基于改进Canny算子的医学图像分割研究  

Research on Medical Image Segmentation Based on Improved Canny Arithmetic

作　　者：徐立[1] 武家敏 杨洋[1] 王宇晨 刘亮[1] XU Li;WU Jiamin;YANG Yang;WANG Yuchen;LIU Liang(Department of Computer Science and Technology,Baotou Medical College,Inner Mongolia University of Science and Technology,Baotou 014040)

机构地区：[1]内蒙古科技大学包头医学院计算机科学与技术学院,包头014040

出　　处：《计算机与数字工程》2023年第2期456-461,共6页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:71363040);内蒙古自治区卫生健康科技计划项目(编号:202201399);包头医学院花蕾计划项目(编号:HL2020018,HL2021010,HLBTMC202210)资助。

摘　　要：为了提升传统Canny算子在医学图像分割中的准确性和有效性,解决图像分割后目标边缘特征点连续性差,难以形成闭合区域等问题,论文结合医学图像特点,提出一种多尺度滤波融合图像特征,结合形态学方法进行图像增强的改进Canny算子边缘检测医学图像分割算法,以实现医学图像的有效分割。实验结果表明,论文提出的方法可以准确地实现细胞显微图像和乳腺钼靶X线图像的目标分割,特别是对复杂细节区域的细碎边缘特征,可以得到较好的分割效果。In order to improve the accuracy and effectiveness of traditional Canny arithmetic in medical image segmentation,to solve the problem of poor continuity of target edge feature points and unable to form a closed area after image segmentation,an improved Canny arithmetic in medical image segmentation is proposed based on multi-scale filtering and fusion image features,morphological method for image enhancement is used to achieve effective segmentation of medical images,combined with the characteristics of medical images.The results show that the proposed method can accurately realize the target segmentation of cell microscopic image and mammary gland molybdenum target X-ray photography,especially for the fine edge features of complex detail regions,which can obtain better segmentation results.

关 键 词：CANNY算子 形态学 图像增强 多尺度滤波 医学图像分割 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]