OpenCV Part VII: License Plate Recognition

Table of contents

1. Adjust the image size and get the grayscale image

 2. Bilateral filtering to remove noise: cv2.bilateralFilter().

3. Edge detection: cv2.Canny (image, threshold1, threshold2)

4. Find the contour: license plate (quadrangle)

​Edit 5. Image bit operations for masking

6. Image cropping

7. Character recognition: OCR

---

1. Adjust the image size and get the grayscale image

import cv2

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)




    # 显示效果
    cv2.imshow('original', img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

 2. Bilateral filtering to remove noise: cv2.bilateralFilter().

import cv2

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray1 = cv2.bilateralFilter(gray, 13, 15, 15)

    # 显示效果
    cv2.imshow('gray', gray)
    cv2.imshow('bilateralFilter', gray1)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

3. Edge detection: cv2.Canny (image, threshold1, threshold2)

Only show edges whose intensity gradient is greater than the minimum threshold threshold1 and less than the maximum threshold threshold2.

import cv2

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray = cv2.bilateralFilter(gray, 13, 15, 15)
    # 边缘检测
    edged = cv2.Canny(gray, 30, 200)




    # 显示效果
    cv2.imshow('gray', gray)
    cv2.imshow('edged', edged)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

4. Find the contour: license plate (quadrangle)

pip install imutils

import cv2
import imutils

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray = cv2.bilateralFilter(gray, 13, 15, 15)
    # 边缘检测
    edged = cv2.Canny(gray, 30, 200)


    # 寻找轮廓（图像矩阵，输出模式，近似方法）
    contours = cv2.findContours(edged.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    # 配合上面一句使用：用来兼容cv2和cv3
    contours = imutils.grab_contours(contours)
    # 根据区域大小排序取前十个
    contours = sorted(contours, key=cv2.contourArea, reverse=True)[:10]
    screenCnt = None
    # 遍历轮廓，找到车牌轮廓
    for c in contours:
        # 计算轮廓周长（轮廓，是否闭合）
        peri = cv2.arcLength(c, True)
        # 折线化（轮廓，阈值（越小越接近曲线），是否闭合）返回折线顶点坐标
        approx = cv2.approxPolyDP(c, 0.018 * peri, True)
        # 获取四个顶点（即四边形）
        if len(approx) == 4:
            screenCnt = approx
            break
    # 如果找到了四边形
    if screenCnt is not None:
        # 根据四个顶点坐标对img画线(图像矩阵，轮廓坐标集，轮廓索引，颜色，线条粗细)
        cv2.drawContours(img, [screenCnt], -1, (0, 0, 255), 3)



    # 显示效果
    cv2.imshow('img', img)
    cv2.imshow('gray', gray)
    cv2.imshow('edged', edged)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

 5. Image bit operation for masking

import cv2
import imutils
import numpy as np

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray = cv2.bilateralFilter(gray, 13, 15, 15)
    # 边缘检测
    edged = cv2.Canny(gray, 30, 200)

    """寻找轮廓（图像矩阵，输出模式，近似方法）"""
    contours = cv2.findContours(edged.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    # 配合上面一句使用：用来兼容cv2和cv3
    contours = imutils.grab_contours(contours)
    # 根据区域大小排序取前十个
    contours = sorted(contours, key=cv2.contourArea, reverse=True)[:10]
    screenCnt = None
    # 遍历轮廓，找到车牌轮廓
    for c in contours:
        # 计算轮廓周长（轮廓，是否闭合）
        peri = cv2.arcLength(c, True)
        # 折线化（轮廓，阈值（越小越接近曲线），是否闭合）返回折线顶点坐标
        approx = cv2.approxPolyDP(c, 0.018 * peri, True)
        # 获取四个顶点（即四边形）
        if len(approx) == 4:
            screenCnt = approx
            break
    # 如果找到了四边形
    if screenCnt is not None:
        # 根据四个顶点坐标对img画线(图像矩阵，轮廓坐标集，轮廓索引，颜色，线条粗细)
        cv2.drawContours(img, [screenCnt], -1, (0, 0, 255), 3)

    """遮罩"""
    # 创建一个灰度图一样大小的图像矩阵
    mask = np.zeros(gray.shape, np.uint8)
    # 将创建的图像矩阵的车牌区域画成白色
    cv2.drawContours(mask, [screenCnt], 0, 255, -1, )
    # 图像位运算进行遮罩
    new_image = cv2.bitwise_and(img, img, mask=mask)


    # 显示效果
    cv2.imshow('img', img)
    cv2.imshow('gray', gray)
    cv2.imshow('edged', edged)
    cv2.imshow('new_image', new_image)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

6. Image cropping

import cv2
import imutils
import numpy as np

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray = cv2.bilateralFilter(gray, 13, 15, 15)
    # 边缘检测
    edged = cv2.Canny(gray, 30, 200)

