def testRollerBlock():
import matplotlib.pyplot as plt
import platform
import cv2
import os
if "Windows" in platform.system():
filename = r"D:\chengxu\python\project\digitRecognise\com\huitong\gasMeterv1\data\img\style1\2.jpg"
elif "Linux" in platform.system():
filename = r"/home/allen/work/digitRecognise/com/huitong/gasMeterv1/data/img/style1/000.jpg"
style1 = GasmeterStyle1(desImageDepth=1)
image = cv2.imread(filename)
ImageTool.showImagePIL(image)
style1.setImage(image)
rollerBlackImage = style1.getRollerBlackArea()
# rollerBlackImage = cv2.cvtColor(rollerBlackImage,cv2.COLOR_BGR2RGB)
title = str(rollerBlackImage.shape) + os.path.basename(filename)
# rollerBlackImage = ImageTool.convertImgBGR2Gray(rollerBlackImage)
# ret,rollerBlackImage = ImageTool.getOTSUGrayImage(rollerBlackImage)
plt.figure()
plt.imshow(rollerBlackImage)
plt.title(title)
plt.show()
def getPredict(hps, mode, gasmeter_filename, save_file_name):
xp = tf.placeholder(
tf.float32,
[None, captchaBoxHeight * captchaBoxWidth * gen.ImageDepth])
yp = tf.placeholder(tf.float32,
[None, captchaCharacterLength * CHAR_SET_LEN])
model = ResNetModel.ResNetModel(hps, xp, yp, mode, captchaBoxHeight,
captchaBoxWidth, gen.ImageDepth)
model.create_graph(captchaCharacterLength)
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
saver = tf.train.Saver()
saver.restore(sess, save_file_name)
# images = gen.get_batch_gasmeter_digit_area_from_filename(gasmeter_filename, hps.batch_nums)
image = cv2.imread(gasmeter_filename)
# gasmeter = GasmeterStyle0(captchaBoxWidth,captchaBoxHeight,desImageDepth=gen.ImageDepth)
gasmeter = GasmeterStyle1(captchaBoxWidth,
captchaBoxHeight,
desImageDepth=gen.ImageDepth)
gasmeter.setImage(image)
image = gasmeter.getRollerBlackArea()
ImageTool.showImagePIL(image)
images = ImageTool.repeatImage2Tensor(image, hps.batch_nums)
feed_dict = {xp: images, model.is_training_ph: False}
outputs = sess.run([model.outputs], feed_dict=feed_dict)
text = get_predict_text(outputs)
return text
def getLCDAreaData(image):
"""
获取 LCD 区域图片数据
:param image: cv2 读进来的图片对象
"""
image = ImageTool.preProcessImage(image)
blackMask = MaskTool.getBlackMaskBGR()
image = blackMask.getInterestImageAreaData(image)
shape = image.shape
grayImage = ImageTool.convertImgBGR2Gray(image)
mid = int(shape[0] / 2)
splitImageBox = (0, mid, shape[1], shape[0])
splitImageGray = ImageTool.getCropImageByBox(grayImage, splitImageBox)
splitImage = ImageTool.getCropImageByBox(image, splitImageBox)
# 显示剪切的 lcd 屏所在的下半屏灰度图
ImageTool.showImagePIL(splitImageGray, "splitImageGray")
retval, otsuImage = ImageTool.getOTSUGrayImage(splitImageGray)
otsuImage = ImageTool.convertImgGray2BGR(otsuImage)
lower = (250, 250, 250)
upper = (255, 255, 255)
lcdBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
otsuImage, lower, upper)
lcdBox = LCDLightDetect.getMinBox(lcdBoxCorner)
# ImageTool.showBoxInImageByBox(splitImage, lcdBox)
lcdImage = ImageTool.getCropImageByBox(splitImage, lcdBox)
return lcdImage
def getPredict(hps, mode, gasmeter_filename, save_file_name):
xp = tf.placeholder(
tf.float32,
[None, captchaBoxHeight * captchaBoxWidth * gen.ImageDepth])
yp = tf.placeholder(tf.float32,
[None, captchaCharacterLength * CHAR_SET_LEN])
