code
产生数据
import os
import cv2 as cv
import numpy as np
from math import *
from PIL import ImageFont
from PIL import Image
from PIL import ImageDraw
index = {"京": 0, "沪": 1, "津": 2, "渝": 3, "冀": 4, "晋": 5, "蒙": 6, "辽": 7, "吉": 8, "黑": 9,
"苏": 10, "浙": 11, "皖": 12, "闽": 13, "赣": 14, "鲁": 15, "豫": 16, "鄂": 17, "湘": 18, "粤": 19,
"桂": 20, "琼": 21, "川": 22, "贵": 23, "云": 24, "藏": 25, "陕": 26, "甘": 27, "青": 28, "宁": 29,
"新": 30, "0": 31, "1": 32, "2": 33, "3": 34, "4": 35, "5": 36, "6": 37, "7": 38, "8": 39,
"9": 40, "A": 41, "B": 42, "C": 43, "D": 44, "E": 45, "F": 46, "G": 47, "H": 48, "J": 49,
"K": 50, "L": 51, "M": 52, "N": 53, "P": 54, "Q": 55, "R": 56, "S": 57, "T": 58, "U": 59,
"V": 60, "W": 61, "X": 62, "Y": 63, "Z": 64}
chars = ["京", "沪", "津", "渝", "冀", "晋", "蒙", "辽", "吉", "黑",
"苏", "浙", "皖", "闽", "赣", "鲁", "豫", "鄂", "湘", "粤",
"桂", "琼", "川", "贵", "云", "藏", "陕", "甘", "青", "宁",
"新", "0", "1", "2", "3", "4", "5", "6", "7", "8",
"9", "A", "B", "C", "D", "E", "F", "G", "H", "J",
"K", "L", "M", "N", "P", "Q", "R", "S", "T", "U",
"V", "W", "X", "Y", "Z"]
def AddSmudginess(img, Smu):
"""
模糊处理
:param img: 输入图像
:param Smu: 模糊图像
:return: 添加模糊后的图像
"""
rows = r(Smu.shape[0] - 50)
cols = r(Smu.shape[1] - 50)
adder = Smu[rows:rows + 50, cols:cols + 50]
adder = cv.resize(adder, (50, 50))
img = cv.resize(img, (50, 50))
img = cv.bitwise_not(img)
img = cv.bitwise_and(adder, img)
img = cv.bitwise_not(img)
return img
def rot(img, angel, shape, max_angel):
"""
添加透视畸变
"""
size_o = [shape[1], shape[0]]
size = (shape[1] + int(shape[0] * cos((float(max_angel) / 180) * 3.14)), shape[0])
interval = abs(int(sin((float(angel) / 180) * 3.14) * shape[0]))
pts1 = np.float32([[0, 0], [0, size_o[1]], [size_o[0], 0], [size_o[0], size_o[1]]])
if angel > 0:
pts2 = np.float32([[interval, 0], [0, size[1]], [size[0], 0], [size[0] - interval, size_o[1]]])
else:
pts2 = np.float32([[0, 0], [interval, size[1]], [size[0] - interval, 0], [size[0], size_o[1]]])
M = cv.getPerspectiveTransform(pts1, pts2)
dst = cv.warpPerspective(img, M, size)
return dst
def rotRandrom(img, factor, size):
"""
添加放射畸变
:param img: 输入图像
:param factor: 畸变的参数
:param size: 图片目标尺寸
:return: 放射畸变后的图像
"""
shape = size
pts1 = np.float32([[0, 0], [0, shape[0]], [shape[1], 0], [shape[1], shape[0]]])
pts2 = np.float32([[r(factor), r(factor)], [r(factor), shape[0] - r(factor)], [shape[1] - r(factor), r(factor)],
[shape[1] - r(factor), shape[0] - r(factor)]])
M = cv.getPerspectiveTransform(pts1, pts2)
dst = cv.warpPerspective(img, M, size)
return dst
def tfactor(img):
"""
添加饱和度光照的噪声
"""
hsv = cv.cvtColor(img, cv.COLOR_BGR2HSV)
hsv[:, :, 0] = hsv[:, :, 0] * (0.8 + np.random.random() * 0.2)
hsv[:, :, 1] = hsv[:, :, 1] * (0.3 + np.random.random() * 0.7)
hsv[:, :, 2] = hsv[:, :, 2] * (0.2 + np.random.random() * 0.8)
img = cv.cvtColor(hsv, cv.COLOR_HSV2BGR)
return img
def random_envirment(img, noplate_bg):
"""
添加自然环境的噪声, noplate_bg为不含车牌的背景图
"""
bg_index = r(len(noplate_bg))
env = cv.imread(noplate_bg[bg_index])
env = cv.resize(env, (img.shape[1], img.shape[0]))
