def create_database(self):
all = ChineseCharacterGenerator.all(self.CHARACTERS_NUM)
print "start len all", len(all)
all = self.all_to_characters(all)
print "unpacked len all", len(all)
print "finish getting characters"
with open(self.database_file, "w+") as file:
file.write(self.heading);
for i, sign in enumerate(all):
image = sign.pilImageColor
cv_image = PILtoCV(image)
h_desc, v_desc = self.picture_descriptioner(cv_image)
self.write_line(file, i, sign.sign, sign.fontName, h_desc, v_desc)
if i % self.reminder_num == 0:
print "Already written:", i
print "Finally written", i
def computeFeaturesInAllSet(cls, binary=True):
allSet = {}
allSet["X"] = []
allSet["y"] = []
picNum = 0
all = ChineseCharacterGenerator.all()
for sign, pic_in_fonts in all.items():
for pic in pic_in_fonts:
sign_features = cls.features(pic)
sign_features = np.array(sign_features)
allSet["X"].append(sign_features)
allSet["y"].append(sign)
picNum += 1
if picNum % 1000 == 0:
print "Already {} pictures".format(picNum)
with open(cls.allSetFile, "wb") as file:
pickle.dump(allSet, file)
def create_database(self):
all = ChineseCharacterGenerator.all(self.CHARACTERS_NUM)
print "start len all", len(all)
all = self.all_to_characters(all)
print "unpacked len all", len(all)
print "finish getting characters"
k = 1
with open(self.database_file, "w+") as file:
file.write(self.heading)
for i, sign in enumerate(all):
image = sign.pilImageColor
cv_image = PILtoCV(image)
desc = self.picture_descriptioner(cv_image, sign.fontName)
if (
sign.fontName != "WCL-07.ttf"
and sign.fontName != "wts11.ttf"
and self.write_line(file, k, sign.sign, sign.fontName, desc)
):
k = k + 1
if k % self.reminder_num == 0:
print "Already written:", k
print "Finally written", k
def set_chinese_sign_picture_right(self):
self.sign_picture = ChineseCharacterGenerator.nextChar(self.sign_picture)
while self.sign_picture.isBlank():
self.sign_picture = ChineseCharacterGenerator.nextChar(self.sign_picture)
self.refresh_setting_to_new_picture()
def set_chinese_sign_picture_random(self):
picture = ChineseCharacterGenerator.random(self.app.picture_actual_size)
while picture.isBlank():
picture = ChineseCharacterGenerator.random(self.app.picture_actual_size)
self.sign_picture = picture
self.refresh_setting_to_new_picture()
for i in range(len(x_points)):
file.write(str(x_points[i]) + ", ")
for j in range(len(y_points) - 1):
file.write(str(y_points[j]) + ", ")
file.write(str(y_points[-1]) + "\n")
with open('line_detection_statistics.csv', 'w+') as file:
file.write("Id, AllNum, Noise, ")
for i in range(30):
file.write("h" + str(i) + ", ")
for j in range(29):
file.write("v" + str(j) + ", ")
file.write("v29\n")
all = ChineseCharacterGenerator.all()
i = 0
all_num = 0
magic_num = 9
for character, inAllFonts in all.items()[:2000]:
for signPicture in inAllFonts:
cv_image = PILtoCV(signPicture.pilImageColor)
x_points, y_points = LineDetector.getLinesOnPicture(cv_image)
file.write(str(i) + ", " + str(all_num) + ",no, ")
writePoints(file, x_points, y_points)
i += 1
for j in xrange(magic_num):
image = ImageNoiser.Noise(cv_image)
x_points, y_points = LineDetector.getLinesOnPicture(image)
file.write(str(i) + ", " + str(all_num) + ",yes, ")
all = ChineseCharacterGenerator.all()
for sign, pic_in_fonts in all.items():
for pic in pic_in_fonts:
sign_features = cls.features(pic)
sign_features = np.array(sign_features)
allSet["X"].append(sign_features)
allSet["y"].append(sign)
picNum += 1
if picNum % 1000 == 0:
print "Already {} pictures".format(picNum)
with open(cls.allSetFile, "wb") as file:
pickle.dump(allSet, file)
@classmethod
def allFeatures(cls):
with open(cls.allSetFile, "r") as file:
allFeaturesSet = pickle.load(file)
print "all features set loaded"
return allFeaturesSet
if __name__ == "__main__":
sign = ChineseCharacterGenerator.random(30)
CellFeatureExtractor.computeFeaturesInAllSet()
allFeatures = CellFeatureExtractor.allFeatures()
print len(allFeatures["X"][0])
print len(allFeatures["X"])
print len(allFeatures["y"])
