用python编写一个验证码识别程序

下面是一个基于Python的验证码识别程序的示例：

1. 首先，需要导入所需的库，包括Pillow、tesseract和pytesseract。Pillow用于图像处理，tesseract和pytesseract用于OCR识别。

from PIL import Image
import pytesseract
import os

2. 然后，读取验证码图片并进行预处理，包括二值化、去除干扰线等操作。

def preprocess(image_path):
    image = Image.open(image_path)
    # 转为灰度图
    gray_image = image.convert('L')
    # 二值化
    threshold = 127
    bin_image = gray_image.point(lambda p: p > threshold and 255)
    # 去除干扰线
    for y in range(1, bin_image.height - 1):
        for x in range(1, bin_image.width - 1):
            if bin_image.getpixel((x, y)) == 0:
                if bin_image.getpixel((x - 1, y)) == 255 and bin_image.getpixel((x + 1, y)) == 255:
                    bin_image.putpixel((x, y), 255)
                elif bin_image.getpixel((x, y - 1)) == 255 and bin_image.getpixel((x, y + 1)) == 255:
                    bin_image.putpixel((x, y), 255)
    return bin_image

3. 接下来，使用pytesseract进行OCR识别。

def recognize(image_path):
    image = preprocess(image_path)
    code = pytesseract.image_to_string(image, lang='eng', config='--psm 7')
    return code

4. 最后，调用recognize函数进行识别，并输出识别结果。

if __name__ == '__main__':
    image_path = 'captcha.png'
    code = recognize(image_path)
    print('识别结果：', code)

完整代码如下：

from PIL import Image
import pytesseract
import os


def preprocess(image_path):
    image = Image.open(image_path)
    # 转为灰度图
    gray_image = image.convert('L')
    # 二值化
    threshold = 127
    bin_image = gray_image.point(lambda p: p > threshold and 255)
    # 去除干扰线
    for y in range(1, bin_image.height - 1):
        for x in range(1, bin_image.width - 1):
            if bin_image.getpixel((x, y)) == 0:
                if bin_image.getpixel((x - 1, y)) == 255 and bin_image.getpixel((x + 1, y)) == 255:
                    bin_image.putpixel((x, y), 255)
                elif bin_image.getpixel((x, y - 1)) == 255 and bin_image.getpixel((x, y + 1)) == 255:
                    bin_image.putpixel((x, y), 255)
    return bin_image


def recognize(image_path):
    image = preprocess(image_path)
    code = pytesseract.image_to_string(image, lang='eng', config='--psm 7')
    return code


if __name__ == '__main__':
    image_path = 'captcha.png'
    code = recognize(image_path)
    print('识别结果：', code)

注意：验证码识别程序的准确率受到多种因素的影响，包括验证码的清晰度、复杂度以及预处理和OCR识别算法的优化等。因此，实际应用中需要根据具体情况进行调试和优化。