bak = (img == 0)
bak = bak.astype(np.uint8) * 255
inv = cv.bitwise_and(bak, env)
img = cv.bitwise_or(inv, img)
return img
def GenCh(f, val):
"""
生成中文字符
"""
img = Image.new("RGB", (45, 70), (255, 255, 255))
draw = ImageDraw.Draw(img)
draw.text((0, 3), val, (0, 0, 0), font=f)
img = img.resize((23, 70))
A = np.array(img)
return A
def GenCh1(f, val):
"""
生成英文字符
"""
img = Image.new("RGB", (23, 70), (255, 255, 255))
draw = ImageDraw.Draw(img)
draw.text((0, 2), val, (0, 0, 0), font=f) # val.decode('utf-8')
A = np.array(img)
return A
def AddGauss(img, level):
"""
添加高斯模糊
"""
return cv.blur(img, (level * 2 + 1, level * 2 + 1))
def r(val):
return int(np.random.random() * val)
def AddNoiseSingleChannel(single):
"""
添加高斯噪声
"""
diff = 255 - single.max()
noise = np.random.normal(0, 1 + r(6), single.shape)
noise = (noise - noise.min()) / (noise.max() - noise.min())
noise *= diff
# noise= noise.astype(np.uint8)
dst = single + noise
return dst
def addNoise(img): # sdev = 0.5,avg=10
img[:, :, 0] = AddNoiseSingleChannel(img[:, :, 0])
img[:, :, 1] = AddNoiseSingleChannel(img[:, :, 1])
img[:, :, 2] = AddNoiseSingleChannel(img[:, :, 2])
return img
class GenPlate:
def __init__(self, fontCh, fontEng, NoPlates):
self.fontC = ImageFont.truetype(fontCh, 43, 0)
self.fontE = ImageFont.truetype(fontEng, 60, 0)
self.img = np.array(Image.new("RGB", (226, 70), (255, 255, 255)))
self.bg = cv.resize(cv.imread("images/template.bmp"), (226, 70)) # template.bmp:车牌背景图
self.smu = cv.imread("images/smu2.jpg") # smu2.jpg:模糊图像
self.noplates_path = []
for parent, parent_folder, filenames in os.walk(NoPlates):
for filename in filenames:
path = parent + "/" + filename
self.noplates_path.append(path)
def draw(self, val):
offset = 2
self.img[0:70, offset + 8:offset + 8 + 23] = GenCh(self.fontC, val[0])
self.img[0:70, offset + 8 + 23 + 6:offset + 8 + 23 + 6 + 23] = GenCh1(self.fontE, val[1])
for i in range(5):
base = offset + 8 + 23 + 6 + 23 + 17 + i * 23 + i * 6
self.img[0:70, base:base + 23] = GenCh1(self.fontE, val[i + 2])
return self.img
def generate(self, text):
if len(text) == 7:
fg = self.draw(text) # decode(encoding="utf-8")
fg = cv.bitwise_not(fg)
com = cv.bitwise_or(fg, self.bg)
com = rot(com, r(60) - 30, com.shape, 30)
com = rotRandrom(com, 10, (com.shape[1], com.shape[0]))
com = tfactor(com)
com = random_envirment(com, self.noplates_path)
#com = AddGauss(com, 1 + r(4))
com = addNoise(com)
return com
@staticmethod
def genPlateString(pos, val):
"""
生成车牌string，存为图片
生成车牌list，存为label
"""
plateStr = ""
plateList = []
box = [0, 0, 0, 0, 0, 0, 0]
if pos != -1:
box[pos] = 1
for unit, cpos in zip(box, range(len(box))):
if unit == 1:
plateStr += val
plateList.append(val)
else:
if cpos == 0:
plateStr += chars[r(31)]
plateList.append(plateStr)
elif cpos == 1:
plateStr += chars[41 + r(24)]
plateList.append(plateStr)
else:
plateStr += chars[31 + r(34)]
plateList.append(plateStr)
plate = [plateList[0]]
b = [plateList[i][-1] for i in range(len(plateList))]
plate.extend(b[1:7])
return plateStr, plate
@staticmethod
def genBatch(batchsize, outputPath, size):