    """寻找轮廓（图像矩阵，输出模式，近似方法）"""
    contours = cv2.findContours(edged.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    # 配合上面一句使用：用来兼容cv2和cv3
    contours = imutils.grab_contours(contours)
    # 根据区域大小排序取前十个
    contours = sorted(contours, key=cv2.contourArea, reverse=True)[:10]
    screenCnt = None
    # 遍历轮廓，找到车牌轮廓
    for c in contours:
        # 计算轮廓周长（轮廓，是否闭合）
        peri = cv2.arcLength(c, True)
        # 折线化（轮廓，阈值（越小越接近曲线），是否闭合）返回折线顶点坐标
        approx = cv2.approxPolyDP(c, 0.018 * peri, True)
        # 获取四个顶点（即四边形）
        if len(approx) == 4:
            screenCnt = approx
            break
    # 如果找到了四边形
    if screenCnt is not None:
        # 根据四个顶点坐标对img画线(图像矩阵，轮廓坐标集，轮廓索引，颜色，线条粗细)
        cv2.drawContours(img, [screenCnt], -1, (0, 0, 255), 3)

    """遮罩"""
    # 创建一个灰度图一样大小的图像矩阵
    mask = np.zeros(gray.shape, np.uint8)
    # 将创建的图像矩阵的车牌区域画成白色
    cv2.drawContours(mask, [screenCnt], 0, 255, -1, )
    # 图像位运算进行遮罩
    new_image = cv2.bitwise_and(img, img, mask=mask)

    """图像剪裁"""
    # 获取车牌区域的所有坐标点
    (x, y) = np.where(mask == 255)
    # 获取底部顶点坐标
    (topx, topy) = (np.min(x), np.min(y))
    # 获取底部坐标
    (bottomx, bottomy,) = (np.max(x), np.max(y))
    # 剪裁
    Cropped = gray[topx:bottomx, topy:bottomy]

    # 显示效果
    cv2.imshow('img', img)
    cv2.imshow('gray', gray)
    cv2.imshow('edged', edged)
    cv2.imshow('Cropped', Cropped)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

7. Character recognition: OCR

import cv2
import imutils
import numpy as np

if __name__ == '__main__':
    img = cv2.imread('2.jpeg')
    # 调整图片大小
    img = cv2.resize(img, (620, 480))
    # 灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 双边滤波
    gray = cv2.bilateralFilter(gray, 13, 15, 15)
    # 边缘检测
    edged = cv2.Canny(gray, 30, 200)

    """寻找轮廓（图像矩阵，输出模式，近似方法）"""
    contours = cv2.findContours(edged.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    # 配合上面一句使用：用来兼容cv2和cv3
    contours = imutils.grab_contours(contours)
    # 根据区域大小排序取前十个
    contours = sorted(contours, key=cv2.contourArea, reverse=True)[:10]
    screenCnt = None
    # 遍历轮廓，找到车牌轮廓
    for c in contours:
        # 计算轮廓周长（轮廓，是否闭合）
        peri = cv2.arcLength(c, True)
        # 折线化（轮廓，阈值（越小越接近曲线），是否闭合）返回折线顶点坐标
        approx = cv2.approxPolyDP(c, 0.018 * peri, True)
        # 获取四个顶点（即四边形）
        if len(approx) == 4:
            screenCnt = approx
            break
    # 如果找到了四边形
    if screenCnt is not None:
        # 根据四个顶点坐标对img画线(图像矩阵，轮廓坐标集，轮廓索引，颜色，线条粗细)
        cv2.drawContours(img, [screenCnt], -1, (0, 0, 255), 3)

    """遮罩"""
    # 创建一个灰度图一样大小的图像矩阵
    mask = np.zeros(gray.shape, np.uint8)
    # 将创建的图像矩阵的车牌区域画成白色
    cv2.drawContours(mask, [screenCnt], 0, 255, -1, )
    # 图像位运算进行遮罩
    new_image = cv2.bitwise_and(img, img, mask=mask)

    """图像剪裁"""
    # 获取车牌区域的所有坐标点
    (x, y) = np.where(mask == 255)
    # 获取底部顶点坐标
    (topx, topy) = (np.min(x), np.min(y))
    # 获取底部坐标
    (bottomx, bottomy,) = (np.max(x), np.max(y))
    # 剪裁
    Cropped = gray[topx:bottomx, topy:bottomy]

    """OCR识别"""
    text = pytesseract.image_to_string(Cropped, config='--psm 11')
    print("车牌结果：", text)

    # 显示效果
    cv2.imshow('img', img)
    cv2.imshow('gray', gray)
    cv2.imshow('edged', edged)
    cv2.imshow('new_image', Cropped)
    cv2.waitKey(0)
    cv2.destroyAllWindows()