model = ResNetModel.ResNetModel(hps, xp, yp, mode, captchaBoxHeight,
captchaBoxWidth, gen.ImageDepth)
model.create_graph(captchaCharacterLength)
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
saver = tf.train.Saver()
saver.restore(sess, save_file_name)
# image = Image.open(gasmeter_filename)
# image = image.resize((captchaBoxWidth,captchaBoxHeight), Image.BICUBIC)
image = cv2.imread(gasmeter_filename)
image = ImageTool.imageResize(image, captchaBoxWidth, captchaBoxHeight)
if gen.ImageDepth == 1:
image = ImageTool.convertImgBGR2Gray(image)
# image = ImageTool.convertImgRGB2Gray(image)
ImageTool.showImagePIL(image)
images = ImageTool.repeatImage2Tensor(image, hps.batch_nums)
feed_dict = {xp: images, model.is_training_ph: False}
outputs = sess.run([model.outputs], feed_dict=feed_dict)
text = get_predict_text(outputs)
return text
def getRollerBlackArea(self):
rollerBox = self.getRollerBox()
rollerImage = ImageTool.getCropImageByBox(self._image, rollerBox)
rollerImage = ImageTool.imageResize(rollerImage, self._desWidth,
self._desHeight)
if self._desImageDepth == 1:
rollerImage = ImageTool.convertImgBGR2Gray(rollerImage)
self._rollerImage = rollerImage
return self._rollerImage
def fun8():
from com.huitong.gasMeterv1.framework.tool.GenDigitsImage import GenDigitsPicture
characterLength = 1
width = 15
height = 30
bkgColor = (20,20,20)
fontColor = (200,200,200)
fontSizes = (29,)
gen = GenDigitsPicture(characterLength, width, height)
while True:
text,image = gen.get_text_and_image(backgroundColor=bkgColor,fontColor=fontColor,fontSizes=fontSizes)
ImageTool.showImagePIL(image,text)
def get_text_and_image(self,
backgroundColor=None,
fontColor=None,
fontSizes=(28, )):
""" 生成字符序列和对应的图片数据 图片颜色通道是(R,G,B)-> text,image"""
charBoxWidth = 15
charBoxHeight = 30
blackBkgColor = ImageCaptcha.random_bkg_color(0, 90)
fontColor = ImageCaptcha.random_font_color(180, 250)
grayBkgColor = ImageCaptcha.random_bkg_color(130, 220)
image = ImageCaptcha(width=charBoxWidth,
height=charBoxHeight,
backgroundColor=blackBkgColor,
fontColor=fontColor,
font_sizes=fontSizes)
captcha_text = self._random_text()
captcha_text = ''.join(captcha_text)
filename = GenImageGasMeterStyle1m2.getRandomFilename()
# print(filename)
captcha_image = Image.open(filename)
captcha_image = ImageTool.imageResizePIL(captcha_image,
self._picBoxWidth,
self._picBoxHeight)
dxs = [7, 32, 58, 83, 107]
dy = 17
i = 0
def randDy():
return 4 - random.randint(1, 8)
for c in captcha_text:
desX = dxs[i]
charImage = image.generate(c)
charImage = Image.open(charImage)
captcha_image.paste(charImage, (desX, dy + randDy()))
i += 1
if self._imageDepth == 1:
# 将彩色图片转换成灰度图图片
captcha_image = ImageTool.convertImgRGB2Gray(captcha_image)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def setImage(self, image):
"""
设置要处理的图片
:param image: 是cv2读进来的图片对象
"""
image = ImageTool.preProcessImage(image)
super(GasmeterStyle0, self).setImage(image)
def getRollerBox(self):
"""
获得黑底白字滚轮区域box，并将该 box 赋值给 self._rollerBox
"""
rollerBoxWidthMax = 425
shape = self._image.shape
mid = int(shape[0] / 2)
quarter = int(shape[0] / 4)
splitImageBox = (70, quarter, 170, mid)
splitImageGray = ImageTool.getCropImageByBox(self._grayImage,
splitImageBox)
retval, otsuImage = ImageTool.getOTSUGrayImage(splitImageGray)