"""
将生成的车牌图片写入文件夹，对应的label写入label.txt
:param batchsize: 批次大小
:param outputPath: 输出图像的保存路径
:param size: 输出图像的尺寸
:return: None
"""
if not os.path.exists(outputPath):
os.mkdir(outputPath)
outfile = open('label.txt', 'w', encoding='utf-8')
for i in range(batchsize):
plateStr, plate = G.genPlateString(-1, -1)
print(plateStr, plate)
img = G.generate(plateStr)
img = cv.resize(img, size)
cv.imwrite(outputPath + "/" + str(i).zfill(2) + ".jpg", img)
outfile.write(str(plate) + "\n")
if __name__ == '__main__':
G = GenPlate("font/platech.ttf", 'font/platechar.ttf', "NoPlates")
G.genBatch(201, 'plate', (272, 72))
训练
import os
import time
import datetime
import numpy as np
import tensorflow as tf
from input_data import OCRIter
import model
os.environ["TF_CPP_MIN_LOG_LEVEL"] = '3'
img_h = 72
img_w = 272
num_label = 7
batch_size = 32
epoch = 10000
learning_rate = 0.0001
logs_path = 'logs '
model_path = 'saved_model'
image_holder = tf.compat.v1.placeholder(tf.float32, [batch_size, img_h, img_w, 3])
label_holder = tf.compat.v1.placeholder(tf.int32, [batch_size, 7])
keep_prob = tf.compat.v1.placeholder(tf.float32)
def get_batch():
data_batch = OCRIter(batch_size, img_h, img_w)
image_batch, label_batch = data_batch.iter()
return np.array(image_batch), np.array(label_batch)
logit1, logit2, logit3, logit4, logit5, logit6, logit7 = model.cnn_inference(
image_holder, keep_prob)
loss1, loss2, loss3, loss4, loss5, loss6, loss7 = model.calc_loss(
logit1, logit2, logit3, logit4, logit5, logit6, logit7, label_holder)
train_op1, train_op2, train_op3, train_op4, train_op5, train_op6, train_op7 = model.train_step(
loss1, loss2, loss3, loss4, loss5, loss6, loss7, learning_rate)
accuracy = model.pred_model(logit1, logit2, logit3, logit4, logit5, logit6, logit7, label_holder)
input_image = tf.compat.v1.summary.image('input', image_holder)
summary_op = tf.compat.v1.summary.merge(tf.compat.v1.get_collection(tf.compat.v1.GraphKeys.SUMMARIES))
init_op = tf.compat.v1.global_variables_initializer()
with tf.compat.v1.Session() as sess:
sess.run(init_op)
train_writer = tf.compat.v1.summary.FileWriter(logs_path, sess.graph)
saver = tf.compat.v1.train.Saver()
start_time1 = time.time()
for step in range(epoch):
# 生成车牌图像以及标签数据
img_batch, lbl_batch = get_batch()
start_time2 = time.time()
time_str = datetime.datetime.now().isoformat()
feed_dict = {image_holder: img_batch, label_holder: lbl_batch, keep_prob: 0.6}
_1, _2, _3, _4, _5, _6, _7, ls1, ls2, ls3, ls4, ls5, ls6, ls7, acc = sess.run(
[train_op1, train_op2, train_op3, train_op4, train_op5, train_op6, train_op7,
loss1, loss2, loss3, loss4, loss5, loss6, loss7, accuracy], feed_dict)
summary_str = sess.run(summary_op, feed_dict)
train_writer.add_summary(summary_str, step)
duration = time.time() - start_time2
loss_total = ls1 + ls2 + ls3 + ls4 + ls5 + ls6 + ls7
if step % 10 == 0:
sec_per_batch = float(duration)
print('%s: Step %d, loss_total = %.2f, acc = %.2f%%, sec/batch = %.2f' %
(time_str, step, loss_total, acc * 100, sec_per_batch))
if step % 5000 == 0 or (step + 1) == epoch:
