首页 | 本学科首页   官方微博 | 高级检索  
     

基于区域特征的胶囊图像分割算法
引用本文:张强,秦勃. 基于区域特征的胶囊图像分割算法[J]. 计算机系统应用, 2015, 24(10): 212-216
作者姓名:张强  秦勃
作者单位:中国海洋大学 信息科学与工程学院, 青岛 266100;中国海洋大学 信息科学与工程学院, 青岛 266100
摘    要:针对胶囊缺陷检测中存在的图像分割效果不理想的问题, 提出了一种基于区域特征的胶囊图像分割算法. 首先将原图像分割成5个子图像, 然后分别在子图图像中分割提取胶囊. 子图图像首先对图像高亮区域作去高光处理、去除噪声, 然后将图像区域的每一行作为一个子区域, 根据胶囊在图像区域中所在的位置特点, 通过判断子区域中链板域与背景域是否存在边界点以及胶囊与链板上的链齿是否连接来识别不同类型的子区域, 寻找子区域中胶囊与非胶囊区域的边界, 然后去除非胶囊区域. 最终对图像区域逐行扫描处理完成后从图像中提取出胶囊. 实验表明该算法与传统方法相比, 不仅速度较快, 准确性和鲁棒性也得到了改善.

关 键 词:机器视觉  胶囊缺陷检测  图像分割  区域特征  子图分割
收稿时间:2015-02-02
修稿时间:2015-04-29

Capsule Image Segmentation Algorithm Based on Region Feature
ZHANG Qiang and QIN Bo. Capsule Image Segmentation Algorithm Based on Region Feature[J]. Computer Systems& Applications, 2015, 24(10): 212-216
Authors:ZHANG Qiang and QIN Bo
Affiliation:College of Information Science and Engineering, Ocean University of China, Qingdao 266100, China;College of Information Science and Engineering, Ocean University of China, Qingdao 266100, China
Abstract:For the image segmentation effect is not ideal in capsule defect detection, a segmentation algorithm of capsule image based on region feature is proposed. First, the original image is divided into 5 sub images, and then capsules are extracted from sub images. First of all, sub-image is removed highlight and noise, then put each row of image region as a sub-region, according to the characteristics of the capsule position in image region, find the boundary of capsule and non-capsule region and remove non-capsule region by judging whether exist boundary points between chain plate and background or not and whether capsule is connected with sprocket or not. Finally the capsule is extracted from sub-image after scanning the sub-image region line by line. Experiments show that the algorithm is fast and its accuracy and robustness are improved noticeably.
Keywords:machine vision  capsule defect detection  image segmentation  region feature  sub-image segmentation
点击此处可从《计算机系统应用》浏览原始摘要信息
点击此处可从《计算机系统应用》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号