python identification verification code

This time I will bring you python identificationverification code, what are the precautions for python identification verification code, the following is a practical case, let's take a look.

In addition to the traditional PIL package to process images, and then use pytessert OCR to identify accidents, you can also use tessorflow training to identify verification codes.

Most of the code in this article is reproduced, with only a few changes.

The code runs in a Linux environment, and tessorflow does not support python 2.7 for windows.

gen_captcha.py code.

#coding=utf-8
from captcha.image import ImageCaptcha # pip install captcha
import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import random
# 验证码中的字符, 就不用汉字了
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
alphabet = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u',
      'v', 'w', 'x', 'y', 'z']
ALPHABET = ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U',
      'V', 'W', 'X', 'Y', 'Z']
'''
number=['0','1','2','3','4','5','6','7','8','9']
alphabet =[]
ALPHABET =[]
'''
# 验证码一般都无视大小写；验证码长度4个字符
def random_captcha_text(char_set=number + alphabet + ALPHABET, captcha_size=4):
  captcha_text = []
  for i in range(captcha_size):
    c = random.choice(char_set)
    captcha_text.append(c)
  return captcha_text
# 生成字符对应的验证码
def gen_captcha_text_and_image():
  while(1):
    image = ImageCaptcha()
    captcha_text = random_captcha_text()
    captcha_text = ''.join(captcha_text)
    captcha = image.generate(captcha_text)
    #image.write(captcha_text, captcha_text + '.jpg') # 写到文件
    captcha_image = Image.open(captcha)
    #captcha_image.show()
    captcha_image = np.array(captcha_image)
    if captcha_image.shape==(60,160,3):
      break
  return captcha_text, captcha_image
if name == 'main':
  # 测试
  text, image = gen_captcha_text_and_image()
  print image
  gray = np.mean(image, -1)
  print gray
  print image.shape
  print gray.shape
  f = plt.figure()
  ax = f.add_subplot(111)
  ax.text(0.1, 0.9, text, ha='center', va='center', transform=ax.transAxes)
  plt.imshow(image)
  plt.show()

train.py code.

