def test_animated_image(self):
frames = []
generated_frames = []
expected = []
for i in self.frame_count:
generated_frames.append(cv2.imread("test_2_generate_circle_{}.png".format(i), -1))
frames.append(cv2.imread("test_2_frame_{}.png".format(i), 1))
expected.append(cv2.imread("test_2_animated_image_{}.png".format(i), -1))
obj = ip.ImageProcessing()
obj.filename = "test_2"
obj.gif_divorce = MagicMock(return_value=frames)
obj.generate_circle = MagicMock(side_effect=generated_frames)
results = obj.animated_image()
for i in self.frame_count:
with self.subTest(i=i):
boolean = np.all(results[i] == expected[i])
if not boolean:
self.assertTrue(boolean)
self.assertTrue(boolean)
def test_generate_circle(self):
frames = [None]*6
desired = [None]*6
for i in self.frame_count:
desired[i] = cv2.imread("test_2_generate_circle_{}.png".format(i), -1)
obj = ip.ImageProcessing()
for i in self.frame_count:
with self.subTest(i=i):
image = cv2.imread("test_2_crop_circle_{}.png".format(i), 1)
obj.crop_circle = MagicMock(return_value=image)
image = cv2.imread("test_2_add_alpha_{}.png".format(i), -1)
obj.add_alpha = MagicMock(return_value=image)
frame = obj.generate_circle(frames)
boolean = np.all(frame == desired[i])
if not boolean:
self.assertTrue(boolean)
self.assertTrue(boolean)
def test():
img = cv2.imread("./img/153g.bmp", 0)
ip = ImageProcessing.ImageProcessing(img)
ip.doHoughTrans()
img = ip.cannyImg[ip.C - 125:ip.C + 125, :]
waveHeight, waveLenght = dp.getWaveParaByPeak(img)
def test_get_frames(self):
gif = Image.open("test_2")
frames = []
obj = ip.ImageProcessing()
try:
i = 0
while True:
gif.seek(i)
imgframe = gif.copy()
if i == 0:
palette = imgframe.getpalette()
else:
imgframe.putpalette(palette)
frames.append(imgframe)
i += 1
except EOFError:
pass
for i in self.frame_count:
frames[i] = obj.get_frames(frames[i])
for i in self.frame_count:
comparable = cv2.imread("test_2_frame_{}.png".format(i), 1)
if np.all(comparable != frames[i]):
self.assertTrue(False)
self.assertTrue(True)
def __init__(self): self.imageProcessing = ImageProcessing.ImageProcessing() self.robotMovement = RobotMovement.RobotMovement() self.Sensors = Sensors.Sensors() self.wiiRemote = WiiRemote.WiiRemote() self.running = True self.mainloop()
def test_get_size(self):
image = cv2.imread("test_1", 1)
obj = ip.ImageProcessing()
dimensions = obj.get_size(image)
self.assertEqual(dimensions, (480, 640))
def onClickedBtnFocus(self):
self.ci.setRoiToDefault()
self.ci.getOneFrame()
self.ci.endCapture()
ip = ImageProcessing.ImageProcessing(self.ci.img)
ip.doHoughTrans()
self.ci.setRoi(self.ROIRANGE, ip.C - self.ROIRANGE/2)
self.ci.startCapture()
def test_add_alpha(self):
input_image = cv2.imread("test_1_crop_circle_0.png", 1)
result_image = cv2.imread("test_1_add_alpha_0.png", -1)
obj = ip.ImageProcessing()
result = obj.add_alpha(input_image)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_get_size(self):
obj = ip.ImageProcessing()
for i in self.frame_count:
with self.subTest(i=i):
image = cv2.imread("test_2_frame_{}.png".format(i), 1)
dimensions = obj.get_size(image)
self.assertEqual(dimensions, (212, 314))
def test_nonanimated_image(self):
