河南科技大学毕业设计(论文) 基于小波变换的医学图像分割系统的研究
摘 要
图像分割是图像处理中一项重要的技术,其目的是把图像分成各具特性的区域并提取出感兴趣的部分。其结果为图像分析和理解提供依据。图像分割是一个经典问题,从发展至今仍没有找到一个通用的方法。
本文通过对医学图像分割技术的相关背景、原理及算法进行研究,采用MATLAB来编程开发一个基于灰度直方图与小波变换的医学图像分割系统。本文在预处理部分采用对图像进行平滑、灰度调整等操作。其中,平滑采用中值滤波算法。第二个部分是基于直方图的小波变换。首先,采用的是基于小波基sym8的滤波算法对直方图进行滤波,能较好的减小噪声对直方图波形的影响。然后,利用小波变换的多尺度特性,对原图灰度直方图采用基于小波基db4的五层小波分解,重构第五层近似分量。第三个部分是多阈值分割。对第二部分中直方图的近似分量采用动态阈值检测,利用检测出的多阈值对原图像进行图像分割。
本文把直方图与小波变换方法结合起来,将小波变换应用于灰度直方图后进行图像的分割,在高尺度上选择分割的阈值,这样使得阈值的选取更加合理。实验结果表明,该方法具有较好的分割效果。
关键词:医学图像分割,预处理,灰度直方图,小波变换,阈值分割
I
河南科技大学毕业设计(论文)
WAVELET-BASED MEDICAL IMAGE SEGMENTATION
SYSTEM
ABSTRACT
Image segmentation is an important image processing technology. Its purpose is to image into regions with different characteristics and extract the interesting part. The results provide the basis for image analysis and understanding. Image segmentation is a classical problem, from the development has yet to find a common approach.
Based on the medical image segmentation technology background, theory and algorithm research, use of MATLAB to develop a program based on histogram and wavelet transform for medical image segmentation system. In this paper, the image preprocessing part is smooth, gray scale adjustment operation. Among them, the smoothing by median filtering algorithm. The second part is based on the histogram of the wavelet transform. First of all, using the wavelet-based filtering algorithm sym8 histogram filtering can be better to reduce the impact of noise on the histogram waveform. Then, using the wavelet transform multi-scale features of the original image histogram using a five-story based on wavelet db4 wavelet decomposition and reconstruction, similar to the fifth floor component. The third part is a multi-thresholding. The second part of the histogram of the approximate weight by dynamic threshold detection, the use of multi-threshold detection of the original image for image segmentation.
This histogram and wavelet transform method to combine the wavelet transform applied to images after histogram segmentation, select the partition in the high-scale threshold, which makes the selection more reasonable threshold. Experimental results show that the method has better segmentation.
KEY WORDS: Medical image segmentation, preprocessing, histogram, wavelet
II
河南科技大学毕业设计(论文) transform, thresholding
III
河南科技大学毕业设计(论文) 目 录
第一章 绪 论 ........................................... 1
§1.1 图像分割技术的现状和发展情况 .................. 1 §1.2 图像分割主要研究方法 .......................... 2 §1.3 论文的内容与结构安排 .......................... 6 第二章 图像增强技术 .................................... 8
§2.1 图像平滑 ...................................... 8
§2.1.1中值滤波原理 ............................. 8 §2.1.2中值滤波主要特性 ......................... 9 §2.1.3平滑效果分析 ............................ 10 §2.2 灰度调整 ..................................... 10
§2.2.1灰度调整原理 ............................ 10 §2.2.2灰度调整效果分析 ........................ 10 §2.3 小结 ......................................... 11 第三章 基于直方图的小波变换应用 ....................... 12
§3.1 直方图的小波滤波 ............................. 12 §3.2 直方图的小波变换 ............................. 13
§3.2.1小波变换理论 ............................ 15 §3.2.2小波算法 ................................ 16 §3.2.3尺度选择 ................................ 16 §3.3小结 ......................................... 16 第四章 图像阈值分割技术 ............................... 17
§4.1阈值分割技术简述 ............................. 17 §4.2阈值分割方法 ................................. 17
§4.2.1 二值化 ................................. 17 §4.2.2 多阈值分割 ............................. 19 §4.3小结 ......................................... 24 第五章 医学图像分割系统设计 ........................... 25
§5.1 设计思路 ..................................... 25 §5.2 GUI界面设计 ................................. 25
IV
河南科技大学毕业设计(论文) §5.3实验仿真结果及评价 ........................... 26 结论 .................................................. 27 参考文献 .............................................. 28 致 谢 ................................................. 31 附 录 ................................................. 32
V