红血球图像表面曲率的计算方法

Method for Curvature Calculation of Red Blood Cell Surface

提出一种基于阴影恢复技术的细胞图像表面各点的曲率计算方法,利用边界轮廓提取技术分离由扫描电镜SEM得到的红血球亮度图像,对单个细胞从跟踪得到的中心点开始进行区域生长,其每个像素点的亮度变化信息决定了与之相对应的高度场深度,阴影恢复技术可以将这种变化的亮度转化为不同的深度,从而得到像素的三维数据点。通过最小二乘法对三维点进行曲面拟合得到分段连续光滑的细胞表面,其形状特征由平均曲率与高斯曲率来表示。结果表明,该方法具有很强的实用性和应用价值。

This paper proposes an approach to compute curvature by means of Shape Form Shading（SFS） technique. Initially it traces the contour of SEM RBCs image, and grows regionally from center point of each cell. The height field depth is derived from the intensity variation with SFS, which is also the 3D points related to the corresponding pixel. As curvature is defined on smooth curved surface, the piece-wise smooth surface can be obtained using least squares method and surface fitting. The shape is featured by mean curvature and Gaussian curvature. Results show that it is promising and practicable.