filename = "test_1"
obj = ip.ImageProcessing()
result = obj.image_processing(filename)
comparable = cv2.imread("test_1_unanimated_image_0.png", -1)
state = np.all(comparable == result[0])
self.assertTrue(state)
def test_trim_square(self):
input_image = cv2.imread("test_1_pre_trim_square_0.png", 1)
result_image = cv2.imread("test_1_trim_square_0.png", 1)
obj = ip.ImageProcessing()
center = 240, 320
result = obj.trim_square(input_image, center, 240)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_find_min(self):
x = [212, 314, 0, -1]
y = [314, 212, 0, -2]
result = [0, 1, 1, 1]
obj = ip.ImageProcessing()
for i in range(len(result)):
if obj.find_min(x[i], y[i]) != result[i]:
self.assertTrue(False)
self.assertTrue(True)
def initCamera(self):
self.ci = Camera.CameraInterface() # 获取界面中的相机接口
self.ci.openCamera()
self.ci.setAttribute()
firstFrame = self.ci.getFirstFrame()
ip = ImageProcessing.ImageProcessing(firstFrame)
if ip.doHoughTrans():
self.HOUGH_LINE_Y = ip.C
else:
self.HOUGH_LINE_Y = self.IMG_HEIGHT/2
self.ci.setRoi(self.ROIRANGE, ip.C-self.ROIRANGE/2)
self.ci.startCapture()
def test_image_processing(self):
result_image = cv2.imread("test_1_image_processing_0.png", -1)
obj = ip.ImageProcessing()
unanimated_image_result = cv2.imread("test_1_generate_circle_0.png", -1)
obj.generate_circle = MagicMock(return_value=unanimated_image_result)
result = obj.image_processing("test_1")
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_add_alpha(self):
for i in self.frame_count:
with self.subTest(i=i):
input_image = cv2.imread("test_2_crop_circle_{}.png".format(i), 1)
result_image = cv2.imread("test_2_add_alpha_{}.png".format(i), -1)
obj = ip.ImageProcessing()
result = obj.add_alpha(input_image)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_animated_image(self):
filename = "test_2"
obj = ip.ImageProcessing()
result = obj.image_processing(filename)
for i, res in enumerate(result):
with self.subTest(i=i):
comparable = cv2.imread(
"test_2_animated_image_{}.png".format(i), -1)
state = np.all(comparable == res)
self.assertTrue(state)
def main():
classifier = ClassifyImage(
'centernet_resnet50_v1_fpn_512x512_coco17_tpu-8/saved_model',
'mscoco_label_map.pbtxt')
camera = ImageProcessing(960, 720)
while True:
coords = classifier.find_person(camera.frameToNumpy())
#print(coords)
if not coords is False:
print(coords)
else:
print("nothing")
time.sleep(2)
self.camera.release()
def test_trim_square(self):
for i in self.frame_count:
with self.subTest(i=i):
input_image = cv2.imread("test_2_pre_trim_square_{}.png".format(i), 1)
result_image = cv2.imread("test_2_trim_square_{}.png".format(i), 1)
obj = ip.ImageProcessing()
center = 106, 157
result = obj.trim_square(input_image, center, 106)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_image_processing(self):
for i in self.frame_count:
with self.subTest(i=i):
result_image = cv2.imread("test_2_image_processing_{}.png".format(i), -1)
obj = ip.ImageProcessing()
unanimated_image_result = cv2.imread("test_2_generate_circle_{}.png".format(i), -1)
obj.generate_circle = MagicMock(return_value=unanimated_image_result)
result = obj.image_processing("test_2")
