摘要
针对透明胶囊的成像特点,提出一种实时性好、健壮性较强的结合阈值分割、腐蚀、轮廓提取及区域填充的图像提取算法.利用旋转和一阶差分分割胶囊本体,通过最小包围矩形测量胶囊的尺寸,并与公认标准进行对比,从而判别胶囊是否存在尺寸缺陷.实验结果表明:胶囊图像提取和尺寸缺陷识别算法的平均漏检率为6.38%,平均误检率为4.13%.
In view of the characteristics of transparent capsule image,this paper proposed the image extraction algorithm combining with threshold segmentation,corrosion,contour extraction and region filling with good real-time and strong robustness.Then by using rotating and first-order difference divided capsule ontology,through the minimum bounding rectangle measured the size of capsule,compared with generally accepted standards we can distinguish whether capsule exists size defect.The experimental results show that：the average missing rate of capsule image extraction and the size defect recognition algorithm was 6.38%,the average error rate was 4.13%.
出处
《华侨大学学报（自然科学版）》
北大核心
2017年第5期716-720,共5页
Journal of Huaqiao University(Natural Science)
基金
福建省科技厅资助项目(2013H2002)
关键词
尺寸缺陷
胶囊
图像处理
区域分割
size defect
capsule
image processing
region segmentation
