python 图片验证码降噪、识别

from PIL import Image
import pytesseract
import os

'''
安装 tesseract-ocr-setup-3.05.01.exe
下面的环境变量根据自己电脑实际情况配置
'''
os.environ['TESSDATA_PREFIX'] = 'F:\\safe-soft\\ocr\\Tesseract-OCR\\tessdata'
pytesseract.pytesseract.tesseract_cmd = 'F:\\safe-soft\\ocr\\Tesseract-OCR\\tesseract.exe'


# 取像素点中间值
def median(img, x, y):
    L = []
    xl = [x - 1, x, x + 1]
    yl = [y - 1, y, y + 1]
    for i in xl:
        for j in yl:
            gray = img.getpixel((i, j))  # 取出灰度值
            L.append(gray)
    L.sort()
    c = L[4]
    return c


# 对图片进行降噪处理
def denoise(path):
    img1 = Image.open(path)  # 图像1
    img1 = img1.convert('L')  # 将图像1转换为灰度图
    w, h = img1.size
    img2 = Image.new('L', (w, h), 'white')  # 图像2
    for x in range(1, w - 1):
        for y in range(1, h - 1):
            c = median(img1, x, y)  # 求中值
            img2.putpixel((x, y), c)  # 将灰度设置为中值
    return img2


# 解析图片验证码
def getYzm(imgPath):
    text = pytesseract.image_to_string(denoise(imgPath))
    return text


print(getYzm('D:/img-test/3.png'))