def test_run_method_hand_drawn_circle(self):
#
#get a list of points
#
folder_script = os.path.dirname(__file__)
filename_input = "NoisyCircle-HandDrawn-001.png"
file_noisy_line = os.path.join(folder_script, "./data/",
filename_input)
np_image = skimage.io.imread(file_noisy_line, as_gray=True)
lst_points = Util.create_points_from_numpyimage(np_image)
#
#initialize RansalHelper
#
helper = RansacCircleHelper()
helper.threshold_error = 20
helper.threshold_inlier_count = 5
helper.add_points(lst_points)
best_model = helper.run()
#
#Superimpose the new line over the image
#
self.superimpose_circle_over_original_image(file_noisy_line,
best_model)
#
#Assertions
#
delta = 15
self.assertAlmostEqual(best_model.X, 89, delta=delta)
self.assertAlmostEqual(best_model.Y, 78, delta=delta)
self.assertAlmostEqual(best_model.R, 45, delta=delta)
def test_large_circle_50X50_no_noise_1(self):
folder_script=os.path.dirname(__file__)
filename_input="NoisyCircle_x_-10_y_-14.png"
file_noisy_line=os.path.join(folder_script,"./data/",filename_input)
np_image=skimage.io.imread(file_noisy_line,as_gray=True)
lst_points=Util.create_points_from_numpyimage(np_image)
helper=BullockCircleFitting(lst_points)
result:CircleModel =helper.FindBestFittingCircle()
#
#Superimpose the new line over the image
#
folder_results=os.path.join(folder_script,"../out/")
count_of_files=len(os.listdir(folder_results))
filename_results=("%s.%d.png" % (__name__,count_of_files) )
file_result=os.path.join(folder_results,filename_results)
new_points=CircleModel.generate_points_from_circle(result)
np_superimposed=Util.superimpose_points_on_image(np_image,new_points,100,255,100)
skimage.io.imsave(file_result,np_superimposed)
delta=2
self.assertAlmostEquals(result.R, 48.0, delta=delta);
self.assertAlmostEquals(result.X, -10.0, delta=delta);
self.assertAlmostEquals(result.Y, -14.0, delta=delta);
pass
def test_run_method_LargeCircle1_50X50_NoNoise(self):
#
#get a list of points
#
folder_script = os.path.dirname(__file__)
filename_input = "NoisyCircle_x_-10_y_-14.png"
file_noisy_circle = os.path.join(folder_script, "./data/",
filename_input)
np_image = skimage.io.imread(file_noisy_circle, as_gray=True)
lst_points = Util.create_points_from_numpyimage(np_image)
#
#initialize RansalHelper
#
helper = RansacCircleHelper()
helper.gradient_descent_max_iterations = 1000
helper.learning_rate = 0.3
helper.threshold_error = 5
helper.threshold_inlier_count = 15
#helper.max_iterations=400 #100
helper.add_points(lst_points)
best_model = helper.run()
self.superimpose_circle_over_original_image(file_noisy_circle,
best_model)
#
#Assertions
#
delta = 15
self.assertAlmostEqual(best_model.X, -2, delta=delta)
self.assertAlmostEqual(best_model.Y, -5, delta=delta)
self.assertAlmostEqual(best_model.R, 37, delta=delta)
pass
def test_large_circle_50X50_no_noise_2(self):
folder_script = os.path.dirname(__file__)
filename_input = "NoisyCircle_x_6_y_-30_r_118.162.png"
file_noisy_line = os.path.join(folder_script, "./data/",
filename_input)
np_image = skimage.io.imread(file_noisy_line, as_gray=True)
lst_points = Util.create_points_from_numpyimage(np_image)
helper = GradientDescentCircleFitting(None,
lst_points,
learningrate=0.4,
iterations=5000)
result: CircleModel = helper.FindBestFittingCircle()
#
#Superimpose the new line over the image
#
folder_results = os.path.join(folder_script, "../out/")
count_of_files = len(os.listdir(folder_results))
filename_results = ("%s.%d.png" % (__name__, count_of_files))
file_result = os.path.join(folder_results, filename_results)
new_points = CircleModel.generate_points_from_circle(result)
np_superimposed = Util.superimpose_points_on_image(
np_image, new_points, 100, 255, 100)
skimage.io.imsave(file_result, np_superimposed)
delta = 10
self.assertAlmostEquals(result.R, +118.0, delta=delta)
