簡(jiǎn)介

有的時(shí)候我們需要將兩張圖片在alpha通道進(jìn)行混合，比如深度學(xué)習(xí)數(shù)據(jù)集增強(qiáng)方式MixUp。OpenCV的addWeighted提供了相關(guān)操作，此篇博客將詳細(xì)介紹這個(gè)函數(shù)，并給出代碼示例。????

函數(shù)原型

$$\mathrm{outputImg}=\mathrm{saturate}(\alpha ?inputImg1+\beta ?inputImg2+\gamma )$$

cv.addWeighted(	src1, alpha, src2, beta, gamma[, dst[, dtype]]	) -> dst

代碼示例

import cv2
import matplotlib.pyplot as plt

# 加載兩張圖片
img1 = cv2.imread(filename="Lenna.png")
img2 = cv2.imread(filename="horses.jpg")

# 將兩張圖片都調(diào)整到640*640
shape1 = img1.shape  # HWC
shape2 = img2.shape  # HWC
max1 = max(shape1[0], shape1[1])
max2 = max(shape2[0], shape2[1])
img1 = cv2.copyMakeBorder(
    src=img1,
    top=int((max1 - shape1[0])/2),
    bottom=int((max1 - shape1[0])/2),
    left=int((max1 - shape1[1])/2),
    right=int((max1 - shape1[1])/2),
    borderType=cv2.BORDER_REFLECT101,
)
img1 = cv2.resize(src=img1, dsize=(640, 640), interpolation=cv2.INTER_LINEAR)
img2 = cv2.copyMakeBorder(
    src=img2,
    top=int((max2 - shape2[0])/2),
    bottom=int((max2 - shape2[0])/2),
    left=int((max2 - shape2[1])/2),
    right=int((max2 - shape2[1])/2),
    borderType=cv2.BORDER_REFLECT101,
)
img2 = cv2.resize(src=img2, dsize=(640, 640), interpolation=cv2.INTER_LINEAR)

# 按照比例將兩張圖片進(jìn)行混合
alpha = 0.5
beta = 1.0 - alpha
img_blending = cv2.addWeighted(src1=img1, alpha=alpha, src2=img2, beta=beta, gamma=.0)

# 繪制圖片
fig = plt.figure(figsize=(9, 3))
fig.suptitle(t="Blend two images")

ax1 = fig.add_subplot(1, 3, 1)
ax2 = fig.add_subplot(1, 3, 2)
ax3 = fig.add_subplot(1, 3, 3)

ax1.set_title(label="image1")
ax1.spines["top"].set_visible(b=False)
ax1.spines["bottom"].set_visible(b=False)
ax1.spines["left"].set_visible(b=False)
ax1.spines["right"].set_visible(b=False)
ax1.axes.xaxis.set_visible(b=False)
ax1.axes.yaxis.set_visible(b=False)
ax1.imshow(X=cv2.cvtColor(src=img1, code=cv2.COLOR_BGR2RGB))

ax2.set_title(label="image2")
ax2.spines["top"].set_visible(b=False)
ax2.spines["bottom"].set_visible(b=False)
ax2.spines["left"].set_visible(b=False)
ax2.spines["right"].set_visible(b=False)
ax2.axes.xaxis.set_visible(b=False)
ax2.axes.yaxis.set_visible(b=False)
ax2.imshow(X=cv2.cvtColor(src=img2, code=cv2.COLOR_BGR2RGB))

ax3.set_title(label="blending image")
ax3.spines["top"].set_visible(b=False)
ax3.spines["bottom"].set_visible(b=False)
ax3.spines["left"].set_visible(b=False)
ax3.spines["right"].set_visible(b=False)
ax3.axes.xaxis.set_visible(b=False)
ax3.axes.yaxis.set_visible(b=False)
ax3.imshow(X=cv2.cvtColor(src=img_blending, code=cv2.COLOR_BGR2RGB))

plt.show()

參考資料

1. Computer Vision: Algorithms and Applications
2. OpenCV文檔：Adding (blending) two images using OpenCV??
3. OpenCV文檔：addWeighted() ??

收集整理和創(chuàng)作不易, 若有幫助??, 請(qǐng)幫忙點(diǎn)贊???收藏??, 謝謝!??????文章來源地址http://www.zghlxwxcb.cn/news/detail-737923.html

到了這里，關(guān)于詳解cv2.addWeighted函數(shù)【使用 OpenCV 添加（混合）兩個(gè)圖像-Python版本】的文章就介紹完了。如果您還想了解更多內(nèi)容，請(qǐng)?jiān)谟疑辖撬阉鱐OY模板網(wǎng)以前的文章或繼續(xù)瀏覽下面的相關(guān)文章，希望大家以后多多支持TOY模板網(wǎng)！