如何在2d numpyarray中用1替换小于或等于0的值？ [英] How can I replace values less than or equal to 0 with 1 in a 2d numpyarray?

问题描述

我需要采用两个图像numpyarrays的比例，不幸的是数据有一堆负值和零点散落。

I need to take the ratio of two image numpyarrays, and unfortunately the data has a bunch of negative values and zeros scattered about.

我找到了一个优雅的方法从numpyarray中删除值， pixels = pixels [pixels> 0] （非常优雅），但遗憾的是不会替换值。

I found one elegant way to remove values from a numpyarray, pixels = pixels[pixels>0] (really elegant), but unfortunately doesn't replace values.

我还尝试使用while循环，找到当前最小值和当前最小索引，将每个指定为1，直到没有值< = 0，但它没有'工作。

I also tried using a while loop, finding the current minimum value, and the current minimum index, assigning each to 1 until there is no value <=0, but it didn't work.

while(np.amin(pixels)<= 0):
     print "found low value of ", np.amin(evenpixels), "at", np.unravel_index(np.argmin(evenpixels), im_sz), "setting to 1"
     pixels[np.unravel_index(np.argmin(pixels), im_sz)] = 1 

我对这个while循环有一些奇怪的行为，它永远无法退出尝试将最终的低值更改为1后的循环：它在（0,0）处找到0并在此处卡住。我无法理解，我想知道它是不是一个无关的bug。请注意，我的数据形状是（2160,2560）。

I got some weird behavior with this while loop and it is never able to exit the loop after trying to change the final low value to 1: it finds a 0 at (0,0) and get stuck here. I can't understand that and I wonder if it is an unrelated bug. Note that the shape of my data is (2160, 2560).

推荐答案

如果像素的最小值可以是1（正整数），那么一个简单的修复是使用 np.clip ，例如：

If the minimum value a pixel can be is 1 (positive integers), then a simple fix is to use np.clip, eg:

pixels = np.clip（pixels，min，np.inf）

在这种情况下， min = 1 。

这篇关于如何在2d numpyarray中用1替换小于或等于0的值？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！