#coding=utf-8
from gen_captcha import gen_captcha_text_and_image
from gen_captcha import number
from gen_captcha import alphabet
from gen_captcha import ALPHABET
import numpy as np
import tensorflow as tf
"""
text, image = gen_captcha_text_and_image()
print "验证码图像channel:", image.shape # (60, 160, 3)
# 图像大小
IMAGE_HEIGHT = 60
IMAGE_WIDTH = 160
MAX_CAPTCHA = len(text)
print  "验证码文本最长字符数", MAX_CAPTCHA # 验证码最长4字符; 我全部固定为4,可以不固定. 如果验证码长度小于4，用'_'补齐
"""
IMAGE_HEIGHT = 60
IMAGE_WIDTH = 160
MAX_CAPTCHA = 4
# 把彩色图像转为灰度图像（色彩对识别验证码没有什么用）
def convert2gray(img):
  if len(img.shape) > 2:
    gray = np.mean(img, -1)
    # 上面的转法较快，正规转法如下
    # r, g, b = img[:,:,0], img[:,:,1], img[:,:,2]
    # gray = 0.2989 * r + 0.5870 * g + 0.1140 * b
    return gray
  else:
    return img
"""
cnn在图像大小是2的倍数时性能最高, 如果你用的图像大小不是2的倍数，可以在图像边缘补无用像素。
np.pad(image,((2,3),(2,2)), 'constant', constant_values=(255,)) # 在图像上补2行，下补3行，左补2行，右补2行
"""
# 文本转向量
char_set = number + alphabet + ALPHABET + ['_'] # 如果验证码长度小于4, '_'用来补齐
CHAR_SET_LEN = len(char_set)
def text2vec(text):
  text_len = len(text)
  if text_len > MAX_CAPTCHA:
    raise ValueError('验证码最长4个字符')
  vector = np.zeros(MAX_CAPTCHA * CHAR_SET_LEN)
  def char2pos(c):
    if c == '_':
      k = 62
      return k
    k = ord(c) - 48
    if k > 9:
      k = ord(c) - 55
      if k > 35:
        k = ord(c) - 61
        if k > 61:
          raise ValueError('No Map')
    return k
  for i, c in enumerate(text):
    #print text
    idx = i * CHAR_SET_LEN + char2pos(c)
    #print i,CHAR_SET_LEN,char2pos(c),idx
    vector[idx] = 1
  return vector
#print text2vec('1aZ_')
# 向量转回文本
def vec2text(vec):
  char_pos = vec.nonzero()[0]
  text = []
  for i, c in enumerate(char_pos):
    char_at_pos = i # c/63
    char_idx = c % CHAR_SET_LEN
    if char_idx < 10:
      char_code = char_idx + ord(&#39;0&#39;)
    elif char_idx < 36:
      char_code = char_idx - 10 + ord(&#39;A&#39;)
    elif char_idx < 62:
      char_code = char_idx - 36 + ord(&#39;a&#39;)
    elif char_idx == 62:
      char_code = ord(&#39;_&#39;)
    else:
      raise ValueError(&#39;error&#39;)
    text.append(chr(char_code))
  return "".join(text)
"""
#向量（大小MAX_CAPTCHA*CHAR_SET_LEN）用0,1编码 每63个编码一个字符，这样顺利有，字符也有
vec = text2vec("F5Sd")
text = vec2text(vec)
print(text) # F5Sd
vec = text2vec("SFd5")
text = vec2text(vec)
print(text) # SFd5
"""
# 生成一个训练batch
def get_next_batch(batch_size=128):
  batch_x = np.zeros([batch_size, IMAGE_HEIGHT * IMAGE_WIDTH])
  batch_y = np.zeros([batch_size, MAX_CAPTCHA * CHAR_SET_LEN])
  # 有时生成图像大小不是(60, 160, 3)
  def wrap_gen_captcha_text_and_image():
    while True:
      text, image = gen_captcha_text_and_image()
      if image.shape == (60, 160, 3):
        return text, image
  for i in range(batch_size):
    text, image = wrap_gen_captcha_text_and_image()
    image = convert2gray(image)
    batch_x[i, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0
    batch_y[i, :] = text2vec(text)
  return batch_x, batch_y
####################################################################
X = tf.placeholder(tf.float32, [None, IMAGE_HEIGHT * IMAGE_WIDTH])
Y = tf.placeholder(tf.float32, [None, MAX_CAPTCHA * CHAR_SET_LEN])
keep_prob = tf.placeholder(tf.float32) # dropout
# 定义CNN
def crack_captcha_cnn(w_alpha=0.01, b_alpha=0.1):
  x = tf.reshape(X, shape=[-1, IMAGE_HEIGHT, IMAGE_WIDTH, 1])
  # w_c1_alpha = np.sqrt(2.0/(IMAGE_HEIGHT*IMAGE_WIDTH)) #
  # w_c2_alpha = np.sqrt(2.0/(3*3*32))
  # w_c3_alpha = np.sqrt(2.0/(3*3*64))
  # w_d1_alpha = np.sqrt(2.0/(8*32*64))
  # out_alpha = np.sqrt(2.0/1024)
  # 3 conv layer
  w_c1 = tf.Variable(w_alpha * tf.random_normal([3, 3, 1, 32]))
  b_c1 = tf.Variable(b_alpha * tf.random_normal([32]))
  conv1 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(x, w_c1, strides=[1, 1, 1, 1], padding=&#39;SAME&#39;), b_c1))
  conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding=&#39;SAME&#39;)
  conv1 = tf.nn.dropout(conv1, keep_prob)
  w_c2 = tf.Variable(w_alpha * tf.random_normal([3, 3, 32, 64]))
  b_c2 = tf.Variable(b_alpha * tf.random_normal([64]))
