测地线主动轮廓模型的改进及其医学应用

The Improvement and Application of the Geodesic Active Contour Model

测地线主动轮廓模型(geodesic active contour)是医学图像分割方法中的最重要内容之一。传统的测地线主动轮廓模型通过高斯平滑的图像预处理使得图像边缘信息模糊甚至消失,以致初始曲线轮廓沿着目标边界逼近的演化过程不稳定,收敛速度慢,分割效果不好。提出基于贝叶斯收缩阈值的小波去噪方法的新测地线模型,由于该去噪方法能在去除噪声的同时,结合噪声与图像信号在频率上的分布特性,增强图像信号细节,从而使能初始轮廓曲线更精确的停止于图像边缘,收敛速度也得到一定程度的提高。

Geodesic active contour model is one of the most important elements of the medical image segmentation methods. The traditional geodesic active contour model with the Gaussian smoothing of image pre-processing makes the image boundary information blurred or even disappear. So the evolution process of approximation that initial curve contour along with the goal border is instable, the rate of convergence is slow, and the segmentation results are not fit well. I will propose a new geodesic active contour model combining wavelet denoising which is based on Bayesian shrinkage threshold, since the denoising method combining the distribution characteristics of noise and image signal in the frequency enhance the image signal details, the test result shows that the convergence of the contour curve of the model will be more accurate, and the speed is also improved to a certain extent.