[英]How to convolve array of arrays with a mask in Python?
Given a t1xt2xn array and a m1xm2 mask, how to obtain the t1xt2xn array where the n-dim arrays are convolved with the mask? 给定一个t1xt2xn数组和一个m1xm2遮罩,如何获得n-dim数组与遮罩卷积的t1xt2xn数组?
The function scipy.signal.convolve is not able to handle this because it only accept inputs with same number of dimensions. 函数scipy.signal.convolve无法处理此问题,因为它仅接受具有相同维数的输入。
Example with the "same" logic: 具有“相同”逻辑的示例:
in1 =
[[[0,1,2],[3,4,5],[6,7,8]],
[[9,10,11],[12,13,14],[15,16,17]],
[[18,19,20],[21,22,23],[24,25,26]]]
in2 =
[[0,1,0],
[0,1,0],
[0,1,0]]
output =
[[[0,0,0],[15,17,19],[0,0,0]],
[[0,0,0],[36,39,42],[0,0,0]],
[[0,0,0],[33,35,37],[0,0,0]]]
I'm very sorry but I haven't strong math background, so my answer could be wrong. 非常抱歉,但是我没有很强的数学背景,所以我的答案可能是错误的。 Anyway, if you need to use mask for selecting you should convert it to bool type.
无论如何,如果需要使用遮罩进行选择,则应将其转换为布尔型。 For example:
例如:
in1 = np.array([[[0,1,2], [3,4,5], [6,7,8]],
[[9,10,11], [12,13,14], [15,16,17]],
[[18,19,20], [21,22,23], [24,25,26]]])
in2 = np.array([[0, 1, 0],
[0, 1, 0],
[0, 1, 0]])
mask = in2.astype(bool)
print(in1[mask])
# [[ 3 4 5]
# [12 13 14]
# [21 22 23]]
in3 = np.zeros(in1.shape)
in3[mask] = np.convolve(in1[mask].ravel(), in2.ravel(), 'same').reshape(mask.shape)
print(in3)
# [[[ 0. 0. 0.]
# [ 15. 17. 19.]
# [ 0. 0. 0.]]
#
# [[ 0. 0. 0.]
# [ 36. 39. 42.]
# [ 0. 0. 0.]]
#
# [[ 0. 0. 0.]
# [ 33. 35. 37.]
# [ 0. 0. 0.]]]
I'm not very sure about last part, especially about reshaping but I hope you get an idea. 我不太确定最后一部分,特别是关于重塑,但希望您有个好主意。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.