首先附上本次识别的图片:(图片是我在百度上找的)
基于OpenCV车牌号识别总体分为四个步骤:
(1)提取车牌位置,将车牌从图中分割出来;
(2)车牌字符的分割;
(3)通过模版匹配识别字符;
(4)将结果绘制在图片上显示出来。
与深度学习相比,传统图像处理的识别有好处又有坏处:
好处:不需要大量的数据集训练模型,通过形态学、边缘检测等操作提取特征
坏处:基于传统图像处理的图像识别代码的泛化性较低,当图像的角度,光照不同时,识别效果有时会不尽人意。
为了方便观察每一步图片的变化,本次代码在Jupyter Notebook上编写,全部代码以上传(可直接运行)。
本次项目中会多次使用到图片显示和图片去噪灰度处理,所以首先定义了显示函数和高斯滤波灰度处理函数,方便后面的调用:
- # 导入所需模块
- import cv2
- from matplotlib import pyplot as plt
- import os
- import numpy as np
- # plt显示彩色图片
- def plt_show0(img):
- #cv2与plt的图像通道不同:cv2为[b,g,r];plt为[r, g, b]
- b,g,r = cv2.split(img)
- img = cv2.merge([r, g, b])
- plt.imshow(img)
- plt.show()
-
- # plt显示灰度图片
- def plt_show(img):
- plt.imshow(img,cmap='gray')
- plt.show()
-
- # 图像去噪灰度处理
- def gray_guss(image):
- image = cv2.GaussianBlur(image, (3, 3), 0)
- gray_image = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
- return gray_image
-
对图片进行阈值化处理、边缘检测及形态学操作,根据得到的轮廓特征识别车牌的具体位置,将车牌分割出来。直接上代码及代码详解:
- # 读取待检测图片
- origin_image = cv2.imread('./image/car.jpg')
- # 复制一张图片,在复制图上进行图像操作,保留原图
- image = origin_image.copy()
- # 图像去噪灰度处理
- gray_image = gray_guss(image)
- # x方向上的边缘检测(增强边缘信息)
- Sobel_x = cv2.Sobel(gray_image, cv2.CV_16S, 1, 0)
- absX = cv2.convertScaleAbs(Sobel_x)
- image = absX
-
- # 图像阈值化操作——获得二值化图
- ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_OTSU)
- # 显示灰度图像
- plt_show(image)
- # 形态学(从图像中提取对表达和描绘区域形状有意义的图像分量)——闭操作
- kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (30, 10))
- image = cv2.morphologyEx(image, cv2.MORPH_CLOSE, kernelX,iterations = 1)
- # 显示灰度图像
- plt_show(image)
-
二值化图以及闭操作(闭合细小的连接,抑制暗细节)的结果如图所示:
- # 腐蚀(erode)和膨胀(dilate)
- kernelX = cv2.getStructuringElement(cv2.MORPH_RECT, (50, 1))
- kernelY = cv2.getStructuringElement(cv2.MORPH_RECT, (1, 20))
- #x方向进行闭操作(抑制暗细节)
- image = cv2.dilate(image, kernelX)
- image = cv2.erode(image, kernelX)
- #y方向的开操作
- image = cv2.erode(image, kernelY)
- image = cv2.dilate(image, kernelY)
- # 中值滤波(去噪)
- image = cv2.medianBlur(image, 21)
- # 显示灰度图像
- plt_show(image)
- # 获得轮廓
- contours, hierarchy = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
-
- for item in contours:
- rect = cv2.boundingRect(item)
- x = rect[0]
- y = rect[1]
- weight = rect[2]
- height = rect[3]
- # 根据轮廓的形状特点,确定车牌的轮廓位置并截取图像
- if (weight > (height * 3.5)) and (weight < (height * 4)):
- image = origin_image[y:y + height, x:x + weight]
- plt_show0(image)
-
- #车牌字符分割
- # 图像去噪灰度处理
- gray_image = gray_guss(image)
- # 图像阈值化操作——获得二值化图
- ret, image = cv2.threshold(gray_image, 0, 255, cv2.THRESH_OTSU)
- plt_show(image)
-
- #膨胀操作,使“苏”字膨胀为一个近似的整体,为分割做准备
- kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (2, 2))
- image = cv2.dilate(image, kernel)
- plt_show(image)
-
- # 查找轮廓
- contours, hierarchy = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
- words = []
- word_images = []
- #对所有轮廓逐一操作
- 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)
- # 排序,车牌号有顺序。words是一个嵌套列表
- words = sorted(words,key=lambda s:s[0],reverse=False)
- i = 0
- #word中存放轮廓的起始点和宽高
- for word in words:
- # 筛选字符的轮廓
- if (word[3] > (word[2] * 1.5)) and (word[3] < (word[2] * 3.5)) and (word[2] > 25):
- i = i+1
- splite_image = image[word[1]:word[1] + word[3], word[0]:word[0] + word[2]]
- word_images.append(splite_image)
- print(i)
- print(words)
-
- for i,j in enumerate(word_images):
- plt.subplot(1,7,i+1)
- plt.imshow(word_images[i],cmap='gray')
- plt.show()
-
模板匹配是一个机械性的流程,所以把机械性的操作设定为函数。
- #模版匹配
- # 准备模板(template[0-9]为数字模板;)
- template = ['0','1','2','3','4','5','6','7','8','9',
- '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']
-
- # 读取一个文件夹下的所有图片,输入参数是文件名,返回模板文件地址列表
- def read_directory(directory_name):