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def displayEdgeImg(self, oriImg): # 边缘显示区域的处理
if self.isMeasuring == 0: # 打开系统，但是未选择测量时
# 边缘显示区域也显示原始图像
img = q2n.gray2qimage(oriImg)
img = img.scaled(self.IMG_WIDTH/2, self.ROIRANGE/2, Qt.KeepAspectRatio)
self.label_canny.setPixmap(QPixmap.fromImage(img)) # 在原始图像显示区域写入图像
return
elif self.isMeasuring == 2: # 仅仅用于演示和显示边缘图像，不对图像进行运算处理
ip = ImageProcessing.ImageProcessing(oriImg)
img = cv2.resize(ip.cannyImg, (self.IMG_WIDTH / 2, self.ROIRANGE / 2), cv2.INTER_NEAREST)
img = q2n.gray2qimage(img) # 此方法显示会将边缘放大，但实际处理的数据不会改变
self.label_canny.setPixmap(QPixmap.fromImage(img)) # 在边缘显示区域写入图像
return
'''单击【自动测量按钮后】 self.isMeasuring==1的情况'''
ip = ImageProcessing.ImageProcessing(oriImg)
topCurve, bottomCurve = dp.getEdgeList(ip.cannyImg)
# waveHeightByPeak, waveLengthByPeak = dp.getWaveParaByPeak(topCurve, bottomCurve)
waveHeightBySin, waveLengthBySin = dp.getWaveParaBySin(topCurve, bottomCurve)
# 显示钢丝波高的值
waveShow = "%0.1f" % (waveHeightBySin * 6.6188 - 7.356)
waveShow += "um"
self.label_wave_height.setText(QString(str(waveShow)))
img = cv2.resize(ip.cannyImg, (self.IMG_WIDTH / 2, self.ROIRANGE / 2), cv2.INTER_NEAREST)
img = q2n.gray2qimage(img) # 此方法显示会将边缘放大，但实际处理的数据不会改变
self.label_canny.setPixmap(QPixmap.fromImage(img)) # 在边缘显示区域写入图像
self.cntImgMeasuring += 1
self.progressBar.setValue(self.cntImgMeasuring) # 更新进度条当前进度
self.waveHeightList.append(waveHeightBySin)
self.waveLengthList.append(waveLengthBySin) # 计数器和数据记录
if self.cntImgMeasuring > self.rotatePeriod: # 测量时间大于设定值
self.completeMeasure() # 测量完成进行相应得清零和整理工作
def start(self, getter):
self.keyboard_input = getter
self.picture_to_show = self.keyboard_input.get()
self.picture_storage = PictureStorage()
self.settings = Settings.Settings()
self.camera = Camera()
self.tools = ImageProcessing.ImageProcessing(self.picture_storage, self.settings)
self.window = Window(self.settings, "Bilder")
self.main_window = Window(self.settings, "Taschenrechner")
self.calculator = Calculator.Calculator()
self.gui = GUI.GUI(self.picture_storage, self.settings)
self.history = History()
self.stage = 0
self.buttons = ButtonGenerator(self.picture_storage)
self.delete_history = History()
self._run()
def test_crop_circle(self):
input_image = cv2.imread("test_1", 1)
result_image = cv2.imread("test_1_crop_circle_0.png", 1)
trim_square_result = cv2.imread("test_1_trim_square_0.png", 1)
obj = ip.ImageProcessing()
obj.get_size = MagicMock(return_value=(480, 640))
obj.find_min = MagicMock(return_value=0)
obj.trim_square = MagicMock(return_value=trim_square_result)
result = obj.crop_circle(input_image)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def main():
ci = Camera.CameraInterface()
ci.openCamera()
ci.setAttribute(mode="Continuous")
img = ci.getFirstFrame()
ip = ImageProcessing.ImageProcessing(img)
ip.doHoughTrans()
print "ip.C:", ip.C
ci.setRoi(250, ip.C - 125)
ci.startCapture()
while True:
ci.getOneFrame()
cv2.imshow("test", ci.img)
k = cv2.waitKey(1)
if k == 0x1b:
cv2.destroyAllWindows()
break
def test_gif_divorce(self):
gif = Image.open("test_2")
frames = []
obj = ip.ImageProcessing()
frames = obj.gif_divorce(gif)
for i in self.frame_count:
