在msys2/mingw下安装及编译opencv

最简单就是直接安装

pacman -S mingw-w64-x86_64-opencv

以下记录一下编译的过程

1.  安装编译工具及第三方库

pacman -S --needed base-devel mingw-w64-x86_64-toolchain unzip gcc pacman -S python mingw-w64-x86_64-python2 mingw-w64-x86_64-gtk3 mingw-w64-x86_64-boost pacman -S mingw-w64-x86_64-ogre3d mingw-w64-x86_64-gtk3 mingw-w64-x86_64-vtk mingw-w64-x86_64-eigen3 mingw-w64-x86_64-ccache mingw-w64-x86_64-python-numpy mingw-w64-x86_64-julia mingw-w64-x86_64-lapack pacman -S unzip vim cmake gcc wget unzip protobuf pacman -S mingw-w64-x86_64-libpng mingw-w64-x86_64-libjpeg mingw-w64-x86_64-libtiff mingw-w64-x86_64-libwebp pacman -S mingw-w64-x86_64-ffmpeg mingw-w64-x86_64-harfbuzz mingw-w64-x86_64-openblas mingw-w64-x86_64-tesseract-ocr mingw-w64-x86_64-tesseract-data-chi_sim mingw-w64-x86_64-tesseract-data-chi_tra

下载 opencv4 源码并解压

wget -O opencv.zip github /opencv/opencv/archive/4.x.zip wget -O opencv_contrib.zip github /opencv/opencv_contrib/archive/4.x.zip unzip opencv.zip unzip opencv_contrib.zip

 编写脚本 build-open.sh

#!/bin/bash # mkdir -p build && cd build # Configure export PATH=/mingw64/lib:/mingw64/bin:/mingw64/include:$PATH cmake -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.x/modules ../opencv-4.x # Build cmake --build . -j 8

由于网络速度慢，有几个文件下载超时，我通过迅雷下载好后 

raw.githubusercontent /opencv/opencv_3rdparty/1224f78da6684df04397ac0f40c961ed37f79ccb/ippicv/ippicv_2021.8_lnx_intel64_20230330_general.tgz

raw.githubusercontent /opencv/opencv_3rdparty/1224f78da6684df04397ac0f40c961ed37f79ccb/ippicv/ippicv_2021.8_win_intel64_20230330_general.zip

github /opencv/ade/archive/v0.1.2b.zip

raw.githubusercontent /opencv/opencv_3rdparty/8afa57abc8229d611c4937165d20e2a2d9fc5a12/face_landmark_model.dat

 我通过迅雷下载后，复制到相应目录

cp ippicv_2021.8_lnx_intel64_20230330_general.tgz opencv-4.x/3rdparty/ippicv/ cp ippicv_2021.8_win_intel64_20230330_general.zip opencv-4.x/3rdparty/ippicv/ cp  -f ade-0.1.2b.zip  opencv-4.x/.cache/ade/4f93a0844dfc463c617d83b09011819a-v0.1.2b.zip cp face_landmark_model.dat opencv-4.x/.cache/data/7505c44ca4eb54b4ab1e4777cb96ac05-face_landmark_model.dat