技术背景
对于各行各业的研究人员来说，经常会面临这样的一个问题：有一篇不错的文章里面有很好的数据，但是这个数据在文章中仅以图片的形式出现。而假如我们希望可以从该图片中提取出数据，这样就可以用我们自己的形式重新来展现这些数据，还可以额外再附上自己优化后的数据。因此从论文图片中提取数据，是一个非常实际的需求。这里以前面写的量子退火的博客为例，博客中有这样的一张图片：
在这篇文章中，我们将介绍如何使用python从图片上把数据抠取出来。
plotdigitizer的安装
这里我们使用pip来安装python第三方库plotdigitizer，该库的主要功能就是可以自动化的从图片中提取出数据，我们可以使用腾讯的pip镜像源来加速我们的安装过程：
[dechin@dechin-manjaro plotdigitizer]$ python3 -m pip install -i https://mirrors.cloud.tencent.com/pypi/simple plotdigitizer Looking in indexes: https://mirrors.cloud.tencent.com/pypi/simple Collecting plotdigitizer Downloading https://mirrors.cloud.tencent.com/pypi/packages/89/bb/ff753093458c05ce3b52fd17527b6b0622ca096aadcf561c6316320ab793/plotdigitizer-0.1.3-py3-none-any.whl (20 kB) Collecting loguru<0.6.0,>=0.5.3 Downloading https://mirrors.cloud.tencent.com/pypi/packages/6d/48/0a7d5847e3de329f1d0134baf707b689700b53bd3066a5a8cfd94b3c9fc8/loguru-0.5.3-py3-none-any.whl (57 kB) |████████████████████████████████| 57 kB 521 kB/s Collecting opencv-python<5.0.0,>=4.5.1 Downloading https://mirrors.cloud.tencent.com/pypi/packages/2a/9a/ff309b530ac1b029bfdb9af3a95eaff0f5f45f6a2dbe37b3454ae8412f4c/opencv_python-4.5.1.48-cp38-cp38-manylinux2014_x86_64.whl (50.4 MB) |████████████████████████████████| 50.4 MB 467 kB/s Collecting numpy<2.0.0,>=1.19.5 Downloading https://mirrors.cloud.tencent.com/pypi/packages/c7/e6/dccac76b7e825915ffb906beeba5a953597b6cfe1fe686b5276e122cb07c/numpy-1.20.1-cp38-cp38-manylinux2010_x86_64.whl (15.4 MB) |████████████████████████████████| 15.4 MB 20.4 MB/s Collecting matplotlib<4.0.0,>=3.3.4 Downloading https://mirrors.cloud.tencent.com/pypi/packages/ab/20/60cfe5d611ac86df07b7b1f9b9582f22f7eda5edbe2124ba85bdf3133822/matplotlib-3.3.4-cp38-cp38-manylinux1_x86_64.whl (11.6 MB) |████████████████████████████████| 11.6 MB 4.4 MB/s Requirement already satisfied: python-dateutil>=2.1 in /home/dechin/anaconda3/lib/python3.8/site-packages (from matplotlib<4.0.0,>=3.3.4->plotdigitizer) (2.8.1) Requirement already satisfied: cycler>=0.10 in /home/dechin/anaconda3/lib/python3.8/site-packages (from matplotlib<4.0.0,>=3.3.4->plotdigitizer) (0.10.0) Requirement already satisfied: pillow>=6.2.0 in /home/dechin/anaconda3/lib/python3.8/site-packages (from matplotlib<4.0.0,>=3.3.4->plotdigitizer) (8.0.1) Requirement already satisfied: kiwisolver>=1.0.1 in /home/dechin/anaconda3/lib/python3.8/site-packages (from matplotlib<4.0.0,>=3.3.4->plotdigitizer) (1.3.0) Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.3 in /home/dechin/anaconda3/lib/python3.8/site-packages (from matplotlib<4.0.0,>=3.3.4->plotdigitizer) (2.4.7) Requirement already satisfied: six>=1.5 in /home/dechin/anaconda3/lib/python3.8/site-packages (from python-dateutil>=2.1->matplotlib<4.0.0,>=3.3.4->plotdigitizer) (1.15.0) Installing collected packages: loguru, numpy, opencv-python, matplotlib, plotdigitizer Attempting uninstall: numpy Found existing installation: numpy 1.19.2 Uninstalling numpy-1.19.2: Successfully uninstalled numpy-1.19.2 Attempting uninstall: matplotlib Found existing installation: matplotlib 3.3.2 Uninstalling matplotlib-3.3.2: Successfully uninstalled matplotlib-3.3.2 Successfully installed loguru-0.5.3 matplotlib-3.3.4 numpy-1.20.1 opencv-python-4.5.1.48 plotdigitizer-0.1.3
通过运行帮助指令，我们可以查看是否安装成功：
[dechin@dechin-manjaro plotdigitizer]$ plotdigitizer -h
usage: plotdigitizer [-h] --data-point DATA_POINT [--location LOCATION] [--plot PLOT] [--output OUTPUT]
[--preprocess] [--debug]
INPUT
Digitize image.
positional arguments:
INPUT Input image file.
optional arguments:
-h, --help show this help message and exit
--data-point DATA_POINT, -p DATA_POINT
Datapoints (min 3 required). You have to click on them later. At least 3 points
are recommended. e.g -p 0,0 -p 10,0 -p 0,1 Make sure that point are comma
separated without any space.
--location LOCATION, -l LOCATION
Location of a points on figure in pixels (integer). These values should appear in
the same order as -p option. If not given, you will be asked to click on the
figure.
--plot PLOT Plot the final result. Requires matplotlib.
--output OUTPUT, -o OUTPUT
Name of the output file else trajectory will be written to <INPUT>.traj.csv
--preprocess Preprocess the image. Useful with bad resolution images.
--debug Enable debug logger
执行指令与输出图片
先把需要抠取数据的图片放到当前目录下，然后运行如下指令：
plotdigitizer ./test1.png -p 0,-1 -p 20,0 -p 0,0.1 --plot output.png
该指令会将test1.png中的数据提取出来，可以使用-o存储为csv格式的数据表格。这里实际使用中我们发现，即使不用plot指令，也会在Manjaro Linux系统下不断的输出打印图片，只有通过kill -9的方式才能强行将进程杀死，有可能是开源库中存在的某个bug。这里展示一下用新的数据绘制出来的效果图：
执行结束后，该图片会被输出到临时文件夹tmp/plotdigitizer/下，但是注意前面产生的图片会被后来的临时文件所覆盖。
总结概要
这里我们仅仅是介绍和演示了plotdigitizer的基本使用方法，这样一个使用python制作的图像数据工具更加符合pythoner的使用习惯和逻辑。虽然实际使用过程中工具可能出现各种各样的问题，但是基本上是一个比较好的工具，值得推荐。
版权声明
本文首发链接为：https://www.cnblogs.com/dechinphy/p/plotdigitizer.html
作者ID：DechinPhy
更多原著文章请参考：https://www.cnblogs.com/dechinphy/
到此这篇关于使用python模块plotdigitizer抠取论文图片中的数据的文章就介绍到这了,更多相关python模块plotdigitizer内容请搜索服务器之家以前的文章或继续浏览下面的相关文章希望大家以后多多支持服务器之家！
原文链接：https://www.cnblogs.com/dechinphy/p/plotdigitizer.html