def test_geodesic_distance2d():
I = np.asarray(Image.open('data/img2d.png').convert('L'), np.float32)
S = np.zeros_like(I, np.uint8)
S[100][100] = 1
t0 = time.time()
D1 = geodesic_distance.geodesic2d_fast_marching(I, S)
t1 = time.time()
D2 = geodesic_distance.geodesic2d_raster_scan(I, S)
dt1 = t1 - t0
dt2 = time.time() - t1
print "runtime(s) of fast marching {0:}".format(dt1)
print "runtime(s) of raster scan {0:}".format(dt2)
plt.subplot(1, 3, 1)
plt.imshow(I, cmap='gray')
plt.autoscale(False)
plt.plot([100], [100], 'ro')
plt.axis('off')
plt.title('input image')
plt.subplot(1, 3, 2)
plt.imshow(D1)
plt.axis('off')
plt.title('fast marching')
plt.subplot(1, 3, 3)
plt.imshow(D2)
plt.axis('off')
plt.title('ranster scan')
plt.show()
def process_frames(self, data):
input_temp = data[0]
indices = np.where(np.isnan(input_temp))
input_temp[indices] = 0.0
if (np.sum(data[1]) > 0):
geoDist = geodesic_distance.geodesic2d_raster_scan(
input_temp, data[1], self.lambda_par, self.iterations)
else:
geoDist = np.float32(np.zeros(np.shape(data[0])))
maxvalues = [np.max(geoDist)]
return [geoDist, np.array([maxvalues])]
def geodesic_distance_2d(I, S, lamb, iter):
'''
get 2d geodesic disntance by raser scanning.
I: input image
S: binary image where non-zero pixels are used as seeds
lamb: weighting betwween 0.0 and 1.0
if lamb==0.0, return spatial euclidean distance without considering gradient
if lamb==1.0, the distance is based on gradient only without using spatial distance
iter: number of iteration for raster scanning.
'''
return geodesic_distance.geodesic2d_raster_scan(I, S, lamb, iter)
def test_geodesic_distance2d():
I = np.asarray(Image.open('../data/img2d.png').convert('L'), np.float32)
S = np.zeros_like(I, np.uint8)
S[100][100] = 1
t0 = time.time()
D1 = geodesic_distance.geodesic2d_fast_marching(I, S)
t1 = time.time()
D2 = geodesic_distance.geodesic2d_raster_scan(I, S)
dt1 = t1 - t0
dt2 = time.time() - t1
print "runtime(s) of fast marching {0:}".format(dt1)
print "runtime(s) of raster scan {0:}".format(dt2)
plt.subplot(1, 3, 1)
plt.imshow(I)
plt.subplot(1, 3, 2)
plt.imshow(D1)
plt.subplot(1, 3, 3)
plt.imshow(D2)
plt.show()
#python setup.py build #python setup.py install #python import geodesic_distance import scipy.misc import numpy as np import matplotlib.pyplot as plt import scipy.ndimage as nd I = np.float32(scipy.misc.ascent()) / 255. Cost = nd.gaussian_gradient_magnitude(I, 1.) Seed = np.zeros_like(Cost) Seed[300, 300] = 1 A = geodesic_distance.geodesic2d_raster_scan(Cost, np.uint8(Seed)) plt.imshow(A) plt.show() I3 = np.stack([I[::2, ::2], I[::-2, ::2]], 2) I3 = np.concatenate([I3, I3[:, :, ::-1]], 2) I3 = np.concatenate([I3, I3[:, :, ::-1]], 2) I3 = np.concatenate([I3, I3[:, :, ::-1]], 2) I3 = np.concatenate([I3, I3[:, :, ::-1]], 2) I3 = np.concatenate([I3, I3[:, :, ::-1]], 2) Cost3 = nd.gaussian_gradient_magnitude(I3, 1.) Seed3 = np.zeros_like(Cost3) Seed3[150, 150, 32] = 1 A3 = geodesic_distance.geodesic3d_raster_scan(Cost3, np.uint8(Seed3), 1., 4.) plt.imshow(A3[:, :, 32]) plt.show()