摘要
提出一种针对不规则多形变红血球图像的自适应曲面拟合并计算曲率的方法,通过阴影恢复技术重构细胞表面的高度场形状,利用三维数据点根据最小二乘法进行曲面拟合,选定深度均方误差阈值来决定参与拟合的邻域点。计算得到的高斯曲率、平均曲率可用来表征某一点的表面类型,主曲率则用来观察曲面变化较大的区域。实验结果表明,该方法具有很强的可行性与实用性。
An adaptive approach for surface fitting and curvature computation oferythrocyte with deformability is proposed. The height field is reconstructed by shape from shading technique, which contributed to surface fitting using least square methods. The points associated with fitting are determined by presetting depth root mean square error value. The mean curvature and Gaussian curvature represented the surface type of a point. The principal curvature indicated points with big difference compared to neighborhood. The results show that this approach is practicable and promising.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第6期1555-1558,共4页
Computer Engineering and Design
关键词
曲率计算
自适应算法
不规则形状
图像分割
阴影恢复
curvature computation
adaptive algorithm
irregular shape
image segmentation
shape from shading
