geomdl库fitting.approximate_surface方法拟合局部误差较大原因

# 测试发现，100个点拟合会出现明显拟合缺陷？？？

# 读取文件数据
data = np.loadtxt(r'E:\My Files\JiQixx\code\data\0229\points\points_even_100_intersections.txt')

# 获取数据尺寸
size_u, size_v = 10, 10

# 导入geomdl库
from geomdl import fitting
from geomdl.visualization import VisMPL as vis

# 设置拟合阶数
degree = 4

# 拟合
surf = fitting.approximate_surface(data, size_u=size_u, size_v=size_v, degree_u=degree, degree_v=degree, ctrlpts_size_u=10, ctrlpts_size_v=10)

# 可视化
surf.vis = vis.VisSurface()
surf.render()

如上，采用100个网格点拟合时结果如下：

使用open3d查看更清晰：

尝试500个点左右就没有这个问题了：

# 读取文件数据
data = np.loadtxt(r'E:\My Files\JiQixx\code\data\0229\points\points_even_500_intersections.txt')

# 获取数据尺寸
size_u, size_v = 22, 22

# 导入geomdl库
from geomdl import fitting
from geomdl.visualization import VisMPL as vis

# 设置拟合阶数
degree = 4

# 拟合
surf = fitting.approximate_surface(data, size_u=size_u, size_v=size_v, degree_u=degree, degree_v=degree, ctrlpts_size_u=10, ctrlpts_size_v=10)

# 可视化
surf.vis = vis.VisSurface()
surf.render()

出图如下：

OK，这样就没有严重偏差的部分了，具体原因是为啥呢，因为拟合阶数不对么？有懂得大佬，多多指教！