创建3D二进制图像

时间:2016-08-30 04:00:34

标签: python numpy indexing binary ndimage

我有一个2D数组,a,包含一组100 x,y,z坐标:

[[ 0.81  0.23  0.52]
 [ 0.63  0.45  0.13]
 ...
 [ 0.51  0.41  0.65]]

我想创建一个3D二进制图像b,每个x,y,z维度有101个像素,坐标范围在0.00到1.00之间。 由a定义的位置处的像素值应为1,所有其他像素的值应为0。

我可以使用b = np.zeros((101,101,101))创建一个正确形状的零数组,但是如何使用a为其创建坐标和切片?

2 个答案:

答案 0 :(得分:4)

你可以这样做 -

# Get the XYZ indices
idx = np.round(100 * a).astype(int)

# Initialize o/p array
b = np.zeros((101,101,101))

# Assign into o/p array based on linear index equivalents from indices array
np.put(b,np.ravel_multi_index(idx.T,b.shape),1)

分配部分的运行时 -

让我们使用更大的网格进行计时。

In [82]: # Setup input and get indices array
    ...: a = np.random.randint(0,401,(100000,3))/400.0
    ...: idx = np.round(400 * a).astype(int)
    ...: 

In [83]: b = np.zeros((401,401,401))

In [84]: %timeit b[list(idx.T)] = 1 #@Praveen soln
The slowest run took 42.16 times longer than the fastest. This could mean that an intermediate result is being cached.
1 loop, best of 3: 6.28 ms per loop

In [85]: b = np.zeros((401,401,401))

In [86]: %timeit np.put(b,np.ravel_multi_index(idx.T,b.shape),1) # From this post
The slowest run took 45.34 times longer than the fastest. This could mean that an intermediate result is being cached.
1 loop, best of 3: 5.71 ms per loop

In [87]: b = np.zeros((401,401,401))

In [88]: %timeit b[idx[:,0],idx[:,1],idx[:,2]] = 1 #Subscripted indexing
The slowest run took 40.48 times longer than the fastest. This could mean that an intermediate result is being cached.
1 loop, best of 3: 6.38 ms per loop

答案 1 :(得分:4)

首先,通过安全地将浮子四舍五入到整数开始。在上下文中,请参阅this问题。

a_indices = np.rint(a * 100).astype(int)

接下来,将b中的索引指定为1.但是请注意使用普通list而不是数组,否则您将触发index arrays的使用。似乎这种方法的性能与替代品的性能相当(感谢@Divakar!: - )

b[list(a_indices.T)] = 1

我创建了一个小尺寸10而不是100的小示例,以及2个尺寸而不是3个,以说明:

>>> a = np.array([[0.8, 0.2], [0.6, 0.4], [0.5, 0.6]])
>>> a_indices = np.rint(a * 10).astype(int)
>>> b = np.zeros((10, 10))
>>> b[list(a_indices.T)] = 1
>>> print(b) 
[[ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  1.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.  0.  0.]]