分水岭算法在CT图像分割中的应用

The Practical Application of Watershed Algorithm in CT Image

目的:分水岭算法在图像分割领域得到了广泛的应用,但单独使用分水岭进行图像分割因为对噪声的抑制能力弱以及对大多数图像易产生过分割现象而变得困难。本文针对分水岭算法存在的过分割问题,提出了一种改进的分水岭算法应用于CT图像,能有效的抑制过分割现象。方法:首先对输入图像进行高斯滤波处理,然后通过Sobel算子求图像的梯度幅值,再求出多尺度灰度图,最后进行阈值分割和多尺度变换而达到对图像进行分割的目的,并将其转化成伪彩色图像显示来优化分割结果,在有效处理过分割问题的同时让图像分割后的效果更加明显。结果:仿真结果表明,与传统的分水岭分割算法比较,缓解了过分割问题,得到的分割效果要好很多。结论:本文实验可以有效地将传统的分水岭算法加以改进,将之应用于医学CT图像分割中,从而使图像各个不同的组织轮廓均得到了很好的区分,减少了图像的过分割点数,使图像的各个区域更易判断。

Objective： The watershed algorithm has been widely used in the field of image segmentation, because the capability of restraining noise is weak and the over segmentation phenomenon of most images which make it is difficult to use watershed image segmentation only. According to the over segmentation of watershed algorithm, we propose an improved watershed algorithm, which can effectively restrain the over segmentation phenomenon. Methods： First, Gauss filter processing the input image, and then uses the Sobel operator of image for gradient magnitude, then calculating the multiscale image segmentation, Finally, to achieve the purpose of the segmentation of image threshold segmentation and multi scale transform, and converting it to the pseudo color image to optimize the segmentation results , at the same time, the effective treatment of the over segmentation problem can make the image segmentation＇s results more apparently. Results： The simulation results show that, compared with the traditional watershed segmentation algorithm, alleviate the over segmentation＇s problem, which obtained better segmentation results. Conclusions： In this paper, the traditional watershed algorithm is effectively improved, and applied in the segmentation of medical image of CT, so that the？different tissue contour of the image is well differentiated. It can reduce the over segmentation points of the image, so that each region of the image is easier to judge.