摘要:这篇文章是在上测试和运行的的下的安装和配置请移步到这里应用程序进程树默认启动时初始化个用于处理图片的工作进程首先安装需要的工具包使用的版本而不是系统自带的创建项目模块图像处理获取宽高把原始的二进制图片数据转换为把转换为的图
这篇文章是在OSX上测试和运行的的, Ubuntu下的安装和配置请移步到这里
应用程序进程树, 默认启动 Poolboy 时, 初始化10个用于处理图片的 Python 工作进程(Worker)
首先安装OpenCV需要的工具包ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)" brew install python brew tap homebrew/science brew install opencv sudo pip install numpy sudo pip install matplotlib
使用 Homebrew 的 Python 版本, 而不是 Mac OS X 系统自带的 Python
alias python="/usr/local/bin/python"创建 Elixir 项目
➜ mix new opencv_thumbnail_server --sup * creating README.md * creating .gitignore * creating mix.exs * creating config * creating config/config.exs * creating lib * creating lib/opencv_thumbnail_server.ex * creating test * creating test/test_helper.exs * creating test/opencv_thumbnail_server_test.exs Your Mix project was created successfully. You can use "mix" to compile it, test it, and more: cd opencv_thumbnail_server mix test Run "mix help" for more commands.
Elixir 模块
require Logger defmodule OpencvThumbnailServer do use Application def start(_type, _args) do Logger.info "Start opencv thumbnail server" OpencvThumbnailServer.Supervisor.start_link() end end
defmodule OpencvThumbnailServer.Supervisor do use Supervisor @config Application.get_env :opencv_thumbnail_server, :settings def start_link() do Supervisor.start_link(__MODULE__, [], name: {:global,__MODULE__}) end def init([]) do pool_options = @config[:poolboy] {_, name} = pool_options[:name] children = [ :poolboy.child_spec(name, pool_options, @config[:module_name]) ] supervise(children, strategy: :one_for_all, max_restarts: 1000, max_seconds: 3600) end end
defmodule OpencvThumbnailServer.Worker do use GenServer @config Application.get_env(:opencv_thumbnail_server, :settings) def start_link(python_module) do GenServer.start_link(__MODULE__, python_module, []) end def call_python(worker, function, args) do GenServer.call(worker, {:call_python, function, args}, 10_000) end def init(python_module) do IO.puts "Start worker" {:ok, pid} = :python.start_link([ {:python_path, @config[:python_path]}, {:python, @config[:python]} ]) state = {python_module, pid} {:ok, state} end def handle_call({:call_python, function, args}, _from, state) do {module, pid} = state result = :python.call(pid, module, function, args) reply = {:ok, result} {:reply, reply, state} end def handle_call(_request, _from, state) do {:stop, :error, :bad_call, state} end def handle_info(_msg, {module,py_pid}) do {:stop, :error, {module,py_pid}} end def terminate(_reason, {_, py_pid}) do :python.stop(py_pid) :ok end end图像处理
获取宽高
# -*- coding: utf-8 -*- import urllib2 as urllib import numpy as np import cv2 def load_image_url(url): resp = urllib.urlopen(url) buf = resp.read() return buf def load_image_file(filename): image = cv2.imdecode(filename, cv2.IMREAD_COLOR) return image def get_photo_sizes(): return [ [160, 160], [320, 320], [640, 640], [1060, 1060], [1280, 1280] ] def show(buf): # print buf # x = cv2.imdecode(image, cv2.IMREAD_COLOR) # d = cv2.cvtColor(c, cv2.COLOR_RGB2BGR) np_ndarray = np.fromstring(buf, dtype=np.uint8) x = cv2.imdecode(np_ndarray, cv2.IMREAD_UNCHANGED) return cv2.imshow("NBA Image", x) def write(buf): nparray = np.fromstring(buf, dtype=np.uint8) img = cv2.imdecode(nparray, cv2.IMREAD_UNCHANGED) return cv2.imwrite("/tmp/imwrite.png", img) # def get_dimension(): # url = "http://img1.gtimg.com/16/1601/160106/16010642_1200x1000_0.jpg" # resp = urllib.urlopen(url) # buf = resp.read() # x = np.fromstring(buf, dtype=np.uint8) # img = cv2.imdecode(x, cv2.IMREAD_UNCHANGED) # # height = np.size(img, 0) # # width = np.size(img, 1) # height, width = image.shape[:2] # return (width, height) def get_dimension(buffer): # 把原始的二进制图片数据转换为NpArray nparray = np.fromstring(buffer, dtype=np.uint8) # 把 nparray 转换为 opencv 的图像格式 image = cv2.imdecode(nparray, cv2.IMREAD_UNCHANGED) height, width = image.shape[:2] return (width, height) def convert_color(): url = "http://ww3.sinaimg.cn/mw690/6941baebgw1epzcuv9vmxj20me0hy0u1.jpg" resp = urllib.urlopen(url) buf = resp.read() x = np.fromstring(buf, dtype=np.uint8) img = cv2.imdecode(x, cv2.IMREAD_UNCHANGED) if __name__ == "__main__": get_dimension()在 Erlang 和 Python 之间传输二进制数据
Erlang 的binary()数据类型和 Python 之间的映射关系, 在Python 2.x 中二进制数据类型为 str() 表示, Python 3.x 中为 bytes()
