How to make adaptive Threshold at part on image in Opencv python?

Question

Maybe my question is strange something but I need to make an adaptive Threshold on part of the image that the user selects with his mouse and that's my code

import cv2
img = cv2.imread("test.png")
# img2 = cv2.imread("flower.jpg")

# variables
ix = -1
iy = -1
drawing = False


def draw_reactangle_with_drag(event, x, y, flags, param):
    global ix, iy, drawing, img
    if event == cv2.EVENT_LBUTTONDOWN:
        drawing = True
        ix = x
        iy = y

    elif event == cv2.EVENT_MOUSEMOVE:
        if drawing == True:
            img2 = cv2.imread("test.png")
            cv2.rectangle(img2, pt1=(ix, iy), pt2=(x, y),
                          color=(0, 255, 255), thickness=1)
            img = img2

    elif event == cv2.EVENT_LBUTTONUP:
        drawing = False
        img2 = cv2.imread("test.png")
        cv2.rectangle(img2, pt1=(ix, iy), pt2=(x, y),
                      color=(0, 255, 255), thickness=1)

        img = img2
        gray = cv2.cvtColor(img2[y: iy, x: ix], cv2.COLOR_BGR2GRAY)

        th = cv2.adaptiveThreshold(gray,
                                   255,  # maximum value assigned to pixel values exceeding the threshold
                                   cv2.ADAPTIVE_THRESH_GAUSSIAN_C,  # gaussian weighted sum of neighborhood
                                   cv2.THRESH_BINARY,  # thresholding type
                                   5,  # block size (5x5 window)
                                   3)  # constant
        img = th


cv2.namedWindow(winname="Title of Popup Window")
cv2.setMouseCallback("Title of Popup Window", draw_reactangle_with_drag)

while True:
    cv2.imshow("Title of Popup Window", img)
    if cv2.waitKey(10) == 27:
        break
cv2.destroyAllWindows()

and that's what I got at attached screen What am I missing?

Create a mask of the area you want (white on black background the size of your image). Do adaptive threshold on the whole image. Use the mask to merge between the two images. np.where(mask==255, adaptive_threshold_image, input_image) — fmw42
– fmw42, Commented Feb 4, 2023 at 23:04
@fmw42 thanks for your reply, sorry I'm a beginner can you edit my code, please? — Emad Younan
– Emad Younan, Commented Feb 5, 2023 at 8:01
I won't edit your code. But if you give me the coordinates of the region that you want, I will provide my code to threshold that area. — fmw42
– fmw42, Commented Feb 5, 2023 at 17:16
What are iy and ix? Are those just the bottom right corner and y,x are the top left corner? — fmw42
– fmw42, Commented Feb 6, 2023 at 23:11

fmw42 · Accepted Answer · 2023-02-08 22:44:24Z

1

Here is one solution for the desired region in Python/OpenCV. It is to use division normalization rather than adaptive thresholding. (This may or may not work for other regions.)

Read the input
Specify crop coordinates for rectangle
Crop the image
Blur the cropped image
Divide the input by the blurred image
Save the result

Input:

import cv2
import numpy as np

# read the input
img = cv2.imread('equador.png')

# specify the crop rectangle
# 364 396 359 453   (y iy x ix)
x1 = 359
y1 = 364
x2 = 453
y2 = 396

# crop the input
crop = img[y1:y2, x1:x2]

# blur
blur = cv2.GaussianBlur(crop, (0,0), sigmaX=99, sigmaY=99)

# divide
divide = cv2.divide(crop, blur, scale=255)

# put the divide back into the input
result = img.copy()
result[y1:y2, x1:x2] = divide

# save results
cv2.imwrite('equador_crop.png', crop)
cv2.imwrite('equador_crop_blur.png', blur)
cv2.imwrite('equador_crop_divide.png', divide)
cv2.imwrite('equador_crop_divide_result.png', result)


# show results
cv2.imshow('crop', crop)
cv2.imshow('blur', blur)
cv2.imshow('divide', divide)
cv2.imshow('result', result)
cv2.waitKey(0)

Cropped Image:

Blurred Image:

Division Normalized Crop:

Division Normalized Replace:

Note: you may prefer to convert the cropped image to grayscale before blurring and then divide the grayscale version by the blurred image.

edited Feb 8, 2023 at 22:44

answered Feb 7, 2023 at 19:36

fmw42

54.1k10 gold badges80 silver badges95 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Emad Younan Over a year ago

it's amazing but still, need a last thing please how to add the Division Normalized Result into the original image, or how to replace that square from the original image with that image?

fmw42 Over a year ago

Somehow my code got deleted. I have added it back and edited it to show how to put the crop back into the image after processing. But please read up on Numpy slicing. It is a core functionality that you will need to master.

Emad Younan Over a year ago

Thank you sir very very much you saved me ^_^ , yup I will read about it :) :)

Collectives™ on Stack Overflow

How to make adaptive Threshold at part on image in Opencv python?

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related