with self.subTest(i=i):
comparable = cv2.imread("test_2_frame_{}.png".format(i), 1)
boolean = np.all(comparable == frames[i])
if not boolean:
self.assertTrue(boolean)
self.assertTrue(boolean)
def __init__(self, settings, connection, cropscoordinates, history):
self.settings = settings
self.connection = connection
self.cropscoordinates = cropscoordinates
self.history = history
self.imageprocessing = ImageProcessing.ImageProcessing(settings)
self.pool = ThreadPool()
self.local = not (self.settings.client_server)
if self.local:
if self.settings.verbosity >= 2:
print("Evaluation init, local model of Darkflow")
self.local_model = self.init_local()
if self.settings.precompute_attention_evaluation:
# initialize hashmaps / dictionaries
self.precomputations_started = {} # <frame_number> => thread id
self.precomputations_finished = {} # <frame_number> => results
def test_crop_circle(self):
for i in self.frame_count:
with self.subTest(i=i):
input_image = cv2.imread("test_2_frame_{}.png".format(i), 1)
result_image = cv2.imread("test_2_crop_circle_{}.png".format(i), 1)
trim_square_result = cv2.imread("test_2_trim_square_{}.png".format(i), 1)
obj = ip.ImageProcessing()
obj.get_size = MagicMock(return_value=(212, 314))
obj.find_min = MagicMock(return_value=0)
obj.trim_square = MagicMock(return_value=trim_square_result)
result = obj.crop_circle(input_image)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def test_generate_circle(self):
input_image = cv2.imread("test_1", 1)
image = []
image.append(input_image)
result_image = cv2.imread("test_1_generate_circle_0.png", -1)
obj = ip.ImageProcessing()
crop_circle_result = cv2.imread("test_1_crop_circle_0.png", 1)
obj.crop_circle = MagicMock(return_value=crop_circle_result)
add_alpha_result = cv2.imread("test_1_add_alpha_0.png", -1)
obj.add_alpha = MagicMock(return_value=add_alpha_result)
result = obj.generate_circle(image)
boolean = np.all(result == result_image)
self.assertTrue(boolean)
def main():
img = cv2.imread("./img/noise2.bmp", 0)
ip = ImageProcessing.ImageProcessing(img)
ip.doHoughTrans()
img = ip.cannyImg[ip.C - 125:ip.C + 125, :]
topCurve, bottomCurve = dp.imgToList(img)
maxTabTopCurve, minTabTopCurve = dp.peakdet(topCurve,
0.5) # 通过函数直接找到极值点，包含了一些噪点。
maxTabBottomCurve, minTabBottomCurve = dp.peakdet(bottomCurve, 0.5)
peakTopCurve = dp.getWavePeak(maxTabTopCurve, line="TopCurve")
peakBottomCurve = dp.getWavePeak(minTabBottomCurve, line="BottomCurve")
waveHeight, waveLength = dp.peakPointToWavePara(peakTopCurve,
peakBottomCurve)
plt.scatter(range(len(topCurve)), topCurve, c="r", s=1)
plt.scatter(range(len(bottomCurve)), bottomCurve, c="g", s=2)
plt.plot(range(2048), [peakTopCurve.mean(axis=0)[1]] * 2048)
plt.plot(range(2048), [peakBottomCurve.mean(axis=0)[1]] * 2048)
for i in range(peakTopCurve.shape[0]):
x = [peakTopCurve[i, 0]] * 2
y = [
peakTopCurve.mean(axis=0)[1] + 5,
peakTopCurve.mean(axis=0)[1] - 5
]
plt.plot(x, y)
for i in range(peakBottomCurve.shape[0]):
x = [peakBottomCurve[i, 0]] * 2
y = [
peakBottomCurve.mean(axis=0)[1] + 5,
peakBottomCurve.mean(axis=0)[1] - 5
]
plt.plot(x, y)
plt.show()
def extract_ines_of_text_from_image_and_match_label(imageArray, text,fontname, folderSave):
