基于opencv的车牌提取项目

初学图像处理，做了一个车牌提取项目，本博客仅仅是为了记录一下学习过程，该项目只具备初级功能，还有待改善

第一部分：车牌倾斜矫正

# 导入所需模块
import cv2
import math
from matplotlib import pyplot as plt

# 显示图片
def cv_show(name,img):
    cv2.imshow(name,img)
    cv2.waitKey()
    cv2.destroyAllWindows()

# 调整图片大小
def resize(image, width=None, height=None, inter=cv2.INTER_AREA):
    dim = None
    (h, w) = image.shape[:2]
    if width is None and height is None:
        return image
    if width is None:
        r = height / float(h)
        dim = (int(w * r), height)
    else:
        r = width / float(w)
        dim = (width, int(h * r))
    resized = cv2.resize(image, dim, interpolation=inter)
    return resized

# 加载图片
origin_Image = cv2.imread("./images/car_09.jpg")
rawImage = resize(origin_Image,height=500)

# 高斯去噪
image = cv2.GaussianBlur(rawImage, (3, 3), 0)
# 灰度处理
gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
# sobel算子边缘检测（做了一个y方向的检测）
Sobel_x = cv2.Sobel(gray_image, cv2.CV_16S, 1, 0)
absX = cv2.convertScaleAbs(Sobel_x)  # 转回uint8
image = absX
# 自适应阈值处理
ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_OTSU)
# 闭运算,是白色部分练成整体
kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (14, 5))
image = cv2.morphologyEx(image, cv2.MORPH_CLOSE, kernelX,iterations = 1)
# 去除一些小的白点
kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (20, 1))
kernelY = cv2.getStructuringElement(cv2.MORPH_RECT, (1, 19))
# 膨胀，腐蚀
image = cv2.dilate(image, kernelX)
image = cv2.erode(image, kernelX)
# 腐蚀，膨胀
image = cv2.erode(image, kernelY)
image = cv2.dilate(image, kernelY)
# 中值滤波去除噪点
image = cv2.medianBlur(image, 15)
# 轮廓检测
# cv2.RETR_EXTERNAL表示只检测外轮廓
# cv2.CHAIN_APPROX_SIMPLE压缩水平方向，垂直方向，对角线方向的元素，只保留该方向的终点坐标，例如一个矩形轮廓只需4个点来保存轮廓信息
thresh_, contours, hierarchy = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
# 绘制轮廓
image1 = rawImage.copy()
cv2.drawContours(image1, contours, -1, (0, 255, 0), 5)
cv_show('image1',image1)

# 筛选出车牌位置的轮廓
# 这里我只做了一个车牌的长宽比在3:1到4：1之间这样一个判断
for i,item in enumerate(contours): # enumerate() 函数用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列，同时列出数据和数据下标，一般用在for循环当中
    # cv2.boundingRect用一个最小的矩形，把找到的形状包起来
    rect = cv2.boundingRect(item)
    x = rect[0]
    y = rect[1]
    weight = rect[2]
    height = rect[3]
    if (weight > (height * 1.5)) and (weight < (height * 4)) and height>50:
        index = i
        image2 = rawImage.copy()
        cv2.drawContours(image2, contours, index, (0, 0, 255), 3)
        cv_show('image2',image2)
# 参数:
#  https://blog.csdn.net/lovetaozibaby/article/details/99482973
# InputArray Points: 待拟合的直线的集合，必须是矩阵形式；
# distType: 距离类型。fitline为距离最小化函数，拟合直线时，要使输入点到拟合直线的距离和最小化。这里的 距离的类型有以下几种：
# cv2.DIST_USER : User defined distance
# cv2.DIST_L1: distance = |x1-x2| + |y1-y2|
# cv2.DIST_L2: 欧式距离，此时与最小二乘法相同
# cv2.DIST_C:distance = max(|x1-x2|,|y1-y2|)
# cv2.DIST_L12:L1-L2 metric: distance = 2(sqrt(1+x*x/2) - 1))
# cv2.DIST_FAIR:distance = c^2(|x|/c-log(1+|x|/c)), c = 1.3998
# cv2.DIST_WELSCH: distance = c2/2(1-exp(-(x/c)2)), c = 2.9846
# cv2.DIST_HUBER:distance = |x|<c ? x^2/2 : c(|x|-c/2), c=1.345
# param： 距离参数，跟所选的距离类型有关，值可以设置为0。
#
# reps, aeps： 第5/6个参数用于表示拟合直线所需要的径向和角度精度，通常情况下两个值均被设定为1e-2.
# output ：
#
# 对于二维直线，输出output为4维，前两维代表拟合出的直线的方向，后两位代表直线上的一点。（即通常说的点斜式直线）
# 其中(vx, vy) 是直线的方向向量，(x, y) 是直线上的一个点。
# 斜率k = vy / vx
# 截距b = y - k * x

