentation method combining color space transformation with GrabCut algorithm is proposed.This method improves the traditional GrabCut algorithm,enhances the ultrasound phased array image by adaptive histogram equalization,then combines the color space transformation and GrabCut algorithm to segment the target image for obtaining the image segmentation result.Experiments show that compared with the traditional GrabCut algorithm,the proposed method can segment the target more accurately,overcome the background noise,and preserve the target image details.