摘要
目的:为了应用计算机对透射电镜膜蛋白颗粒图像进行颗粒的自动识别与计数处理工作,探索一种改善颗粒图像反差不均匀的校正算法。方法:尝试将同态滤波技术应用于电子束入射-透射模型的膜蛋白颗粒图像处理中。针对颗粒图像明暗反差不均、颗粒形态边界模糊不清、难以辨认的特点,通过实验改造和调整高斯、巴特沃斯、指数高通同态滤波器的参数,并在此基础上重新构造了巴特沃斯带阻同态滤波和指数-巴特沃斯带阻同态滤波方法。通过压缩图像中反映背景的低频信号,展宽表现颗粒细节的高频分量,实现改善图像反差不均的目的。结果:实验所构造的同态滤波器处理颗粒图像的效果各有不同,但均能突出颗粒的细节,减轻背景干扰的影响,实现对图像的非线性校正。结论:经过综合比较5种同态滤波函数对颗粒图像的处理效果,认为指数-巴特沃斯带阻同态滤波方法获得的改善反差不均的效果相对更理想,能更好地满足颗粒的识别与分割的要求。
Objective In order to identify and count the membrane protein particles automatically in TEM image, this paper ex- plores a correction algorithm to improve uneven particle image. Methods This work is about Homomorphic filtering technology applied in electron beam incident - transmission model of membrane protein particle image processing. According to the image characteristics of uneven contrast and illegible particle morphology boundary, parameters of Gauss, Butterworth, exponential high-pass filter were reformed and adjusted through experiments. On this basis, methods of Butterworth band-stop homomorph- ic filtering and index Butterworth stop-band homomorphic filtering were reconstructed. Uneven image contrast was improved by compressing the low frequency signals reflecting the image background and spreading the high frequency components express- ing the details. Results The effects of particle images processed by homomorphic filters reconstructed in the experiment are of some difference. And the filters are able to make details of particles prominent and reduce background interference to achieve nonlinear correction of the image. Conclusion After a comprehensive comparison of five kinds of homomorphic filtering functi- on, it was thought that the method of index - Butterworth stop-band filtering performs better in improving uneven image contra- st. This method can better identify and segment particles.
出处
《中国医学物理学杂志》
CSCD
2015年第2期187-192,共6页
Chinese Journal of Medical Physics
基金
福建省教育厅A类科技项目(JA10149)
关键词
同态滤波
膜蛋白颗粒
图像增强
反差不均校正
homomorphic filtering
membrane protein particles
image enhancement
uneven contrast correction
