将两幅图添加起来，构成一幅新的图像，并尝试将一个PNG的透明背景转换为特定的颜色。

生成纯背景

以下代码生成一个纯色背景JPG，颜色为照片底板的蓝（R:60 G:140 B:220）。在用OpenCV创建图像时，颜色按BGR。

import cv2 import numpy as np  def create_blank(width, height, rgb_color=(60, 140, 220)):     """Create new image(numpy array) filled with certain color in RGB"""     # Create black blank image     image = np.zeros((height, width, 3), np.uint8)      # Since OpenCV uses BGR, convert the color first     color = tuple(reversed(rgb_color))     # Fill image with color     image[:] = color      return image  img = cv2.imread('dog.png') (rows, cols) = img.shape[:2] blank = create_blank(cols, rows) cv2.imwrite('blue.jpg', blank) 

两个图像相加

两个图像大小相同的情况下，可以通过矩阵叠加方法——OpenCV Python中，为cv2.add，合并在一起，代码如下：

import cv2  img1 = cv2.imread('dog.png') img2 = cv2.imread('blue.jpg')  # Add 2 images img_dst = cv2.add(img1, img2)  # Display the processed images  cv2.imshow('img', img_dst ) cv2.imwrite('add.jpg', img_dst)  # Wait for a key press and then close the windows cv2.waitKey(0) cv2.destroyAllWindows() 

显然，这是矩阵个元素数值的相加，那么，假如要把透明背景用别的颜色替换，而不是这样相加，又该“如何应对”？

透明色转换

读入一个有透明背景的png图片，然后我们会发现，它的α通道是0。

import cv2  # Read a png with alpha channel img = cv2.imread('dog.png', cv2.IMREAD_UNCHANGED) cv2.imshow('img', img) cv2.imwrite('dog.png', img) # Wait for a key press and then close the windows cv2.waitKey(0) cv2.destroyAllWindows() 

对于图像的透明部分，其值为[0,0,0,0]，那么我们可以将该透明部分修改成对应的颜色，具体见函数transparent2bgra_color。

import cv2 import numpy as np  def transparent2bgra_color(img, bgra_color=(255, 255, 255, 255)):     # Avoid the side-effect     res = img.copy()     res[img[:, :, 3] == 0] = bgra_color     return res  # Read the png image with alpha channel origin_img = cv2.imread('dog.png', cv2.IMREAD_UNCHANGED) dst_img = transparent2bgra_color(origin_img, bgra_color=(220, 140, 60, 255)) black_dog = transparent2bgra_color(origin_img, bgra_color=(0, 0, 0, 255))  cv2.imshow('Origin', origin_img) cv2.imshow('Converted', dst_img) cv2.imshow('Black', black_dog) cv2.imwrite('new_dog.png', dst_img) cv2.imwrite('black_dog.png', black_dog) # Wait for a key press and then close the windows cv2.waitKey(0) cv2.destroyAllWindows() 

Origin:

Converted:

Black: