Mac/Linux使用hyperLPR+opencv从监控中进行车牌检测--含完整测试用例

一.需要使用的库

1.使用hyperLPR文件包所依赖的库

pip3 install opencv-python==3.4.6.27
pip3 install pyinstaller -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
pip3 install Keras
pip3 install Theano
pip3 install Numpy
pip3 install Scipy
pip3 install scikit-image
pip3 install pillow
pip3 install tensorflow

这几个库树莓派装的时候会报错，scipy装不上，hyper也可以正常工作。
2.hyperLPR

pip3 install hyperLPR

3.OpenCV 3.4.3并编译
https://github.com/opencv/opencv/releases/tag/3.4.3
下载源代码文件包source code(zip)

二.环境编译

1.编译opencv3.4.3
Mac版
进入解压后的opencv文件

cd opencv-3.4.3
mkdir build
cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..
make -j4
sudo make install

Linux版
先安装编译环境

sudo apt install build-essential
sudo apt install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
sudo apt install python-dev python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev
sudo apt install libqt4-test
sudo apt install libatlas-base-dev
sudo apt install libjasper-dev
sudo apt install libqtgui4
sudo apt install libhdf5-dev
sudo apt install libhdf5-serial-dev

编译opencv3.4.3

cd opencv-3.4.3
mkdir build
cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..
sudo make -j
sudo make install
sudo /bin/bash -c 'echo "/usr/local/lib" > /etc/ld.so.conf.d/opencv.conf' 
sudo ldconfig

2.hyperLPR
https://github.com/zeusees/HyperLPR
下载git源文件，然后进入文件并编译 mac/linux通用

cd hyperLPR
cd Prj-Linux
mkdir build 
cd build
cmake ../
sudo make -j

3.树莓派编译出错问题，增加交换空间内存

cd /opt
sudo mkdir image
cd image
sudo touch swap    #创建文件
sudo dd if=/dev/zero of=/opt/image/swap bs=1024 count=1024000  #添加交换文件并设置为1G
#过段时间会返回(这个略慢)
#1024000+0 records in
#1024000+0 records out
#大小 copied, 所用时间 s, 速度 MB/s
sudo mkswap /opt/image/swap  #设置交换空间

sudo swapon /opt/image/swap  #启用新增的交换空间
free -m    #确认是否已经生效
# 修改/etc/fstab文件
sudo nano /etc/fstab
/opt/image/swap    /swap    swap    defaults 0 0 

三.测试
1.使用pip3安装的hyperlpr库

from hyperlpr import *
import cv2


class CarNumRecognition(object):

    def __init__(self, image_path, video_path):
        self._image_path = image_path
        self._video_path = video_path

    def video(self):
        print('[INFO] starting video stream')
        stream = cv2.VideoCapture(self._video_path)
        while 1:
            grabbed, frame = stream.read()
            if not grabbed:
                print('NO DATA')
                break
            res = HyperLPR_plate_recognition(frame)
            print(res)
            key = cv2.waitKey(5) & 0xFF
            if key == ord('q'):
                break
            stream.release()
            cv2.destroyAllWindows()

    def image(self):
        print('[INFO] starting image')
        image = cv2.imread(self._image_path)
        res = HyperLPR_plate_recognition(image)
        print(res)


if __name__ == '__main__':

    _video = 'test.mp4'
    _image = '1.png'

    _test = CarNumRecognition(video_path=_video, image_path=_image)
    # _test.video()
    _test.image()

2.使用编译完的hyperlpr包
用pycharm打开hyperlpr文件
然后找到test_image文件，点击运行。

四.从监控视频中提取车牌号

from hyperlpr import *
import cv2


class CarNumRecognize(object):

    def __init__(self):
        pass

    def recognize(self):
        cap = cv2.VideoCapture(0) # 打开自身摄像头
        print(f'IP摄像头是否开启：{cap.isOpened()}')
        cap.set(cv2.CAP_PROP_BUFFERSIZE, 1)  # 设置buffer
        cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1920)  # 设置视频画面宽
        cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 1080)  # 设置视频画面高
        while 1:
            grabbed, frame = cap.read()
            if not grabbed:
                print('NO DATA')
                break
            res = HyperLPR_plate_recognition(frame)
            print(res)
            cv2.imshow('Window', frame)
            key = cv2.waitKey(2000) & 0xFF  #  检测时间
            if key == ord('q'):
                break
            cap.release()
            cv2.destroyAllWindows()


if __name__ == '__main__':
    _test = CarNumRecognize()
    _test.recognize()