self.assertAlmostEquals(result.X, +06.0, delta=delta)
self.assertAlmostEquals(result.Y, -30.0, delta=delta)
pass
def run(self, image):
print(
f"Going to run RANSAC on a patch square with dimension:{self.patchdimension} and stride:{self.patchstride}"
)
self._image = image
xtractor = ImagePatchExtractor(self.image, self._cropdimension,
self._stride)
patch_results: PatchResults = xtractor.extract_patches()
patchcount_x = patch_results.patch_indices.shape[1]
patchcount_y = patch_results.patch_indices.shape[0]
good_patches: List[_PatchAnalysis] = list()
for x in range(0, patchcount_x):
for y in range(0, patchcount_y):
patchinfo: PatchInfo = patch_results.get_patch_xy(x, y)
img_patchregion = patchinfo.image
lst_all_points = Util.create_points_from_numpyimage(
img_patchregion)
line = self.find_line_using_ransac(lst_all_points,
img_patchregion)
if (line == None):
continue
print("Got a line X=%d Y=%d , line=%s" % (x, y, str(line)))
#add to a collection of patch regions+ransacline
patch_analysis_result = _PatchAnalysis(patchinfo,
lst_all_points, line)
good_patches.append(patch_analysis_result)
pass
print("Total interesting patches with lines found = %d" %
(len(good_patches)))
#you have all the patches that yielded some lines, superimpose on the original
for good_patch in good_patches:
print(str(good_patch))
self.__superimpose(good_patches)
def run_image2matplot(filename):
folder_script = os.path.dirname(__file__)
absolute_path = os.path.join(folder_script, "./input/", filename)
try:
np_image = skimage.io.imread(absolute_path, as_gray=True)
lst_all_points = Util.create_points_from_numpyimage(np_image)
plot_new_points_over_existing_points(lst_all_points, [], "Input data",
"Original points", "")
lrate = 0.3
iterations = 5000
helper = GradientDescentCircleFitting(None,
points=lst_all_points,
learningrate=lrate,
iterations=iterations)
start_time = time.time()
model: CircleModel = helper.FindBestFittingCircle()
new_points = CircleModel.generate_points_from_circle(model)
plot_new_points_over_existing_points(
lst_all_points, new_points, "Gradient descent circle fitting",
"Original points", "Gradient descent")
except Exception as e:
tb = traceback.format_exc()
print("Error:%s while doing RANSAC on the file: %s , stack=%s" %
(str(e), filename, str(tb)))
print("------------------------------------------------------------")
pass
pass
def test_run_with_100x100_image(self):
#
#get a list of points
#
folder_script=os.path.dirname(__file__)
filename_input="Line_100x100.png"
file_noisy_line=os.path.join(folder_script,"./data/",filename_input)
np_image=skimage.io.imread(file_noisy_line,as_gray=True)
lst_points=Util.create_points_from_numpyimage(np_image)
#
#initialize RansalHelper
#
helper1=RansacLineHelper()
helper1.add_points(lst_points)
helper1.max_iterations=20
helper1.min_points_for_model=2
helper1.threshold_error=10
helper1.threshold_inlier_count=3
result_model=helper1.run()
print("RANSAC-complete")
print("Found model %s , polar=%s" % (result_model,result_model.display_polar()))
#
#Superimpose the new line over the image
#
folder_results=os.path.join(folder_script,"../out/")
count_of_files=len(os.listdir(folder_results))
filename_results=("%s.Run.%d.png" % (filename_input,count_of_files) )
file_result=os.path.join(folder_results,filename_results)
x_lower=0
x_upper=np_image.shape[1]-1
y_lower=0
y_upper=np_image.shape[0]-1
#
#Superimpose a line over the inliers only
#
new_points=Util.generate_plottable_points_from_projection_of_points(result_model,result_model.points)
np_superimposed=Util.superimpose_points_on_image(np_image,new_points,100,255,100)
skimage.io.imsave(file_result,np_superimposed)
#11 inlier points in total which give us the good model
self.assertEqual(len(result_model.points),11)
#
#No of detected inliers must be more than or equal to threshold