buf = resp.read(), 其中变量 buf 的类型为
在 Elixir 我们看到变量 buf 的值为:
{:ok, <<255, 216, 255, 224, 0, 16, 74, 70, 73, 70, 0, 1, 1, 1, 0, 72, 0, 72, 0, 0, 255, 219, 0, 67, 0, 8, 6, 6, 7, 6, 5, 8, 7, 7, 7, 9, 9, 8, 10, 12, 20, 13, 12, 11, 11, 12, 25, 18, 19, ...>>}调用 Python 函数
{:ok, data} = OpencvThumbnailServer.Api.load_image_url("https://segmentfault.com/img/bVwhAW") OpencvThumbnailServer.Api.get_dimension(data) {:ok, {800, 431}}创建 Python 模块
之前的 Python 图像处理模块可以组织到一个项目中多带带维护. 这里使用工具 cookiecutter 创建 Python 一个基本的项目骨架, 用于实现缩略图的功能
cookiecutter 可以通过多种方式安装, 包括pip, easy_install, conda, brew
pip install cookiecutter easy_install cookiecutter conda install -c https://conda.binstar.org/pydanny cookiecutter brew install cookiecutter(Mac OS X)
目录结构
➜ opencv_thumbnail git:(master) tree . ├── AUTHORS.rst ├── CONTRIBUTING.rst ├── HISTORY.rst ├── LICENSE ├── MANIFEST.in ├── Makefile ├── README.rst ├── build │ ├── bdist.macosx-10.11-x86_64 │ └── lib │ └── opencv_thumbnail │ ├── __init__.py │ └── opencv_thumbnail.py ├── dist │ └── opencv_thumbnail-0.1.0-py2.7.egg ├── docs │ ├── Makefile │ ├── authors.rst │ ├── conf.py │ ├── contributing.rst │ ├── history.rst │ ├── index.rst │ ├── installation.rst │ ├── make.bat │ ├── readme.rst │ └── usage.rst ├── opencv_thumbnail │ ├── __init__.py │ ├── __init__.pyc │ ├── opencv_thumbnail.py │ └── opencv_thumbnail.pyc ├── opencv_thumbnail.egg-info │ ├── PKG-INFO │ ├── SOURCES.txt │ ├── dependency_links.txt │ ├── not-zip-safe │ └── top_level.txt ├── requirements_dev.txt ├── setup.cfg ├── setup.py ├── tests │ ├── __init__.py │ └── test_opencv_thumbnail.py ├── tox.ini └── travis_pypi_setup.py 9 directories, 36 filesAPI实现
调用需要从 Poolboy 池中取出一个工作进程, 并调用工作进程的call_python, 进程使用完成后返还给 Poolboy 进程池, 这里对调用过程封装一下, 以简化使用.
defmodule OpencvThumbnailServer.Api do alias OpencvThumbnailServer.Worker def get_dimension(data) do worker = :poolboy.checkout(:opencv_thumbnail_server_pool) {w, h} = Worker.call_python(worker, :get_dimension, [data]) :poolboy.checkin(:opencv_thumbnail_server_pool, worker) {w, h} end def load_image_url(url) do worker = :poolboy.checkout(:opencv_thumbnail_server_pool) image_bin = Worker.call_python(worker, :load_image_url, [url]) :poolboy.checkin(:opencv_thumbnail_server_pool, worker) image_bin end end源码
https://github.com/developerworks/opencv_thumbnail_server
参考资料利用Python和OpenCV将URL直接转换成OpenCV格式
How to read raw png from an array in python opencv?
Install OpenCV for Python on Mac OS X
Installing scikit-image
How can i read an image from an internet url in python cv2 , scikit image and mahotas
Using Elixir, erlport with Python 2.7.9, receiving an arity error
How to read image from in memory buffer (StringIO) or from url with opencv python library
Python OpenCV convert image to byte string?
文章版权归作者所有,未经允许请勿转载,若此文章存在违规行为,您可以联系管理员删除。
转载请注明本文地址:https://www.ucloud.cn/yun/37987.html
摘要:图片中有人脸,将自动视为人脸区域为重要区域,将不会被裁剪掉。自动识别其它重要区域。如果图片中未识别出人脸,则会根据特征分布计算出重区域。源码地址安装根据上的说明,在上安装没有问题,但在我的上就挂了。首先遇到的问题就是,安装不过去。 Tclip 用于图片裁剪,有以下特点: 能进行人脸识别。图片中有人脸,将自动视为人脸区域为重要区域,将不会被裁剪掉。 自动识别其它重要区域。如果图片中...
摘要:图像指纹与汉明距离在介绍下面其他判别相似度的方法前,先补充一些概念。汉明距离为,即代表两张图片完全一样。下一次将讲述利用和以训练好的模型来进行人脸识别。本文参考文章和图片来源的文章赖勇浩的文章下一篇地址利用进行识别相似图片二 文章简介 在网上看到python做图像识别的相关文章后,真心感觉python的功能实在太强大,因此将这些文章总结一下,建立一下自己的知识体系。当然了,图像识别这个...
阅读 1698·2023-04-26 01:02
阅读 4838·2021-11-24 09:39
阅读 1803·2019-08-30 15:44
阅读 2872·2019-08-30 11:10
阅读 1782·2019-08-30 10:49
阅读 983·2019-08-29 17:06
阅读 608·2019-08-29 16:15
阅读 902·2019-08-29 15:17