# 直线拟合找斜率
cnt = contours[index]
image3 = rawImage.copy()
h, w = image3.shape[:2]
[vx, vy, x, y] = cv2.fitLine(cnt, cv2.DIST_L2, 0, 0.01, 0.01)
k = vy/vx
b = y-k*x
lefty = b
righty = k*w+b
img = cv2.line(image3, (w, righty), (0, lefty), (0, 255, 0), 2)
cv_show('img',img)

a = math.atan(k)
a = math.degrees(a)
image4 = origin_Image.copy()
# 图像旋转
h,w = image4.shape[:2]
print(h,w)
#第一个参数旋转中心，第二个参数旋转角度，第三个参数：缩放比例
M = cv2.getRotationMatrix2D((w/2,h/2),a,1)
#第三个参数：变换后的图像大小
dst = cv2.warpAffine(image4,M,(int(w*1),int(h*1)))
cv_show('dst',dst)
cv2.imwrite('car_09.jpg',dst)

第二部分：车牌号码提取

 # 导入所需模块
 import cv2
 from matplotlib import pyplot as plt
 import os
 import numpy as np

 # 显示图片
 def cv_show(name,img):
     cv2.imshow(name,img)
     cv2.waitKey()
     cv2.destroyAllWindows()

 def resize(image, width=None, height=None, inter=cv2.INTER_AREA):
     dim = None
     (h, w) = image.shape[:2]
     if width is None and height is None:
         return image
     if width is None:
         r = height / float(h)
         dim = (int(w * r), height)
     else:
         r = width / float(w)
         dim = (width, int(h * r))
     resized = cv2.resize(image, dim, interpolation=inter)
     return resized

 #读取待检测图片
 origin_image = cv2.imread('./car_09.jpg')
 resize_image = resize(origin_image,height=600)
 ratio = origin_image.shape[0]/600
 print(ratio)
 cv_show('resize_image',resize_image)

 #高斯滤波，灰度化
 image = cv2.GaussianBlur(resize_image, (3, 3), 0)
 gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 cv_show('gray_image',gray_image)

 #梯度化
 Sobel_x = cv2.Sobel(gray_image, cv2.CV_16S, 1, 0)
 absX = cv2.convertScaleAbs(Sobel_x)
 image = absX
 cv_show('image',image)

 #闭操作
 ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_OTSU)
 kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (15, 1))
 image = cv2.morphologyEx(image, cv2.MORPH_CLOSE, kernelX, iterations=2)
 cv_show('image',image)
 kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (15, 1))
 kernelY = cv2.getStructuringElement(cv2.MORPH_RECT, (1, 1))
 image = cv2.dilate(image, kernelX)
 image = cv2.erode(image, kernelX)
 image = cv2.erode(image, kernelY)
 image = cv2.dilate(image, kernelY)
 image = cv2.medianBlur(image, 9)
 cv_show('image',image)

 #绘制轮廓
 thresh, contours, hierarchy = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
 print(type(contours))
 print(len(contours))

 cur_img = resize_image.copy()
 cv2.drawContours(cur_img,contours,-1,(0,0,255),3)
 cv_show('img',cur_img)

 for item in contours:
     rect = cv2.boundingRect(item)
     x = rect[0]
     y = rect[1]
     weight = rect[2]
     height = rect[3]
     if (weight > (height * 2.5)) and (weight < (height * 4)):
         if height > 40 and height < 80:
             image = origin_image[int(y*ratio): int((y + height)*ratio), int(x*ratio) : int((x + weight)*ratio)]
 cv_show('image',image)
 cv2.imwrite('chepai_09.jpg',image)

第三部分：车牌号码分割：

 # 导入所需模块
 import cv2
 from matplotlib import pyplot as plt
 import os
 import numpy as np

 # 显示图片
 def cv_show(name,img):
     cv2.imshow(name,img)
     cv2.waitKey()
     cv2.destroyAllWindows()

 #车牌灰度化
 chepai_image = cv2.imread('chepai_09.jpg')
 image = cv2.GaussianBlur(chepai_image, (3, 3), 0)
 gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 cv_show('gray_image',gray_image)

 ret, image = cv2.threshold(gray_image, 0, 255, cv2.THRESH_OTSU)
 cv_show('image',image)

 #计算二值图像黑白点的个数，处理绿牌照问题，让车牌号码始终为白色
 area_white = 0
 area_black = 0
 height, width = image.shape
 for i in range(height):
     for j in range(width):
         if image[i, j] == 255:
             area_white += 1
         else:
             area_black += 1
 print(area_black,area_white)
 if area_white > area_black:
     ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_OTSU | cv2.THRESH_BINARY_INV)
     cv_show('image',image)

