Opencv 图片的OCR识别的实战示例_Python

一、图片变换

0、导入模块

导入相关函数，遇到报错的话，直接pip install 函数名。

				?

									import numpy as np

									import argparse

									import cv2

参数初始化

				?

									ap = argparse.ArgumentParser()

									ap.add_argument("-i", "--image", required = True,

									  help = "Path to the image to be scanned") 

									args = vars(ap.parse_args())

Parameters:

--image images\page.jpg

1、重写resize函数

				?

									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

2、预处理

读取图片后进行重置大小，并计算缩放倍数；进行灰度化、高斯滤波以及Canny轮廓提取

				?

									image = cv2.imread(args["image"])

									ratio = image.shape[0] / 500.0

									orig = image.copy()

									image = resize(orig, height = 500)

									gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

									gray = cv2.GaussianBlur(gray, (5, 5), 0)

									edged = cv2.Canny(gray, 75, 200)

3、边缘检测

检测轮廓并排序，遍历轮廓。

				?

									cnts = cv2.findContours(edged.copy(), cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)[0]# 轮廓检测

									cnts = sorted(cnts, key = cv2.contourArea, reverse = True)[:5]#保留前5个轮廓

									# 遍历轮廓

									for c in cnts:

									  # 计算轮廓近似

									  peri = cv2.arcLength(c, True)# 计算轮廓长度，C表示输入的点集，True表示轮廓是封闭的

									  #(C表示输入的点集，epslion判断点到相对应的line segment 的距离的阈值,曲线是否闭合的标志位)

									  approx = cv2.approxPolyDP(c, 0.02 * peri, True)

									  # 4个点的时候就拿出来

									  if len(approx) == 4:

									   screenCnt = approx

									   break

4、透视变换

画出近似轮廓，透视变换，二值处理

				?

									cv2.drawContours(image, [screenCnt], -1, (0, 255, 0), 2)

									warped = four_point_transform(orig, screenCnt.reshape(4, 2) * ratio)#透视变换

									# 二值处理

									warped = cv2.cvtColor(warped, cv2.COLOR_BGR2GRAY)

									ref = cv2.threshold(warped, 100, 255, cv2.THRESH_BINARY)[1]

									cv2.imwrite('scan.jpg', ref)

二、OCR识别

0、安装tesseract-ocr

链接: 下载

在环境变量、系统变量的Path里面添加安装路径，例如：E:\Program Files (x86)\Tesseract-OCR

				?

									tesseract -v#打开命令行，进行测试

									tesseract XXX.png result#得到结果 

									pip install pytesseract#安装依赖包

打开python安装路径里面的python文件，例如C:\ProgramData\Anaconda3\Lib\site-packages\pytesseract\pytesseract.py
将tesseract_cmd 修改为绝对路径即可，例如：tesseract_cmd = ‘C:/Program Files (x86)/Tesseract-OCR/tesseract.exe'

1、导入模块

				?

									from PIL import Image

									import pytesseract

									import cv2

									import os

2、预处理

读取图片、灰度化、滤波

				?

									image = cv2.imread('scan.jpg')

									gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

									gray = cv2.medianBlur(gray, 3)

3、输出结果

				?

									filename = "{}.png".format(os.getpid())

									cv2.imwrite(filename, gray)  

									text = pytesseract.image_to_string(Image.open(filename))

									print(text)

									os.remove(filename)