[PYTHON] Try blurring the image with opencv2

Purpose

Because of the article the other day, I became a little interested in image processing. However, the last time I touched mathematics was the final exam of high school 3. Moreover, 2 points. There were some things that I couldn't understand, but first I wrote the code to try it out.

First of all, installation and preparation of images

opencv2 seems to be a module for image processing. So it seems that I can't find it even if I do "pip install open cv2" easily. Apparently this is the correct one.

$ pip install opencv-python

The blurred image uses a country-like image of free material as an eye candy.

Let's blur this guy.

code

Let's write the code.

Import part + image reading

main.py

import cv2

#Load image
image = cv2.imread("inaka.jpg ")

(Supplement) If you want to convert from BGR to RGB

This opencv2 seems to express the image in BGR. And matplotlib, a library that can be used to save and draw processed images, is in RGB format. If you use matplotlib, ignore the differences here and proceed! Lol

So, when converting from BGR to RGB, do this. This time, the processed image is not passed to matplotlib, so it is unnecessary.

main.py

image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

Blur

It's finally time to start blurring. That said, just call the method.

main.py

blurr_image = cv2.GaussianBlur(image, (5, 5), 0)

This Guassian Blur, I didn't quite understand the argument. The first argument is an image object, the second argument is a numeric array called a kernel, and the third argument seems to be the standard deviation. It all sounds cosmic to me. I should have studied properly.

Export image

Now export the completed image.

main.py

cv2.imwrite("new_image.jpg ", blurr_image)

Code final form

The whole code looks like this.

main.py

import cv2

image = cv2.imread("inaka.jpg ")
blurr_image = cv2.GaussianBlur(image, (5, 5), 0)
cv2.imwrite("new_image.jpg ", blurr_image)

I will check

It's changed ...?

I tried playing with Gaussian Blur's arguments a little bit, but I couldn't grasp the rules and gave up today because I didn't have much time.

Gori push breakthrough

As an aside, I hate glasses because my eyes are too bad, and I'm afraid of contact lenses, so I can't put them in. Everything looks blurry from usual. I don't know at all whether it really blurred me. Perhaps the above image is also blurred, but I couldn't feel it, so I thought I'd turn it about 100 times repeatedly (I think it's definitely wrong).

main.py

import cv2
image = cv2.imread("inaka.jpg ")

for i in range(1,100):
    image = cv2.GaussianBlur(image, (5, 5), 0)

cv2.imwrite("new_image.jpg ", image)

I feel like it's blurred properly.

By the way, even if I increased the number of repetitions any more, the degree of blur did not increase dramatically. I feel like there was someone who was familiar with the juniors who left the company, so I'd like to ask.

Conclusion

・ Image blur can be realized with opencv2 ・ I don't know how to use Guassian Blur, but it looks like it after repeating the muddy blurring process.