摘要
通常 ,尿囊膜的血管图像质量较差 ,原因在于血管纵横交错地分布于尿囊膜上 ,图像的对比度低 ,从而导致血管检测工作十分困难。通过分析血管空间和光学特性 ,了解到沿血管的横向断面方向 ,其灰度呈高斯分布。利用二维高斯滤波器进行模板匹配 ,每隔 15°取一个检测方向 ,共建立 12个 13× 13方向检测模板矩阵。经处理后 ,血管从背景中显现出来 ,取得较令人满意的对比度和清晰度。
The quality of vascular images of allantois membrane is usually poor because of the poor local contrast and the complex distribution of the vesseles. Therefore, the detection of vascular tree is an uneasy task. By analyzing the spatial and optical properties of blood vessels, we knew that the gray-level profile of the cross section of a blood vessel was approximated by a Gaussian shaped curve. The method of template matching was applied. By obtaining detecting orientations every 15°, we then constructed 12 different 13 x 13 templates that could be used to search for vessel segments along all possible directions. After the implementation of two-dimensional (2-D) Gaussian filters, blood vessels clearly appeared from the background. The contrast of vascular images has been highly improved. Based on these processed images, we can do further computerized quantitative analyses.
出处
《中国生物医学工程学报》
EI
CAS
CSCD
北大核心
2003年第6期513-517,487,共6页
Chinese Journal of Biomedical Engineering
关键词
尿囊膜
血管图像
血管检测
二维高斯滤波
模板匹配
Biofilters
Biological membranes
Blood vessels
Image analysis
Optical properties
