直接上代碼

#####################################構建網(wǎng)絡模型以及訓練部分##################

coding:utf-8

fromgen_captchaimportgen_captcha_text_and_image

fromgen_captchaimportnumber

fromgen_captchaimportalphabet

fromgen_captchaimportALPHABET

importmatplotlib.pyplotasplt

fromPILimportImage

importnumpyasnp

importtensorflowastf

text, image = gen_captcha_text_and_image()

print("驗證碼圖像channel:", image.shape)# (60, 160, 3)

# 圖像大小

IMAGE_HEIGHT =60

IMAGE_WIDTH =160

MAX_CAPTCHA =len(text)

print("驗證碼文本最長字符數(shù)", MAX_CAPTCHA)# 驗證碼最長4字符; 我全部固定為4,可以不固定. 如果驗證碼長度小于4固蛾，用'_'補齊

# 把彩色圖像轉為灰度圖像（色彩對識別驗證碼沒有什么用）

defconvert2gray(img):

iflen(img.shape) >2:

gray = np.mean(img, -1)

# 上面的轉法較快跃惫，正規(guī)轉法如下

# r, g, b = img[:,:,0], img[:,:,1], img[:,:,2]

# gray = 0.2989 * r + 0.5870 * g + 0.1140 * b

returngray

else:

returnimg

"""

cnn在圖像大小是2的倍數(shù)時性能最高, 如果你用的圖像大小不是2的倍數(shù)，可以在圖像邊緣補無用像素腊徙。

np.pad(image【,((2,3),(2,2)), 'constant', constant_values=(255,))? # 在圖像上補2行，下補3行蝌焚，左補2行简识，右補2行

"""

#############@@@@@@@@@@@@@@@@

#文本轉向量

char_set = number + alphabet + ALPHABET + ['_']

CHAR_SET_LEN? =len(char_set)

deftext2vec(text):

text_len =len(text)

iftext_len > MAX_CAPTCHA:

raiseValueError('驗證碼最長四個字符')

##########@@@@@@@@@@@@@@@@

# 文本轉向量

char_set = number + alphabet + ALPHABET + ['_']# 如果驗證碼長度小于4, '_'用來補齊

CHAR_SET_LEN =len(char_set)

deftext2vec(text):

text_len =len(text)

iftext_len > MAX_CAPTCHA:

raiseValueError('驗證碼最長4個字符')

vector = np.zeros(MAX_CAPTCHA*CHAR_SET_LEN)

defchar2pos(c):

ifc =='_':

k =62

returnk

k =ord(c)-48

ifk >9:

k =ord(c) -55

ifk >35:

k =ord(c) -61

ifk >61:

raiseValueError('No Map')

returnk

fori, cinenumerate(text):

idx = i * CHAR_SET_LEN + char2pos(c)

vector[idx] =1

returnvector

# 向量轉回文本

defvec2text(vec):

char_pos = vec.nonzero()[0]

text=[]

fori, cinenumerate(char_pos):

char_at_pos= i#c/63

char_idx = c % CHAR_SET_LEN

ifchar_idx <10:

char_code = char_idx +ord('0')

elifchar_idx <36:

char_code = char_idx -10+ord('A')

elifchar_idx <62:

char_code = char_idx-36+ord('a')

elifchar_idx ==62:

char_code =ord('_')

else:

raiseValueError('error')

text.append(chr(char_code))

return"".join(text)

"""

#向量（大小MAX_CAPTCHA*CHAR_SET_LEN）用0,1編碼 每63個編碼一個字符冲秽，這樣順利有，字符也有

vec = text2vec("F5Sd")

text = vec2text(vec)

print(text)? # F5Sd

vec = text2vec("SFd5")

text = vec2text(vec)

print(text)? # SFd5

"""

#生成一個訓練batch

defget_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])

# 有時生成圖像大小不是(60, 160, 3)

defwrap_gen_captcha_text_and_image():

whileTrue:

text, image = gen_captcha_text_and_image()

ifimage.shape == (60,160,3):

returntext, image

foriinrange(batch_size):

text, image = wrap_gen_captcha_text_and_image()

image = convert2gray(image)

batch_x[i,:] = image.flatten() /255# (image.flatten()-128)/128? mean為0

batch_y[i,:] = text2vec(text)

returnbatch_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

# 定義CNN

defcrack_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='SAME'), b_c1))

conv1 = tf.nn.max_pool(conv1,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

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='SAME'), b_c2))

conv2 = tf.nn.max_pool(conv2,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

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='SAME'), b_c3))

conv3 = tf.nn.max_pool(conv3,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

conv3 = tf.nn.dropout(conv3, keep_prob)

# Fully connected layer

w_d = tf.Variable(w_alpha*tf.random_normal([8*20*64,1024]))

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)

returnout

# 訓練

deftrain_crack_captcha_cnn():

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))

# 最后一層用來分類的softmax和sigmoid有什么不同镊靴？

# optimizer 為了加快訓練 learning_rate應該開始大铣卡，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()

withtf.Session()assess:

sess.run(tf.global_variables_initializer())

step =0

whileTrue:

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(step, loss_)

# 每100 step計算一次準確率

ifstep %10==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(step, acc)

# 如果準確率大于50%,保存模型,完成訓練

ifacc >0.98:

saver.save(sess,"./crack_capcha.model",global_step=step)

break

step +=1

defcrack_captcha(captcha_image):

output = crack_captcha_cnn()

saver = tf.train.Saver()

withtf.Session()assess:

saver.restore(sess, tf.train.latest_checkpoint('.'))

predict = tf.argmax(tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN]),2)

text_list = sess.run(predict,feed_dict={X: [captcha_image], keep_prob:1})

text = text_list[0].tolist()

vector = np.zeros(MAX_CAPTCHA*CHAR_SET_LEN)

i =0

fornintext:

vector[i*CHAR_SET_LEN + n] =1

i +=1

returnvec2text(vector)

if__name__ =='__main__':

text, image = gen_captcha_text_and_image()

img111 = image

image = convert2gray(image)

image = image.flatten() /255

predict_text = crack_captcha(image)

print("正確: {}? 預測: {}".format(text, predict_text))

plt.imshow(img111)

plt.show()

#? train_crack_captcha_cnn()

#####################################生成驗證碼部分##################

#coding=utf-8

fromcaptcha.imageimportImageCaptcha# pip install captcha

importnumpyasnp

importmatplotlib.pyplotasplt

fromPILimportImage

importrandom

importmatplotlib.imageasmpimg

importtensorflowastf

importcv2

# 驗證碼中的字符, 就不用漢字了

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 =[]

'''

# 驗證碼一般都無視大小寫；驗證碼長度4個字符

defrandom_captcha_text(char_set=number + alphabet + ALPHABET, captcha_size=4):

captcha_text = []

foriinrange(captcha_size):

c = random.choice(char_set)

captcha_text.append(c)

returncaptcha_text

# 生成字符對應的驗證碼

defgen_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')? # 寫到文件

captcha_image = Image.open(captcha)

#captcha_image.show()

captcha_image = np.array(captcha_image)

ifcaptcha_image.shape==(60,160,3):

break

lena = mpimg.imread('jiqi10.jpg')

shrink = cv2.resize(lena, (160,60),interpolation=cv2.INTER_AREA)

print("testttttttt:",shrink.shape)

returncaptcha_text, captcha_image

if__name__ =='__main__':

# 測試

text, image = gen_captcha_text_and_image()

printimage

gray = np.mean(image, -1)

printgray

printimage.shape

printgray.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()

#################################預測結果#######################

原文參考鏈接：http://www.cnblogs.com/ydf0509/p/6916435.html