基于多小波的医学图像阈值分割算法研究

The Medical Image Threshold Value Segmentation Algorithm Study Based on Multi-Wavelet Analysis

图像分割技术是医学图像处理和分析中的关键技术。本文通过构造一个正交多小波,把图像在尺度空间展开,对不同分辨级下的小波系数采用不同阈值进行分割处理。应用自动和交互相结合的分割方法,通过对目标区域图像特征进行分析,交互确定目标区域的大致灰度范围。在综合考虑图像的边缘和区域的基础上,确定精确的阈值,将目标组织从背景中分割出来。通过对计算机断层扫描医学图像进行分割处理和对比实验,得到的图像清晰度明显提高,证明该方法不仅能得到较清晰的感兴趣区域,而且还能获得较清晰的边缘特征。

Recently, image segmentation technology has become the key technology of the medical image processing and analysis. In this paper, constructing a orthogonal multi-wavelet, spreading images in the scale space, using the different threshold value to do the segmentation processing of wavelet coefficients on different resolution grades, applying the segmentation method of combining the automatic and interactive method together, we confirm the roughly gray areas of the target areas interactively through analyzing the regional characteristics of image, on the basis of considering the edge and the region of the image synthetically to ascertain the precise threshold, and separate the target organizations from the background. Through image segmentation processing and comparative experiments of the computer tomography medical images, we prove that this method could get not only comparatively clearer interested areas, but also relatively legibled edge features, and the definition of the image also has been improved obviously.