  conv2 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv1, w_c2, strides=[1, 1, 1, 1], padding=&#39;SAME&#39;), b_c2))
  conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding=&#39;SAME&#39;)
  conv2 = tf.nn.dropout(conv2, keep_prob)
  w_c3 = tf.Variable(w_alpha * tf.random_normal([3, 3, 64, 64]))
  b_c3 = tf.Variable(b_alpha * tf.random_normal([64]))
  conv3 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv2, w_c3, strides=[1, 1, 1, 1], padding=&#39;SAME&#39;), b_c3))
  conv3 = tf.nn.max_pool(conv3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding=&#39;SAME&#39;)
  conv3 = tf.nn.dropout(conv3, keep_prob)
  # Fully connected layer
  w_d = tf.Variable(w_alpha * tf.random_normal([8 * 32 * 40, 1024]))
  b_d = tf.Variable(b_alpha * tf.random_normal([1024]))
  dense = tf.reshape(conv3, [-1, w_d.get_shape().as_list()[0]])
  dense = tf.nn.relu(tf.add(tf.matmul(dense, w_d), b_d))
  dense = tf.nn.dropout(dense, keep_prob)
  w_out = tf.Variable(w_alpha * tf.random_normal([1024, MAX_CAPTCHA * CHAR_SET_LEN]))
  b_out = tf.Variable(b_alpha * tf.random_normal([MAX_CAPTCHA * CHAR_SET_LEN]))
  out = tf.add(tf.matmul(dense, w_out), b_out)
  # out = tf.nn.softmax(out)
  return out
# 训练
def train_crack_captcha_cnn():
  import time
  start_time=time.time()
  output = crack_captcha_cnn()
  # loss
  #loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(output, Y))
  loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=output, labels=Y))
  # 最后一层用来分类的softmax和sigmoid有什么不同？
  # optimizer 为了加快训练 learning_rate应该开始大，然后慢慢衰
  optimizer = tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
  predict = tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN])
  max_idx_p = tf.argmax(predict, 2)
  max_idx_l = tf.argmax(tf.reshape(Y, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
  correct_pred = tf.equal(max_idx_p, max_idx_l)
  accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
  saver = tf.train.Saver()
  with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    step = 0
    while True:
      batch_x, batch_y = get_next_batch(64)
      _, loss_ = sess.run([optimizer, loss], feed_dict={X: batch_x, Y: batch_y, keep_prob: 0.75})
      print time.strftime(&#39;%Y-%m-%d %H:%M:%S&#39;,time.localtime(time.time())),step, loss_
      # 每100 step计算一次准确率
      if step % 100 == 0:
        batch_x_test, batch_y_test = get_next_batch(100)
        acc = sess.run(accuracy, feed_dict={X: batch_x_test, Y: batch_y_test, keep_prob: 1.})
        print u&#39;***************************************************************第%s次的准确率为%s&#39;%(step, acc)
        # 如果准确率大于50%,保存模型,完成训练
        if acc > 0.9:         ##我这里设了0.9，设得越大训练要花的时间越长，如果设得过于接近1，很难达到。如果使用cpu，花的时间很长，cpu占用很高电脑发烫。
          saver.save(sess, "crack_capcha.model", global_step=step)
          print time.time()-start_time
          break
      step += 1
train_crack_captcha_cnn()

Test code:

output = crack_captcha_cnn()
saver = tf.train.Saver()
sess = tf.Session()
saver.restore(sess, tf.train.latest_checkpoint('.'))
while(1):
  
  text, image = gen_captcha_text_and_image()
  image = convert2gray(image)
  image = image.flatten() / 255
  predict = tf.argmax(tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
  text_list = sess.run(predict, feed_dict={X: [image], keep_prob: 1})
  predict_text = text_list[0].tolist()
  vector = np.zeros(MAX_CAPTCHA * CHAR_SET_LEN)
  i = 0
  for t in predict_text:
    vector[i * 63 + t] = 1
    i += 1
    # break
  print("正确: {} 预测: {}".format(text, vec2text(vector)))

If you want to test the code effect quickly, do not set too many characters in the verification code, for example, 0123 is enough.

I believe you have mastered the method after reading the case in this article. For more exciting information, please pay attention to other related articles on the php Chinese website!

Recommended reading:

How to implement Mahalanobis distance in Python

##How to batch read txt files into DataFrame format in Python

The above is the detailed content of python identification verification code. For more information, please follow other related articles on the PHP Chinese website!