从 jpeg 中删除文本

我有一个包含 alpha 混合文本的 jpeg。知道字体和大小后，我推导出一个代表文本的 png 文件

使用 ImageMagick，我可以获得原始图片的近似值吗？

实现此目的的一种方法是使用一种称为修复的技术。您可以在 (Python) Skimage 中找到它

http://scikit-image.org/docs/dev/api/skimage.restoration.html#inpaint-biharmonic http://scikit-image.org/docs/dev/api/skimage.restoration.html#inpaint-biharmonic

或者在 OpenCV 中

https://docs.opencv.org/3.0-beta/modules/photo/doc/inpainting.html https://docs.opencv.org/3.0-beta/modules/photo/doc/inpainting.html https://docs.opencv.org/3.4.0/df/d3d/tutorial_py_inpainting.html https://docs.opencv.org/3.4.0/df/d3d/tutorial_py_inpainting.html

这是Python Skimage修复处理：

小猫图片：

水印图片：

Skimage 修复需要二值掩模图像。所以我可以通过以下方式将你的水印转换为这样的蒙版：

convert watermark.png -alpha extract -threshold 0 mask.png

蒙版图像：

这是Python代码：

#!/opt/local/bin/python3.6

import numpy as np
import skimage.io
import skimage.restoration
import skimage.exposure

img = skimage.io.imread('/Users/fred/desktop/kitty.png')
msk = skimage.io.imread('/Users/fred/desktop/mask.png')
msk = skimage.exposure.rescale_intensity(msk, in_range='image', out_range=(0,1))
newimg = skimage.restoration.inpaint_biharmonic(img, msk, multichannel=True)
skimage.io.imsave('/Users/fred/desktop/kitty_inpaint_biharmonic.png', newimg)

Imagemagick 没有官方版本。但 Imagemagick 论坛上的用户 snibgo 已经实现了一个他称之为“漏洞填充”的自定义版本http://im.snibgo.com/fillholespri.htm http://im.snibgo.com/fillholespri.htm。他展示了一个例子.

此外，在同一页面上，他展示了一些聪明的 Imagemagick 代码，可以重复进行少量的大小调整。这实现了与修复有些相似的结果。但总的来说，它不会像修复那么好。尽管如此，它确实对你的形象起到了一定的作用。

小猫图片：

水印图片：

首先，我必须获取您的水印图像并从中提取二进制图像，其中文本为白色，背景为黑色。然后我用它使文本所在的小猫图像透明。然后我裁剪掉文本区域只是为了让后续处理更快。

convert kitty.png \
\( watermark.png -alpha extract -threshold 0 -negate \) \
-alpha off -compose copy_opacity -composite \
-crop 490x102+235+150 +repage tmp1.png

然后我运行他相当长的连续调整图像大小的序列，然后合并所有图层并将大小调整回原始大小。

convert tmp1.png \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
\( +clone -resize 90.9091% \) \
-layers RemoveDups \
-filter Gaussian -resize 490x102! \
-background None \
-compose DstOver -layers merge \
-alpha opaque \
tmp2.png

最后，我将这个结果合成回小猫图像上我裁剪它的位置。

convert kitty.png tmp2.png -geometry +235+150 -compose over -composite kitty2.png

在全分辨率下，您仍然可以看出该图像中残留的非常微弱的文本。通过快速交替两张图像可以看出，Skimage 结果更好。