otsuImage = ImageTool.convertImgGray2BGR(otsuImage)
lower = (250, 250, 250)
upper = (255, 255, 255)
splitBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
otsuImage, lower, upper)
splitBox = ImageTool.getBoxFromBoxCorner(splitBoxCorner)
rollerBoxUp = int(quarter + splitBox[1])
rollerBoxDown = int(quarter + splitBox[3])
trollerBox = (0, rollerBoxUp, shape[1], rollerBoxDown)
rollerImageGray = ImageTool.getCropImageByBox(self._grayImage,
trollerBox)
retval, otsuImage = ImageTool.getOTSUGrayImage(rollerImageGray)
otsuImage = ImageTool.convertImgGray2BGR(otsuImage)
trollerBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
otsuImage, lower, upper)
trollerBox = ImageTool.getBoxFromBoxCorner(trollerBoxCorner)
rollerBox = (trollerBox[0], rollerBoxUp + 1, trollerBox[2],
rollerBoxDown - 1)
getWidth = rollerBox[2] - rollerBox[0]
if rollerBoxWidthMax < getWidth:
rollerBox = (rollerBox[0], rollerBox[1],
rollerBox[0] + rollerBoxWidthMax, rollerBox[3])
right = int(rollerBox[2] * 5 / 8) + 20
rollerBox = (rollerBox[0], rollerBox[1], right, rollerBox[3])
# ImageTool.showBoxInImageByBox(self._image,rollerBox)
self._rollerBox = rollerBox
return rollerBox
def getInterestBoxCornerPoint(self, bkgImage):
"""
获得感兴趣的图片区域所在矩形的四个顶角坐标，如果图片中没有在所选区域范围内，返回 None
:param bkgImage: 通过cv2 读进来的图像对象
"""
boxCornerPoint = ImageTool.getInterestBoxCornerPointByColor(
bkgImage, self._colorLower, self._colorUpper)
return boxCornerPoint
def setImage(self, image):
"""
设置要处理的图片
:param image: 是cv2读进来的图片对象
提取表头黑色背景区域图片保存到 self._image 域中，方便以后提取条形码、液晶屏、黑底白字滚轮区域数字
"""
if image is None:
raise ValueError("image is None")
image = ImageTool.preProcessImage(image)
# ImageTool.showImagePIL(image)
super(GasmeterStyle1, self).setImage(image)
blackMask = MaskTool.getBlackMaskBGR()
self._image = blackMask.getInterestImageAreaData(self._image)
self._grayImage = ImageTool.convertImgBGR2Gray(self._image)
def fun7():
baseDir = r"D:\chengxu\python\project\digitRecognise\com\huitong\gasMeterv1"
imgdirname = ["data", "img","gasmeterHead","style1"]
dirname = FileNameUtil.getDirname(baseDir,imgdirname)
filelist = FileNameUtil.getPathFilenameList(dirname,".*\.jpg")
interestLoweer = (0,0,100)
interestUpper = (60,45,255)
interestLoweer = (40, 35, 80)
interestUpper = (105, 105, 230)
for each in filelist:
image = cv2.imread(each)
cornerpoint = ImageTool.getInterestBoxCornerPointByColor(image,interestLoweer,interestUpper)
title = os.path.basename(each)
ImageTool.showBoxInImageByBoxCornerPoint(image, cornerpoint, title)
def getPredict(hps, mode, save_file_name, gen):
xp = tf.placeholder(
tf.float32,
[None, captchaBoxHeight * captchaBoxWidth * gen.ImageDepth])
yp = tf.placeholder(tf.float32,
[None, captchaCharacterLength * CHAR_SET_LEN])
model = ResNetModel.ResNetModel(hps, xp, yp, mode, captchaBoxHeight,
captchaBoxWidth, gen.ImageDepth)
model.create_graph(captchaCharacterLength)
gen1 = GenImageGasMeterStyle1m1(captchaCharacterLength,
captchaBoxWidth,
captchaBoxHeight,
imageDepth=1)
gen2 = GenImageGasMeterStyle1m2(captchaCharacterLength,
captchaBoxWidth,
captchaBoxHeight,
imageDepth=1)
gen3 = GenImageGasMeterStyle1m3(captchaCharacterLength,
captchaBoxWidth,
captchaBoxHeight,
imageDepth=1)
gens = [gen1, gen2, gen3]
