Python Workshop: OpenCV
Introduction
OpenCV (Open Source Computer Vision Library) is an open source computer vision and machine learning software library. OpenCV was built to provide a common infrastructure for computer vision applications and to accelerate the use of machine perception in the commercial products. [https://opencv.org/about/]
Basic usage
In [1]:
import cv2 # opencv for python package
import matplotlib.pyplot as plt
figsize = (10, 10)
In [2]:
# to run in google colab
import sys
if "google.colab" in sys.modules:
def download_from_web(url):
import requests
response = requests.get(url)
if response.status_code == 200:
with open(url.split("/")[-1], "wb") as file:
file.write(response.content)
else:
raise Exception(f"Failed to download the image. Status code: {response.status_code}")
download_from_web("https://github.com/YoniChechik/AI_is_Math/raw/master/c_01_intro_to_CV_and_Python/Lenna.png")
download_from_web(
"https://github.com/YoniChechik/AI_is_Math/raw/master/c_01_intro_to_CV_and_Python/opencv_logo.png"
)
This is how to read and plot an image with opencv
In [3]:
img = cv2.imread("Lenna.png")
In [4]:
plt.figure(figsize=figsize)
plt.imshow(img)
plt.title("Lenna orig")
plt.show()
We got a weird image colors... This is because OpenCV uses image reading convention of BGR and matplotlib uses RGB.
The fix is easy:
In [5]:
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
plt.figure(figsize=figsize)
plt.imshow(img)
plt.title("Lenna RGB")
plt.show()
In [6]:
# some image info:
print(type(img))
print(img.shape)
img
Out[6]:
In [7]:
# show only red channel
plt.figure(figsize=figsize)
plt.imshow(img[:, :, 0])
plt.title("Lenna red channel")
plt.show()
In [8]:
# gray color-mapping
fig, ax_arr = plt.subplots(1, 2, figsize=figsize)
ax_arr[0].imshow(img[:, :, 0], cmap="gray")
ax_arr[0].set_title("Auto-adjusted pixel\n scale intensity")
ax_arr[1].imshow(img[:, :, 0], cmap="gray", vmin=0, vmax=255) # 255 is the max of uint8 type number (== 2**8 -1)
ax_arr[1].set_title("Absolute pixel\n scale intensity")
plt.show()
In [9]:
# image blurring
img_blurred = cv2.GaussianBlur(img, (15, 15), 7) # use a 15x15 Gaussian kernel with standard deviation 7
plt.figure(figsize=figsize)
plt.imshow(img_blurred)
plt.title("Lenna blurred")
plt.show()
In [10]:
# edge detection
img_canny = cv2.Canny(img, 180, 200) # end args are the lower & upper TH of hysteresis
plt.figure(figsize=figsize)
plt.imshow(img_canny, cmap="gray")
plt.title("Lenna edges")
plt.show()
In [11]:
# detect circles
img = cv2.imread("opencv_logo.png")
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img_gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
circles = cv2.HoughCircles(img_gray, cv2.HOUGH_GRADIENT, 0.1, 50, param1=50, param2=38)
for x, y, r in circles[0, :]:
# draw the outer circle
cv2.circle(img, (int(x), int(y)), int(r), (0, 0, 0), 2)
# draw the center of the circle
cv2.circle(img, (int(x), int(y)), 2, (0, 0, 255), 3)
plt.figure(figsize=figsize)
plt.imshow(img)
plt.title("finding circles")
plt.show()