lines_txt = text.split('\n')
num_lines_txt = len(lines_txt)
print(num_lines_txt)
num_line=0
newImagesForTrain = []
width = imageArray.shape[1]
height = imageArray.shape[0]
boundries = ImageProcessing.GetLineBounds(imageArray)
print(len(boundries))
num_lines_image = len(boundries)
for i in range(len(boundries)):
x, y, w, h = boundries[i]
cutImage = imageArray[max(0, min(y, y+h)-NUMPIXELS):min(max(y, y+h)+NUMPIXELS,height) , max(0, min(x, x+w)-NUMPIXELS) :min(width, max(x, x+w)+NUMPIXELS)]
while lines_txt[num_line]=="" or lines_txt[num_line]==" " and num_line<num_lines_txt-1:
num_line+=1
if num_line<=num_lines_txt-1:
Label = lines_txt[num_line]
num_line += 1
newImagesForTrain.append(ImageProcessing.ImageProcessing(cutImage, imagePath=None, handwrite_ID=fontname, Label = Label))
Insert_to_folder(newImagesForTrain, folderSave)
def begin(self):
#print("Starting processing")
imageProcessor = ImageProcessing(self.meeting, self.arrayStudents,
self.size)
confidenceMatrix = imageProcessor.processImageAndGetConfidenceMatrix()
#print("done confidence:")
#for i in range(len(confidenceMatrix)):
# print(confidenceMatrix[i])
useDelete = True
useSocial = True
matcher = Matcher(self.meeting, self.arrayStudents)
attendance = matcher.matchStudents(confidenceMatrix, useDelete,
useSocial)
output = Output(self.meeting, self.arrayStudents,
self.meeting.getCroppedFaces(), attendance)
imageOrginalWithAttedance, imageAttendancePath = output.createAndWriteAttendacePiture(
"Matching")
#write the attedance picture to the db
self.database.writeImageWithAttendance(
self.db, imageOrginalWithAttedance,
self.meeting.getMeetingDirectory(), imageAttendancePath)
#write the attedance to the db
self.database.writeAttendance(self.db, self.arrayStudents)
#write the average social matrix to the db if its
if (self.meeting.getFirstMeeting() == True):
finalAverageSocialMatrix = output.findSocialMatrixFirstMeeting()
else:
finalAverageSocialMatrix = output.findAverageSocialMatrix()
self.database.writeSocialMatrix(self.db, self.meeting,
finalAverageSocialMatrix,
self.arrayStudents)
output.printAttendance()
if (1 == 2):
useDelete = False
useSocial = True
matcher = Matcher(self.meeting, self.arrayStudents)
attendance = matcher.matchStudents(confidenceMatrix, useDelete,
useSocial)
output4 = Output(self.meeting, self.arrayStudents,
self.meeting.getCroppedFaces(), attendance)
output4.createAndWriteAttendacePiture("SocialNoDelete")
output4.printAttendance()
useDelete = True
useSocial = False
matcher = Matcher(self.meeting, self.arrayStudents)
attendance = matcher.matchStudents(confidenceMatrix, useDelete,
useSocial)
output3 = Output(self.meeting, self.arrayStudents,
self.meeting.getCroppedFaces(), attendance)
output3.createAndWriteAttendacePiture("Social")
output3.printAttendance()
useDelete = False
useSocial = False
matcher = Matcher(self.meeting, self.arrayStudents)
attendance = matcher.matchStudents(confidenceMatrix, useDelete,
useSocial)
output2 = Output(self.meeting, self.arrayStudents,
self.meeting.getCroppedFaces(), attendance)
output2.createAndWriteAttendacePiture("MatchingNoDelete")
output2.printAttendance()
