我有一个使用numpy的python脚本,它应该在返回单个值之前拍摄图像并执行一些计算。当我单独执行每一行时,它按预期工作。当我将它们放在一个.py脚本中并从命令行或Canopy中运行时,它会返回一个数组。
我稍微修改了代码,不需要通常的图像输入,结果是一样的:
import numpy as np
# Instead of loading an image, generate a test case (w or wo structured noise)
roi = np.random.poisson(38,(256,256));
blob = np.random.poisson(5,(128,128));
roi[64:192,64:192] = roi[64:192,64:192]+blob;
# Load the other variables if necessary (i.e., no DICOM to load)
[xDim,yDim] = [512,512];
roiLength = xDim/2;
pix = 1.18958;
# Declare memory for the FFTs
sizeFFT = xDim;
NPS2D = np.zeros((sizeFFT,sizeFFT)); # declare memory for fft results
fftslice = np.zeros((sizeFFT,sizeFFT));
# Set the dimension of the ROI and pull the pixel size. This will be
# used for the scaling factor in the 2D NPS.
deltaX = pix;
deltaY = pix;
scaleFactor = (deltaX/roiLength)*(deltaY/roiLength);
# Calculate the NPS
roiMean = np.mean(roi);
fftslice = np.fft.fft2((roi-roiMean),s=[sizeFFT,sizeFFT]);
NPS2D = scaleFactor*np.fft.fftshift(np.multiply(fftslice,np.conj(fftslice)));
NPS2D = NPS2D.real;
# Subtract the white noise from the NPS to get the structured NPS
stNPS = NPS2D - roiMean*deltaX*deltaY;
# Calculate SNI
SNI=sum(stNPS)/sum(NPS2D);
# Display the result
print SNI;
如果我执行每一行的结果是0.107213670449(或类似的,因为它每次都重新生成一个随机数组)。如果我使用python foo.py
从命令行运行脚本或单击Canopy中的播放按钮,结果是512长度数组[4.64940089e-03 ... -4.59789051e-02 -7.15113682e-02]
,其中我手动删除了509个条目。
有什么想法?我错过了一些明显的东西吗
答案 0 :(得分:1)
SNI=sum(stNPS)/sum(NPS2D)
以默认的python方式对列进行求和。所以,你得到一个长度为512的数组
而是从numpy
尝试sum
SNI=stNPS.sum()/NPS2D.sum()
答案 1 :(得分:1)
使用内置sum
功能与使用数组的numpy.sum
或sum
方法不同。
对于> 1d数组,python的sum
会产生非常不同的结果:
In [1]: import numpy as np
In [2]: x = np.arange(100).reshape(10, 10)
In [3]: x
Out[3]:
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, 27, 28, 29],
[30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
[40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
[50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
[60, 61, 62, 63, 64, 65, 66, 67, 68, 69],
[70, 71, 72, 73, 74, 75, 76, 77, 78, 79],
[80, 81, 82, 83, 84, 85, 86, 87, 88, 89],
[90, 91, 92, 93, 94, 95, 96, 97, 98, 99]])
In [4]: sum(x)
Out[4]: array([450, 460, 470, 480, 490, 500, 510, 520, 530, 540])
In [5]: x.sum()
Out[5]: 4950
In [6]: np.sum(x)
Out[6]: 4950
这是因为python的总和基本上是对象的for循环。
在> 1d数组上循环将沿第一轴返回切片。 E.g。
In [7]: for item in x:
...: print item
...:
[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 27 28 29]
[30 31 32 33 34 35 36 37 38 39]
[40 41 42 43 44 45 46 47 48 49]
[50 51 52 53 54 55 56 57 58 59]
[60 61 62 63 64 65 66 67 68 69]
[70 71 72 73 74 75 76 77 78 79]
[80 81 82 83 84 85 86 87 88 89]
[90 91 92 93 94 95 96 97 98 99]
在这种情况下,Python的sum
实际上为您提供了列的总和(即row1 + row2 + row3 ...
)