Opencv fills the rectangle with a transparent color

Fill an irregular figure in an image

Fixed color fill

use cv2.fillPoly function

img = cv2.imread('./001.png')
area = np.array([[1000, 200], [1500, 200], [1500, 400], [1000, 400]])
cv2.fillPoly(img,[area], color=(255, 255, 255))

The effect is as shown in the picture , There can be multiple area Coordinate group , Set the color yourself

Transparent color fill

There is no function dedicated to filling transparent colors
But we can use image fusion cv2.addWeighted
（ Note that this method can only fill the rectangle with transparent color , Irregular shapes cannot ）

import cv2
import numpy as np


img = cv2.imread('./001.png')
#  Create a three channel image   The image size can be set by yourself  np.zeros(H,W,C)
create_green = np.zeros((100,100,3),dtype=np.uint8)
#  Set the color values of the three channels of the image 
# img[ y: y+h , x : x+w ]  Split the image 
# numpy yes BGR  So the first 0 The first channel is B  Blue  1 yes G  green  2 yes R  Red 
create_green[:,:,0] = 0
create_green[:,:,1] = 255 # Here I create a pure green image 
create_green[:,:,2] = 0

#  Image fusion   The two images must be the same size to merge  0.7  and  0.3 Is the parameter 
#  The formula for adding images is  R=a*x1+b*x2+c x1 x2 Represents two images  a b Two parameters  c Is a bias item. 
#  The fusion degree parameter can be adjusted   To control “ transparency ”
img_add = cv2.addWeighted(img[100:200,100:200], 0.7 ,create_green, 0.3, 0)
#  The fused image is then put back to the original 
img[100:200,100:200] = img_add
#  Show it after modifying the size 
h_h = img.shape[0]
w_w = img.shape[1]
resize = cv2.resize(img,( int(w_w / 2) ,int(h_h / 2) ) ,interpolation = cv2.INTER_CUBIC)
cv2.imshow("test",resize)
cv2.waitKey(0)

The effect is as follows ：