 #绘制轮廓
 kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (20, 5))
 image = cv2.dilate(image, kernel)
 cv_show('image', image)
 thresh, contours, hierarchy = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
 cur_img = chepai_image.copy()
 cv2.drawContours(cur_img,contours,-1,(0,0,255),3)
 cv_show('img',cur_img)
 words = []
 word_images = []
 print(len(contours))
 for item in contours:
     word = []
     rect = cv2.boundingRect(item)
     x = rect[0]
     y = rect[1]
     weight = rect[2]
     height = rect[3]
     word.append(x)
     word.append(y)
     word.append(weight)
     word.append(height)
     words.append(word)
 words = sorted(words, key=lambda s:s[0], reverse=False)
 print(words)
 i = 0
 for word in words:
     if (word[3] > (word[2] * 1)) and (word[3] < (word[2] * 5)):
         i = i + 1
         splite_image = chepai_image[word[1]:word[1] + word[3], word[0]:word[0] + word[2]]
         word_images.append(splite_image)

 for i, j in enumerate(word_images):
     cv_show('word_images[i]',word_images[i])
     cv2.imwrite("./chepai_09/0{}.png".format(i),word_images[i])

第四部分：字符匹配：

 # 导入所需模块
 import cv2
 from matplotlib import pyplot as plt
 import os
 import numpy as np

 # 准备模板
 template = ['','','','','','','','','','',
             'A','B','C','D','E','F','G','H','J','K','L','M','N','P','Q','R','S','T','U','V','W','X','Y','Z',
             '藏','川','鄂','甘','赣','贵','桂','黑','沪','吉','冀','津','晋','京','辽','鲁','蒙','闽','宁',
             '青','琼','陕','苏','皖','湘','新','渝','豫','粤','云','浙']

 # 显示图片
 def cv_show(name,img):
     cv2.imshow(name,img)
     cv2.waitKey()
     cv2.destroyAllWindows()

 # 读取一个文件夹下的所有图片，输入参数是文件名，返回文件地址列表
 def read_directory(directory_name):
     referImg_list = []
     for filename in os.listdir(directory_name):
         referImg_list.append(directory_name + "/" + filename)
     return referImg_list

 # 中文模板列表（只匹配车牌的第一个字符）
 def get_chinese_words_list():
     chinese_words_list = []
     for i in range(34,64):
         c_word = read_directory('./refer1/'+ template[i])
         chinese_words_list.append(c_word)
     return chinese_words_list

 #英文模板列表(只匹配车牌的第二个字符)
 def get_english_words_list():
     eng_words_list = []
     for i in range(10,34):
         e_word = read_directory('./refer1/'+ template[i])
         eng_words_list.append(e_word)
     return eng_words_list

 # 英文数字模板列表（匹配车牌后面的字符）
 def get_eng_num_words_list():
     eng_num_words_list = []
     for i in range(0,34):
         word = read_directory('./refer1/'+ template[i])
         eng_num_words_list.append(word)
     return eng_num_words_list

 #模版匹配
 def template_matching(words_list):
     if words_list == 'chinese_words_list':
         template_words_list = chinese_words_list
         first_num = 34
     elif words_list == 'english_words_list':
         template_words_list = english_words_list
         first_num = 10
     else:
         template_words_list = eng_num_words_list
         first_num = 0
     best_score = []
     for template_word in template_words_list:
         score = []
         for word in template_word:
             template_img = cv2.imdecode(np.fromfile(word, dtype=np.uint8), 1)
             template_img = cv2.cvtColor(template_img, cv2.COLOR_RGB2GRAY)
             ret, template_img = cv2.threshold(template_img, 0, 255, cv2.THRESH_OTSU)
             height, width = template_img.shape
             image = image_.copy()
             image = cv2.resize(image, (width, height))
             result = cv2.matchTemplate(image, template_img, cv2.TM_CCOEFF)
             score.append(result[0][0])
         best_score.append(max(score))
     return template[first_num + best_score.index(max(best_score))]

 referImg_list = read_directory("chepai_13")
 chinese_words_list = get_chinese_words_list()
 english_words_list = get_english_words_list()
 eng_num_words_list = get_eng_num_words_list()
 chepai_num = []

 #匹配第一个汉字
 img = cv2.imread(referImg_list[0])
 # 高斯去噪
 image = cv2.GaussianBlur(img, (3, 3), 0)
 # 灰度处理
 gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 # 自适应阈值处理
 ret, image_ = cv2.threshold(gray_image, 0, 255, cv2.THRESH_OTSU)
 #第一个汉字匹配
 chepai_num.append(template_matching('chinese_words_list'))
 print(chepai_num[0])

 #匹配第二个英文字母
 img = cv2.imread(referImg_list[1])
 # 高斯去噪
 image = cv2.GaussianBlur(img, (3, 3), 0)
 # 灰度处理
 gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 # 自适应阈值处理
 ret, image_ = cv2.threshold(gray_image, 0, 255, cv2.THRESH_OTSU)
 #第二个英文字母匹配
 chepai_num.append(template_matching('english_words_list'))
 print(chepai_num[1])

 #匹配其余5个字母，数字
 for i in range(2,7):
     img = cv2.imread(referImg_list[i])
     # 高斯去噪
     image = cv2.GaussianBlur(img, (3, 3), 0)
     # 灰度处理
     gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
     # 自适应阈值处理
     ret, image_ = cv2.threshold(gray_image, 0, 255, cv2.THRESH_OTSU)
     #其余字母匹配
     chepai_num.append(template_matching('eng_num_words_list'))
     print(chepai_num[i])
 print(chepai_num)