一种改进Canny算法在肺炎图像边缘检测中的研究与应用  

Research and Application of an Improved Canny Algorithm in Pneumonia Image Edge Detection

作　　者：王煜 朱硕 吕宗奎 WANG Yu;ZHU Shuo;LYU Zongkui(School of Electronic Information Engineering,Nanjing University of Information Engineering,Nanjing 210044;School of Electronic Information Engineering,Wuxi University,Wuxi 214105)

机构地区：[1]南京信息工程大学电子与信息工程学院,南京210044 [2]无锡学院电子信息工程学院,无锡214105

出　　处：《计算机与数字工程》2025年第3期851-856,共6页Computer & Digital Engineering

基　　金：江苏省双创人才计划双创博士(编号:JSSCBS20210871);无锡学院人才启动资金(编号:2021r014)资助。

摘　　要：新冠肺炎患者在确诊及治疗过程中,通过医生对肺部X光的病灶情况进行判断,由医生根据专业知识与经验给出诊断结果,为了保证医生在诊断过程当中给出正确、可靠的诊断结果,提供边缘清晰、信噪比高的肺部影像图片是新冠肺炎患者确诊及治疗的关键步骤。目前,利用Canny边缘检测算法与深度学习结合,实现新冠肺炎x光图像检测的方法,虽然检测效果好,但是泛化能力差、诊断效率低。针对传统Canny算法中使用高斯滤波不能消除椒盐噪声,且人工选取高低阈值自适应性不足的问题,提出一种改进的Canny算法,首先使用对比度受限的自适应直方图均衡化(CLAHE)对待检测图像进行图片增强,其次进行双边滤波的预处理,由于阈值的选取需要人为干预,采用最大类间方差法实现自适应选取高低阈值。通过对比实验发现,改进后的算法性能优于传统的Canny边缘检测方法,可以检测出更多的边缘细节,轮廓更加明显,并且图像质量提升了7%~15%。保持了Canny算法自适应强、去除干扰能力强的优点。In the process of diagnosis and treatment,patients with COVID-19 are judged by doctors on the lesions of lung X-rays,and doctors give diagnosis results based on professional knowledge and experience.In order to ensure that doctors give correct and reliable diagnosis results during the diagnosis process,providing lung image pictures with clear edges and high signal-to-noise ratio is a key step in the diagnosis and treatment of new coronary pneumonia patients.At present,the method of using the Canny edge detection algorithm combined with deep learning to realize the COVID-19 X-ray image detection method has good detection effect,but the generalization ability is poor and the diagnosis efficiency is low.Aiming at the problem that Gaussian filtering in the traditional Canny algorithm cannot eliminate the salt and pepper noise,and the adaptability of manual selection of high and low thresholds is insufficient,an improved Canny algorithm is proposed,and the contrast-limited adaptive histogram equalization(CLAHE)is first used to enhance the picture of the detected image.Secondly,the preprocessing of bilateral filtering is carried out,and since the selection of thresholds requires human intervention,the maximum between-class variance method is used to realize adaptive selection of high and low thresholds.Through comparative experiments,it is found that the improved algorithm performs better than the traditional Canny detection method,which can detect more edge details,more obvious contours,and improve image quality by 7%~15%.The advantages of Canny's algorithm with strong adaptability and strong ability to remove interference are maintained.

关 键 词：边缘检测 新冠肺炎 双边滤波 最大类间方差法 

分 类 号：TN91[电子电信—通信与信息系统]