#
self.assertTrue(len(result_model.points) >= helper1.threshold_inlier_count,"Number of inliers should be >= threshold")
#
#There should be no-duplicates in the RANSAC inlier points
#
set_ids=set(map(lambda x: x.ID, result_model.points))
list_ids=list(map(lambda x: x.ID, result_model.points))
self.assertEqual(len(set_ids),len(list_ids),"Inliers should be unique")
#
#All the RANSAC linlier points must be within the threshold distance from the RANSAC line
#
for inlier_pt in result_model.points:
distance=result_model.compute_distance(inlier_pt)
self.assertTrue(distance < helper1.threshold_error,"Distance of inlier from RANSAC line must be less than threshold")
def run(filename, threshold, inlier, sampling_fraction=0.25, matplot=False):
print("Going to process file:%s" % (filename))
folder_script = os.path.dirname(__file__)
file_noisy_circle = os.path.join(folder_script, "./input/", filename)
try:
np_image = skimage.io.imread(file_noisy_circle, as_gray=True)
#
#Iterate over all cells of the NUMPY array and convert to array of Point classes
#
lst_all_points = Util.create_points_from_numpyimage(np_image)
#
#begin RANSAC
#
helper = RansacCircleHelper()
helper.threshold_error = threshold
helper.threshold_inlier_count = inlier
helper.add_points(lst_all_points)
helper.sampling_fraction = sampling_fraction
best_model = helper.run()
print("RANSAC-complete")
if (best_model == None):
print(
"ERROR! Could not find a suitable model. Try altering ransac-threshold and min inliner count"
)
return
#
#Generate an output image with the model circle overlayed on top of original image
#
now = datetime.datetime.now()
filename_result = ("%s-%s.png" %
(filename, now.strftime("%Y-%m-%d-%H-%M-%S")))
file_result = os.path.join(folder_script, "./out/", filename_result)
#Load input image into array
np_image_result = skimage.io.imread(file_noisy_circle, as_gray=True)
new_points = CircleModel.generate_points_from_circle(best_model)
np_superimposed = Util.superimpose_points_on_image(
np_image_result, new_points, 100, 255, 100)
#Save new image
skimage.io.imsave(file_result, np_superimposed)
print("Results saved to file:%s" % (file_result))
print("------------------------------------------------------------")
if (matplot == True):
plot_new_points_over_existing_points(
lst_all_points, new_points, "Outcome of RANSAC algorithm",
"Original points", "RANSAC")
except Exception as e:
tb = traceback.format_exc()
print("Error:%s while doing RANSAC on the file: %s , stack=%s" %
(str(e), filename, str(tb)))
print("------------------------------------------------------------")
pass
def test_run_with_very_simple_image(self):
#
#get a list of points
#
folder_script=os.path.dirname(__file__)
filename_input="Line_50x30.png"
file_noisy_line=os.path.join(folder_script,"./data/",filename_input)
np_image=skimage.io.imread(file_noisy_line,as_gray=True)
lst_points=Util.create_points_from_numpyimage(np_image)
#
#initialize RansalHelper
#
helper1=RansacLineHelper()
helper1.add_points(lst_points)
helper1.max_iterations=1000
#10000 did not work
helper1.min_points_for_model=2
helper1.threshold_error=3 #10
helper1.threshold_inlier_count=3
result_model=helper1.run()
print("RANSAC-complete")
print("Found model %s , polar=%s" % (result_model,result_model.display_polar()))
#
#Superimpose the new line over the image
#
folder_results=os.path.join(folder_script,"../out/")
count_of_files=len(os.listdir(folder_results))
filename_results=("Line_50x30.%d.png" % (count_of_files) )
file_result=os.path.join(folder_results,filename_results)
x_lower=0
x_upper=np_image.shape[1]-1
y_lower=0
y_upper=np_image.shape[0]-1
new_points=LineModel.generate_points_from_line(result_model,x_lower,y_lower,x_upper,y_upper)
np_superimposed=Util.superimpose_points_on_image(np_image,new_points,100,255,100)
skimage.io.imsave(file_result,np_superimposed)
#
#Asserts!