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
saver = tf.train.Saver()
saver.restore(sess, save_file_name)
while True:
gen = random.choice(gens)
oriText, image = gen.get_text_and_image()
images = ImageTool.repeatImage2Tensor(image, hps.batch_nums)
feed_dict = {xp: images, model.is_training_ph: False}
outputs = sess.run([model.outputs], feed_dict=feed_dict)
predictText = get_predict_text(outputs)
title = "text:%s, predict:%s" % (oriText, predictText)
ImageTool.showImagePIL(image, title)
print(title)
def getInterestBox(self, bkgImage):
"""
获得感兴趣的区域box，box 是由 (left, upper, right, lower ) pixel coordinate tuple 来表示
如果没有返回 None
:param bkgImage: 通过cv2 读进来的图像对象
"""
boxCornerPoint = self.getInterestBoxCornerPoint(bkgImage)
if boxCornerPoint is None:
return None
box = ImageTool.getBoxFromBoxCorner(boxCornerPoint)
return box
def test():
import cv2
import platform
import os
style = GasmeterStyle0(desImageDepth=1)
if "Linux" in platform.system():
filename = r'/home/allen/work/digitRecognise/com/huitong/gasMeterv1/data/img/style0/10.jpg'
elif "Windows" in platform.system():
filename = r"D:\chengxu\python\project\digitRecognise\com\huitong\gasMeterv1\data\img\style0\10.jpg"
image = cv2.imread(filename)
title = os.path.basename(filename)
style.setImage(image)
image = style.getRollerBlackArea()
# ret, image = ImageTool.getOTSUGrayImage(image)
# image = cv2.cvtColor(image,cv2.COLOR_BGR2RGB)
ImageTool.showImagePIL(image, str(image.shape) + title)
def getInterestImageAreaData(self, bkgImage):
"""
获得图片中感兴趣的颜色范围区域数据，如果没有返回 None
:param bkgImage: 通过cv2 读进来的图像对象
"""
boxCornerPoint = self.getInterestBoxCornerPoint(bkgImage)
if boxCornerPoint is None:
return None
image = ImageTool.getCropImageByBoxCornerPoint(bkgImage,
boxCornerPoint)
return image
def get_text_and_image(self,
backgroundColor=None,
fontColor=None,
fontSizes=(28, )):
filename = GenImageGasMeterStyle1m2.getRandomFilename()
# print(filename)
captcha_image = Image.open(filename)
captcha_image = ImageTool.imageResizePIL(captcha_image,
self._picBoxWidth,
self._picBoxHeight)
t = FileNameUtil.getFilenameFromFullFilepathname(filename)
captcha_text = t[:5]
if self._imageDepth == 1:
# 将彩色图片转换成灰度图图片
captcha_image = ImageTool.convertImgRGB2Gray(captcha_image)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def getRollerBlackArea(self):
if self._image is None:
raise ValueError("应先通过setImage()函数设置image，然后获取感兴趣数据")
blackMask = MaskTool.getBlackMaskBGR()
blackImage = blackMask.getInterestImageAreaData(self._image)
redMask = MaskTool.getRedMaskBGR()
redBox = redMask.getInterestBox(blackImage)
if redBox is not None:
rollerBlackArea = ImageTool.removeRightArea(blackImage, redBox[0])
else:
rollerBlackArea = blackImage
rollerBlackArea = ImageTool.imageResize(rollerBlackArea,
self._desWidth,
self._desHeight)
if self._desImageDepth == 1:
rollerBlackArea = ImageTool.convertImgBGR2Gray(rollerBlackArea)
return rollerBlackArea
def getLCDBox(self):
"""
"""
shape = self._image.shape
mid = int(shape[0] / 2)
splitImageBox = (0, mid, shape[1], shape[0])
splitImageGray = ImageTool.getCropImageByBox(self._grayImage,
splitImageBox)
ImageTool.showImagePIL(splitImageGray)
splitImage = ImageTool.getCropImageByBox(self._image, splitImageBox)