- referImg_list = []
- for filename in os.listdir(directory_name):
- referImg_list.append(directory_name + "/" + filename)
- return referImg_list
-
- # 获得中文模板列表(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):
- #将模板存放在字典中
- c_word = read_directory('./refer1/'+ template[i])
- chinese_words_list.append(c_word)
- return chinese_words_list
- chinese_words_list = get_chinese_words_list()
-
-
- # 获得英文模板列表(只匹配车牌的第二个字符)
- def get_eng_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
- eng_words_list = get_eng_words_list()
-
-
- # 获得英文和数字模板列表(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
- eng_num_words_list = get_eng_num_words_list()
-
-
- # 读取一个模板地址与图片进行匹配,返回得分
- def template_score(template,image):
- #将模板进行格式转换
- template_img=cv2.imdecode(np.fromfile(template,dtype=np.uint8),1)
- template_img = cv2.cvtColor(template_img, cv2.COLOR_RGB2GRAY)
- #模板图像阈值化处理——获得黑白图
- 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))
- image_ = image.copy()
- #获得待检测图片的尺寸
- height, width = image_.shape
- # 将模板resize至与图像一样大小
- template_img = cv2.resize(template_img, (width, height))
- # 模板匹配,返回匹配得分
- result = cv2.matchTemplate(image_, template_img, cv2.TM_CCOEFF)
- return result[0][0]
-
-
- # 对分割得到的字符逐一匹配
- def template_matching(word_images):
- results = []
- for index,word_image in enumerate(word_images):
- if index==0:
- best_score = []
- for chinese_words in chinese_words_list:
- score = []
- for chinese_word in chinese_words:
- result = template_score(chinese_word,word_image)
- score.append(result)
- best_score.append(max(score))
- i = best_score.index(max(best_score))
- # print(template[34+i])
- r = template[34+i]
- results.append(r)
- continue
- if index==1:
- best_score = []
- for eng_word_list in eng_words_list:
- score = []
- for eng_word in eng_word_list:
- result = template_score(eng_word,word_image)
- score.append(result)
- best_score.append(max(score))
- i = best_score.index(max(best_score))
- # print(template[10+i])
- r = template[10+i]
- results.append(r)
- continue
- else:
- best_score = []
- for eng_num_word_list in eng_num_words_list:
- score = []
- for eng_num_word in eng_num_word_list:
- result = template_score(eng_num_word,word_image)
- score.append(result)
- best_score.append(max(score))
- i = best_score.index(max(best_score))
- # print(template[i])
- r = template[i]
- results.append(r)
- continue
- return results
-
-
- word_images_ = word_images.copy()
- # 调用函数获得结果
- result = template_matching(word_images_)
- print(result)
- # "".join(result)函数将列表转换为拼接好的字符串,方便结果显示
- print( "".join(result))
-
- Output:
- ['苏', 'E', '0', '5', 'E', 'V', '8']
- 苏E05EV8
-
最后,利用PIL库将结果绘制在原图上,获得的最终输出图片如下:
- from PIL import ImageFont, ImageDraw, Image
-
- height,weight = origin_image.shape[0:2]
- print(height)
- print(weight)
-
- image_1 = origin_image.copy()
- cv2.rectangle(image_1, (int(0.2*weight), int(0.75*height)), (int(weight*0.9), int(height*0.95)), (0, 255, 0), 5)
-
- #设置需要显示的字体
- fontpath = "font/simsun.ttc"
- font = ImageFont.truetype(fontpath,64)
- img_pil = Image.fromarray(image_1)
- draw = ImageDraw.Draw(img_pil)
- #绘制文字信息
- draw.text((int(0.2*weight)+25, int(0.75*height)), "".join(result), font = font, fill = (255, 255, 0))
- bk_img = np.array(img_pil)
- print(result)
- print( "".join(result))
- plt_show0(bk_img)
-
大功告成!!!!!
(一) 、OpenCV的车牌号码识别一共分为四步走:
1--提取车牌位置,将车牌从图中分割出来;
2--车牌字符的分割;
3--通过模版匹配识别字符;
4--将结果绘制在图片上显示出来。
(二)、图像处理的识别泛化性较低,对图片的角度光照有要求,所以要理解图像处理每一步的作用,根据自己图像的特点调整参数,更改操作顺序等等,以达到最好的效果。
(三)、车牌号识别的模板连接如下,需要的可以下载,有了模板就可以识别自己的图片了
链接:https://pan.baidu.com/s/1QBjy7c0klv_PBUwJjA8ynA
提取码:v53d