#
x_intercept=result_model.xintercept()
y_intercept=result_model.yintercept()
self.assertTrue ( x_intercept > 30,"X intercept below threshold")
self.assertTrue ( x_intercept < 50,"X intercept above threshold")
self.assertTrue ( y_intercept > 30,"Y intercept above threshold")
self.assertTrue ( y_intercept < 45,"Y intercept below threshold")
self.assertTrue(len(result_model.points),5)
for pt in result_model.points:
distance_from_line=result_model.compute_distance(pt)
self.assertTrue(distance_from_line <= helper1.threshold_error)
def run_image2image(filename):
print("Going to fit circle in the file:%s" % (filename))
folder_script = os.path.dirname(__file__)
absolute_path = os.path.join(folder_script, "./input/", filename)
try:
np_image = skimage.io.imread(absolute_path, as_gray=True)
lst_all_points = Util.create_points_from_numpyimage(np_image)
lrate = 0.3
iterations = 5000
helper = GradientDescentCircleFitting(None,
points=lst_all_points,
learningrate=lrate,
iterations=iterations)
start_time = time.time()
model: CircleModel = helper.FindBestFittingCircle()
print("--- %s seconds for gradient descent algo ---" %
(time.time() - start_time))
#
#Generate an output image with the model circle overlayed on top of original image
#
now = datetime.datetime.now()
filename_result = ("gradient-descent-%s.png" % (filename))
file_result = os.path.join(folder_script, "./out/", filename_result)
#Load input image into array
np_image_result = skimage.io.imread(absolute_path, as_gray=True)
new_points = CircleModel.generate_points_from_circle(model)
np_superimposed = Util.superimpose_points_on_image(
np_image_result, new_points, 100, 255, 100)
#Save new image
skimage.io.imsave(file_result, np_superimposed)
print("Results saved to file:%s" % (file_result))
print("------------------------------------------------------------")
except Exception as e:
tb = traceback.format_exc()
print("Error:%s while doing RANSAC on the file: %s , stack=%s" %
(str(e), filename, str(tb)))
print("------------------------------------------------------------")
pass
pass
def test_create_points_from_numpyimage(self):
pass
folder_script = os.path.dirname(__file__)
filename = "Util_unittest.png"
file_noisy_line = os.path.join(folder_script, "./data/", filename)
np_image = skimage.io.imread(file_noisy_line, as_gray=True)
height = np_image.shape[0]
width = np_image.shape[1]
lst_points = Util.create_points_from_numpyimage(np_image)
np_shape = np_image.shape
self.assertEqual(len(lst_points), 3)
for pt_any in lst_points:
if pt_any.X == 0 and pt_any.Y == height - 1:
pass
elif (pt_any.X == width - 1 and pt_any.Y == height - 1):
pass
elif (pt_any.X == width - 1 and pt_any.Y == 0):
pass
else:
raise Exception("Point '%s' was not expected." % (pt_any))
def run_ransac(filename):
folder_script = os.path.dirname(__file__)
#Images which did not generate good results:
# NoisyImage_3.png
# NoisyLine-Gaussian-sp-0.80.111.png
file_noisy_line = os.path.join(folder_script, "./input/", filename)
np_image = skimage.io.imread(file_noisy_line, as_gray=True)
#
#Iterate over all cells of the NUMPY array and convert to array of Point classes
#
lst_all_points = Util.create_points_from_numpyimage(np_image)
#
#begin RANSAC
#
ransac_maxiterations = 12000
#12000
#6000
#12000 worked well
ransac_minpoints = 5
#5 worked well
#2 gave very bad results
#20 worked well
ransac_threshold = 5
#25 worked well for 'NoisyLine-Gaussian-sp-0.80.104.png' 15 and 5 did not
#Nothing worked well for 'NoisyLine-Gaussian-sp-0.80.111.png" , tried increasing to 35
#3 for first set when points were much closer
#5 produced too much deviation
ransac_mininliers = 10
helper = RansacLineHelper()
helper.max_iterations = ransac_maxiterations
helper.min_points_for_model = ransac_minpoints
helper.threshold_error = ransac_threshold
helper.threshold_inlier_count = ransac_mininliers
helper.add_points(lst_all_points)
model = helper.run()
#Display the model , you could render over the original picture
print("-------------------------------------")
print("RANSAC-complete")
print("Found model %s , polar=%s" % (model, model.display_polar()))
#
#Generate an output image with the model line
#
filename_noextension = no_extension = os.path.splitext(filename)[0]
now = datetime.datetime.now()
filename_result = ("%s-%s.result.png") % (
filename_noextension, now.strftime("%Y-%m-%d-%H-%M-%S"))
file_result = os.path.join(folder_script, "./out/", filename_result)
#Load input image into array
np_image_result = skimage.io.imread(file_noisy_line, as_gray=True)
new_points = LineModel.generate_points_from_line(
model, 0, 0, np_image_result.shape[1] - 1,
np_image_result.shape[0] - 1)
np_superimposed = Util.superimpose_points_on_image(np_image_result,
new_points, 100, 255,
100)
#Save new image
skimage.io.imsave(file_result, np_superimposed)