OpenCV 是开源计算机视觉库，提供图像处理、特征检测、视频分析等功能，支持 C++、Python、Java。
安装
# 基础安装
pip install opencv-python
# 完整版（含视频处理、GUI）
pip install opencv-contrib-python常用示例
import cv2
# 图像读取与灰度转换
img = cv2.imread('image.jpg')
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.imwrite('gray_image.jpg', gray)
# ORB 特征检测
orb = cv2.ORB_create()
keypoints = orb.detect(gray, None)
img_keypoints = cv2.drawKeypoints(gray, keypoints, None, color=(0, 255, 0))
# Haar 人脸检测
face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
faces = face_cascade.detectMultiScale(gray, 1.3, 5)
for (x, y, w, h) in faces:
cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2)
# 视频读取
cap = cv2.VideoCapture('video.mp4')
while cap.isOpened():
ret, frame = cap.read()
if not ret:
break
cv2.imshow('Frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()CUDA 加速编译
从源码编译启用 GPU 加速：
# 安装依赖
apt-get install build-essential cmake git pkg-config
apt-get install libjpeg-dev libtiff-dev libpng-dev
apt-get install libavcodec-dev libavformat-dev libswscale-dev
apt-get install libgtk-3-dev libatlas-base-dev gfortran
apt-get install python3-dev libcuda1-11-6 libnpp-dev nvidia-cuda-toolkit
# 编译
git clone https://github.com/opencv/opencv.git
git clone https://github.com/opencv/opencv_contrib.git
cd opencv && mkdir build && cd build
cmake -D CMAKE_BUILD_TYPE=Release \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D OPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules \
-D WITH_CUDA=ON \
-D CUDA_ARCH_BIN=9.0 ..
make -j8 && make install && ldconfigCUDA 架构对照：
| GPU | 架构 | Compute Capability |
|---|---|---|
| H800 | Hopper | 9.0 |
| A100 | Ampere | 8.0 |
| V100 | Volta | 7.0 |
验证 CUDA 支持：
import cv2
print(cv2.getBuildInformation()) # 确认 WITH_CUDA=YES