OpenCV中cv2.imshow显示正常但cv2.imwrite保存结果异常求助

我尝试将一张PNG图像覆盖到另一张PNG图像上，用cv2.imshow能看到正确的叠加结果，但cv2.imwrite保存后的结果却异常。

相关代码 #

coverImg = cv2.imread('./images/cover.png', cv2.IMREAD_UNCHANGED)
back = cv2.imread('./images/back.png', cv2.IMREAD_UNCHANGED)

x_offset = y_offset = 0

y1, y2 = y_offset, y_offset + coverImg.shape[0]
x1, x2 = x_offset, x_offset + coverImg.shape[1]

alpha_s = coverImg[:, :, 3] / 255.0
alpha_l = 1.0 - alpha_s

result = back.copy()

for c in range(0, 3):
    result[y1:y2, x1:x2, c] = (alpha_s * coverImg[y1:y2, x1:x2, c] +
                               alpha_l * result[y1:y2, x1:x2, c])

cv2.imshow("result", result)
res2 = cv2.imwrite("./result.png", result)

注：result的dtype为uint8

图像情况 #

• cv2.imshow显示结果：覆盖图半透明叠加在背景图上，效果符合预期
• cv2.imwrite保存结果：叠加区域出现异常色块，颜色失真
• 原始背景图back.png：纯灰色背景的PNG图像
• 原始覆盖图cover.png：带透明通道的半白色圆形PNG图像

问题原因 #

核心问题是浮点数运算后的数值截断：计算alpha_s * coverImg[...] + alpha_l * result[...]得到的是浮点数，但直接赋值给uint8类型的数组时，OpenCV会直接舍弃小数部分（而非四舍五入），导致像素值计算错误。cv2.imshow会自动适配浮点数到整数的显示，而cv2.imwrite对数据类型和数值范围的校验更严格，因此保存结果异常。

解决方案 #

方案1：手动处理浮点数计算与类型转换

先以浮点数精度完成计算，再通过四舍五入转换为uint8类型，避免截断错误：

import numpy as np
import cv2

coverImg = cv2.imread('./images/cover.png', cv2.IMREAD_UNCHANGED)
back = cv2.imread('./images/back.png', cv2.IMREAD_UNCHANGED)

x_offset = y_offset = 0
y1, y2 = y_offset, y_offset + coverImg.shape[0]
x1, x2 = x_offset, x_offset + coverImg.shape[1]

alpha_s = coverImg[:, :, 3] / 255.0
alpha_l = 1.0 - alpha_s

result = back.copy()

# 创建浮点数临时数组存储计算结果
temp = np.zeros_like(result, dtype=np.float32)
for c in range(3):
    temp[y1:y2, x1:x2, c] = alpha_s * coverImg[y1:y2, x1:x2, c] + alpha_l * result[y1:y2, x1:x2, c]

# 四舍五入后转换为uint8类型
result[y1:y2, x1:x2] = np.round(temp[y1:y2, x1:x2]).astype(np.uint8)

cv2.imshow("result", result)
cv2.imwrite("./result.png", result)

方案2：使用cv2.addWeighted（推荐）

利用OpenCV内置的加权叠加函数，自动处理数值计算与类型转换，代码更简洁：

import cv2

coverImg = cv2.imread('./images/cover.png', cv2.IMREAD_UNCHANGED)
back = cv2.imread('./images/back.png', cv2.IMREAD_UNCHANGED)

x_offset = y_offset = 0
y1, y2 = y_offset, y_offset + coverImg.shape[0]
x1, x2 = x_offset, x_offset + coverImg.shape[1]

# 分离覆盖图的RGB通道和alpha通道
cover_rgb = coverImg[:, :, :3]
alpha = coverImg[:, :, 3] / 255.0

# 加权叠加
result = back.copy()
result[y1:y2, x1:x2] = cv2.addWeighted(cover_rgb, alpha, result[y1:y2, x1:x2], 1 - alpha, 0)

cv2.imshow("result", result)
cv2.imwrite("./result.png", result)

内容的提问来源于stack exchange，提问作者pigLoveRabbit520