retval, otsuImage = ImageTool.getOTSUGrayImage(splitImageGray)
otsuImage = ImageTool.convertImgGray2BGR(otsuImage)
lower = (250, 250, 250)
upper = (255, 255, 255)
lcdBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
otsuImage, lower, upper)
lcdBox = ImageTool.getBoxFromBoxCorner(lcdBoxCorner)
self._lcdBox = lcdBox
ImageTool.showBoxInImageByBoxCornerPoint(splitImage, lcdBoxCorner,
"lcd")
def get_text_and_image(self,backgroundColor = None, fontColor = None, fontSizes = None):
""" 生成字符序列和对应的图片数据 图片颜色通道是(R,G,B)-> text,image"""
image = ImageCaptcha(width=self._picBoxWidth, height=self._picBoxHeight,
backgroundColor=backgroundColor, fontColor=fontColor, font_sizes=fontSizes)
captcha_text = self._random_text()
captcha_text = ''.join(captcha_text)
captcha = image.generate(captcha_text)
captcha_image = Image.open(captcha)
if self._imageDepth == 1:
# 将彩色图片转换成灰度图图片
captcha_image = ImageTool.convertImgRGB2Gray(captcha_image)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def get_text_and_image(self,
backgroundColor=None,
fontColor=None,
fontSizes=None):
""" 生成字符序列和对应的图片数据 图片颜色通道是(R,G,B)-> text,image"""
charBoxWidth = 20
charBoxHeight = 50
blackBkgColor = ImageCaptcha.random_bkg_color(0, 65)
fontColor = ImageCaptcha.random_font_color(180, 250)
grayBkgColor = ImageCaptcha.random_bkg_color(130, 220)
image = ImageCaptcha(width=charBoxWidth,
height=charBoxHeight,
backgroundColor=blackBkgColor,
fontColor=fontColor)
captcha_text = self._random_text()
captcha_text = ''.join(captcha_text)
captcha_image = Image.new('RGB',
(self._picBoxWidth, self._picBoxHeight),
grayBkgColor)
dx = 4
dy = 7
desX = dx
for c in captcha_text:
charImage = image.generate(c)
charImage = Image.open(charImage)
captcha_image.paste(charImage, (desX, dy))
desX += dx + charBoxWidth
if self._imageDepth == 1:
# 将彩色图片转换成灰度图图片
captcha_image = ImageTool.convertImgRGB2Gray(captcha_image)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def lcdLighted(lcdImage):
"""
检测lcd 区域图片是否点亮，如果亮返回True，如果不亮返回False
:param lcdImage: cv2 读进来的图片对象
"""
lower = (0, 0, 0)
upper = (50, 50, 50)
blackBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
lcdImage, lower, upper)
# 如果获得的 lcd 区域有黑色区域，说明没有点亮。
if blackBoxCorner is not None:
return False
lcdImageGray = ImageTool.convertImgBGR2Gray(lcdImage)
retval, otsuImage = ImageTool.getOTSUGrayImage(lcdImageGray)
otsuImage = ImageTool.convertImgGray2BGR(otsuImage)
notLcdBoxCorner = ImageTool.getInterestBoxCornerPointByColor(
otsuImage, lower, upper)
ImageTool.showImagePIL(lcdImage, "lcdimage")
return notLcdBoxCorner is None
def fun9():
filename = r"D:\chengxu\python\project\digitRecognise\com\huitong\gasMeterv1\data\img\gasmeterRoller\000041.jpg"
image = Image.open(filename)
image = image.resize((128,64),Image.CUBIC)
ImageTool.showImagePIL(image,"des")
def fun5():
filename = r"D:\chengxu\python\project\digitRecognise\com\huitong\gasMeterv1\data\img\style0\17.jpg"
image = cv2.imread(filename)
ImageTool.showImageCv2(image)
def getBlackMaskBGR():
"""获得黑色区域的 Mask"""
blkLower, blkUpper = ImageTool.getBlackColorRangeBGR()
blackMask = BaseMask(blkLower, blkUpper)
return blackMask
def getRedMaskBGR():
"""获得红色区域的 Mask"""
redLower, redUpper = ImageTool.getRedColorRangeBGR()
redMask = BaseMask(redLower, redUpper)
return redMask