def test_dct():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100. * data / np.amax(data)
y = data + np.random.normal(0, 20, data.shape)
y = y.astype(np.float32)
out = dct_8x8(y, 60)
outs = [calcPSNR(data, dct_8x8(y, s)) for s in np.linspace(10, 60, 20)]
return out
def test_bilateral():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100.*data/np.amax(data)
y = data + np.random.normal(0,20,data.shape)
y = y.astype(np.float32)
outs = [calcPSNR(data,bilateral(y,2,sigs)) for sigs in np.linspace(1.,200,20)]
return outs
def test_dct():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100.*data/np.amax(data)
y = data + np.random.normal(0,20,data.shape)
y = y.astype(np.float32)
out = dct_8x8(y,60)
outs = [calcPSNR(data,dct_8x8(y,s)) for s in np.linspace(10,60,20)]
return out
def test_bilateral():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100. * data / np.amax(data)
y = data + np.random.normal(0, 20, data.shape)
y = y.astype(np.float32)
outs = [
calcPSNR(data, bilateral(y, 2, sigs))
for sigs in np.linspace(1., 200, 20)
]
return outs
def test_nlm():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100. * data / np.amax(data)
y = data + np.random.normal(0, 20, data.shape)
y = y.astype(np.float32)
out = nlm(y, 2, 3, 25.)
sigs = np.linspace(2, 70, 10)
outs = [calcPSNR(data, nlm(y, 2, 3, s)) for s in sigs]
print "nlm: sig_max = %s" % sigs[np.argmax(outs)]
return outs
def test_nlm():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100.*data/np.amax(data)
y = data + np.random.normal(0,20,data.shape)
y = y.astype(np.float32)
out = nlm(y,2,3,25.)
sigs = np.linspace(2,70,10)
outs = [calcPSNR(data,nlm(y,2,3,s)) for s in sigs]
print "nlm: sig_max = %s" %sigs[np.argmax(outs)]
return outs
def test_nlm_fast():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100. * data / np.amax(data)
y = data + np.random.normal(0, 20, data.shape)
y = y.astype(np.float32)
out = nlm_fast(y, 2, 3, 5.)
sigs = np.linspace(3, 70, 30)
outs = [calcPSNR(data, nlm_fast(y, 2, 3, s)) for s in sigs]
bests = []
for s0 in np.linspace(2, 20, 10):
y = data + np.random.normal(0, s0, data.shape)
y = y.astype(np.float32)
ind = np.argmax([calcPSNR(data, nlm_fast(y, 2, 3, s)) for s in sigs])
print ind
bests.append([s0, sigs[ind]])
print "nlm fast: sig_max = %s" % sigs[np.argmax(outs)]
return bests
def test_nlm_fast():
from imgtools import test_images, calcPSNR
data = test_images.lena()
data = 100.*data/np.amax(data)
y = data + np.random.normal(0,20,data.shape)
y = y.astype(np.float32)
out = nlm_fast(y,2,3,5.)
sigs = np.linspace(3,70,30)
outs = [calcPSNR(data,nlm_fast(y,2,3,s)) for s in sigs]
bests = []
for s0 in np.linspace(2,20,10):
y = data + np.random.normal(0,s0,data.shape)
y = y.astype(np.float32)
ind=np.argmax([calcPSNR(data,nlm_fast(y,2,3,s)) for s in sigs])
print ind
bests.append([s0,sigs[ind]])
print "nlm fast: sig_max = %s" %sigs[np.argmax(outs)]
return bests
