创建与原图片一样的透明图片

import cv2 
import numpy as np

def vertical_grad(src,color_start,color_end):
    h = src.shape[0]
    print( type(src))
    # 创建一幅与原图片一样大小的透明图片
    grad_img = np.ndarray(src.shape,dtype=np.uint8)
    # opencv 默认采用 BGR 格式而非 RGB 格式
    g_b = float(color_end[0] - color_start[0]) / h
    g_g = float(color_end[1] - color_start[1]) / h
    g_r = float(color_end[2] - color_start[2]) / h

    for i in range(h):
        for j in range(src.shape[1]):

                grad_img[i,j,0] = color_start[0] + i * g_b
                grad_img[i,j,1] = color_start[1] + i * g_g
                grad_img[i,j,2] = color_start[2] + i * g_r
    return grad_img

if __name__ == '__main__':
    img = cv2.imread('bd01.jpg')
    grad_img = vertical_grad(img,(0,0,255),(0,0,0))

    cv2.imshow('lena',img)
    cv2.imshow('gradients',grad_img)

    cv2.waitKey()
    cv2.destroyAllWindows()