python - 为什么 patch.contains_point() 在检查点是否在多边形内时与 patch.get_path().contains_point() 的行为不同？

Question

对于 matplotlib.patches，patch.contains_point(xy) 方法似乎与 patch.get_path().contains_point(xy) 不同，至少在拥有之后将补丁添加到轴上。请参阅下面的区别 True/True 和 True/False。我找不到关于这种差异的任何文档。有人知道吗？我也很难看出 contains_point() 如何确定点是否在路径内，因为在这种情况下路径的顶点是单位矩形而不是我指定的矩形。

fig, ax = plt.subplots()
rect = patches.Rectangle([0.2, 0.3], 0.8, 0.5)

pnt = [0.4, 0.45] # point inside rect
print("Before adding patch to axes:")
print(rect.get_path().vertices)
print(rect.contains_point(pnt))
print(rect.get_path().contains_point(pnt))

print("After adding patch to axes")
ax.add_patch(rect)
print(rect.get_path().vertices)
print(rect.contains_point(pnt))
print(rect.get_path().contains_point(pnt))

plt.show()

Before adding patch to axes:
[[0. 0.]
 [1. 0.]
 [1. 1.]
 [0. 1.]
 [0. 0.]]
True
True
After adding patch to axes
[[0. 0.]
 [1. 0.]
 [1. 1.]
 [0. 1.]
 [0. 0.]]
False
True

Answer 1

虽然这个问题很老，但我遇到了同样的问题并解决了它。

问题是在向轴添加补丁后，您需要在显示引用框架中给出坐标/点。这可以通过以下方式执行:

ax.transData.transform()

我在忽略导入语句的代码中添加了一行。所以代码放在这里:

import matplotlib.pyplot as plt
import matplotlib.patches as patches

fig, ax = plt.subplots()
rect = patches.Rectangle([0.2, 0.3], 0.8, 0.5)

pnt = [0.4, 0.45] # point inside rect
print("Before adding patch to axes:")
print(rect.get_path().vertices)
print(rect.contains_point(pnt))
print(rect.get_path().contains_point(pnt))

print("After adding patch to axes")
ax.add_patch(rect)
print(rect.get_path().vertices)
# added lines
pnt_in_display_coordiantes = ax.transData.transform(pnt)
print(rect.contains_point(pnt_in_display_coordiantes))
print(rect.get_path().contains_point(pnt))

plt.show()

输出:

Before adding patch to axes:
[[0. 0.]
[1. 0.]
[1. 1.]
[0. 1.]
[0. 0.]]
True
True
After adding patch to axes
[[0. 0.]
[1. 0.]
[1. 1.]
[0. 1.]
[0. 0.]]
True
True

更多信息:https://matplotlib.org/stable/tutorials/advanced/transforms_tutorial.html