checkpoint_path = os.path.join(model_path, 'model.ckpt')
saver.save(sess, checkpoint_path, global_step=step)
end_time = time.time()
print("Training over. It costs {:.2f} minutes".format((end_time - start_time1) / 60))
测试
import os
import cv2 as cv
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
from PIL import Image
import model
os.environ["TF_CPP_MIN_LOG_LEVEL"] = '3' # 只显示 Error
index = {"京": 0, "沪": 1, "津": 2, "渝": 3, "冀": 4, "晋": 5, "蒙": 6, "辽": 7, "吉": 8, "黑": 9,
"苏": 10, "浙": 11, "皖": 12, "闽": 13, "赣": 14, "鲁": 15, "豫": 16, "鄂": 17, "湘": 18, "粤": 19,
"桂": 20, "琼": 21, "川": 22, "贵": 23, "云": 24, "藏": 25, "陕": 26, "甘": 27, "青": 28, "宁": 29,
"新": 30, "0": 31, "1": 32, "2": 33, "3": 34, "4": 35, "5": 36, "6": 37, "7": 38, "8": 39,
"9": 40, "A": 41, "B": 42, "C": 43, "D": 44, "E": 45, "F": 46, "G": 47, "H": 48, "J": 49,
"K": 50, "L": 51, "M": 52, "N": 53, "P": 54, "Q": 55, "R": 56, "S": 57, "T": 58, "U": 59,
"V": 60, "W": 61, "X": 62, "Y": 63, "Z": 64}
chars = ["京", "沪", "津", "渝", "冀", "晋", "蒙", "辽", "吉", "黑",
"苏", "浙", "皖", "闽", "赣", "鲁", "豫", "鄂", "湘", "粤",
"桂", "琼", "川", "贵", "云", "藏", "陕", "甘", "青", "宁",
"新", "0", "1", "2", "3", "4", "5", "6", "7", "8",
"9", "A", "B", "C", "D", "E", "F", "G", "H", "J",
"K", "L", "M", "N", "P", "Q", "R", "S", "T", "U",
"V", "W", "X", "Y", "Z"]
def get_one_image(test):
""" 随机获取单张车牌图像 """
n = len(test)
rand_num = np.random.randint(0, n)
img_dir = test[rand_num]
image_show = Image.open(img_dir)
plt.imshow(image_show) # 显示车牌图片
image = cv.imread(img_dir)
image = image.reshape(1,72, 272, 3)
image = np.multiply(image, 1 / 255.0)
return image
batch_size = 1
x = tf.compat.v1.placeholder(tf.float32, [batch_size, 72, 272, 3])
keep_prob = tf.compat.v1.placeholder(tf.float32)
test_dir = 'plate/'
test_image = []
for file in os.listdir(test_dir):
test_image.append(test_dir + file)
test_image = list(test_image)
image_array = get_one_image(test_image)
logit1, logit2, logit3, logit4, logit5, logit6, logit7 = model.cnn_inference(x, keep_prob)
model_path = 'saved_model/'
saver = tf.compat.v1.train.Saver()
with tf.compat.v1.Session() as sess:
print("Reading checkpoint...")
ckpt = tf.train.get_checkpoint_state(model_path)
if ckpt and ckpt.model_checkpoint_path:
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
saver.restore(sess, ckpt.model_checkpoint_path)
print('Loading success, global_step is %s' % global_step)
else:
print('No checkpoint file found')
pre1, pre2, pre3, pre4, pre5, pre6, pre7 = sess.run(
[logit1, logit2, logit3, logit4, logit5, logit6, logit7],
feed_dict={x: image_array, keep_prob: 1.0})
prediction = np.reshape(np.array([pre1, pre2, pre3, pre4, pre5, pre6, pre7]), [-1, 65])
max_index = np.argmax(prediction, axis=1)
print(max_index)
line = ''
result = np.array([])
for i in range(prediction.shape[0]):
if i == 0:
result = np.argmax(prediction[i][0:31])
if i == 1:
result = np.argmax(prediction[i][41:65]) + 41
if i > 1:
result = np.argmax(prediction[i][31:65]) + 31
line += chars[result] + " "
print('predicted: ' + line)
plt.show()
