plt.imshow（）无法正确显示numpy矩阵

Question

I'm having trouble drawing a rectangle on an image with OpenCV, and then displaying the image with matplotlib. This code in a Jupyter Notebook results in a black image being displayed, with no green rectangle:

import numpy as np
import cv2
import matplotlib.pyplot as plt
import matplotlib.image as mpimg

%matplotlib inline
image = np.zeros((720, 1280, 3))
cv2.rectangle(image, (100,100), (200,200), color=(0,255,0), thickness=2) # Green
plt.imshow(image)

I believe I can see the rectangle if I display the image with cv2.imshow() , but I don't know that OpenCV images can be displayed inline in Jupyter Notebooks like matplotlib images can.

EDIT: I think what's really going on is explained in this excerpt from the documentation :

The value for each component of MxNx3 and MxNx4 float arrays should be in the range 0.0 to 1.0.

It can handle values between 0 and 255 if it's an integer array.

Answer 1

By default when you create a numpy array using np.zeros() or np.ones() , the default data type of the matrix is set to be float, or more precisely np.float64 , And there seems to be some problem in matplotlib to render a matrix with float data type (As per convention the RGB values in an image are always integral, so we can't blame matplotlib for this.)

Now you have 2 options:

• Create the numpy matrix with np.uint8 datatype

   image = np.zeros((720, 1280, 3), dtype = np.uint8) 

• Convert the matrix in integral type before rendering it on plt.

   image = np.astype(image, np.uint8) 

Also you need to add plt.show() at the end.