def loadBSQ(path = '/home/crob/HyperSpec_Data/WBC v ALL/WBC25', debug=False):
d31 = []
d31_norm = []
d25 = []
d25_norm = []
l25 = []
l = []
l3 = []
lam = []
for root, dirs, files in os.walk(path):
print(dirs)
for name in sorted(files): #os walk iterates arbitrarily, sort fixes it
print(name)
if name.endswith(".png"):
# Import label image
im = np.array(Image.open(os.path.join(root,name)),'f')
print np.shape(im)
im = im[:,:,0:3] # > 250
# generate a mask for 3x3 conv layer (probably not needed)
#conv3bw = signal.convolve2d(bw, np.ones([22,22],dtype=np.int), mode='valid') >= 464
print(np.shape(im))
#p = open(name+'_22sqMask.png','wb')
#w = png.Writer(255)
#bw = np.flipud(bw)
im = np.flipud(im)
#l3.append(np.reshape(conv3bw, ))
#l.append(np.reshape(bw, 138659))
l.append(im)
print(np.shape(im))
print("Name = " + name)
if name.endswith(".bsq"):
bs = bsq.readbsq(os.path.join(root,name))
print(np.shape(bs[0]))
print(len(bs[1]))
#separate bsq files by prism
if len(bs[1]) == 31:
print('BSQ is size 31')
print(len(bs[1]))
lam = bs[1]
#d31.append(np.reshape(np.transpose(bs[0], (1, 2, 0)), 4298429))
d31.append(bs[0].astype(np.float32))
d31_norm.append(bs[0].astype(np.float32)/np.amax(bs[0]))
if len(bs[1]) == 25:
print('BSQ is size 25')
print(len(bs[1]))
lam = bs[1]
d25.append(bs[0].astype(np.float32))
d25_norm.append(bs[0].astype(np.float32)/np.amax(bs[0]))
#d25.append(np.reshape(bs[0],[138659,25]).astype(np.float32))
# old don't use #d25.append(np.reshape(np.transpose(bs[0], (1, 2, 0)), 3466475))
out = collections.namedtuple('examples', ['data31', 'data31_norm', 'data25', 'data25_norm', 'labels', 'lambdas'])
o = out(data31=np.dstack(d31),data31_norm=np.dstack(d31_norm), data25=d25, data25_norm=d25_norm, labels=np.dstack(l), lambdas=lam) #np.vstack(d25), labels=np.hstack(l)
return o
def readBsqToObj(path,mPath):
dCube = dataCube()
(dCube.HSData, dCube.Lambdas) = BSQ.readbsq(dir)
#add in line to read in mask image
return dCube
scipy.misc.imsave(pathName + '\\processed_masks\\' + maskNames[i][2:-8] + '.png', im)
# im = skimage.io.imread(mFile, plugin='tifffile')
A = dataCube(maskNames[i][2:-8], None, None, im)
print(A.MaskData.shape)
maskList.append(A)
## insert for loop here (add if statement)
#This loop runs when looking in BSQ Directory
if dirs == 'bsq':
for i in range(0,bsqCount):
dir = os.path.dirname(os.path.join(sDir,objNames[i]))
print(str(maskCount))
if not os.path.exists(dir+'/TIF/'):
os.mkdir(dir+'/TIF/')
#outfile = dir + '/TIF/' + os.path.basename(os.path.join(sDir,objNames[i]))[:-4]
(dcb, lambdas) = BSQ.readbsq(os.path.join(sDir,objNames[i]))
#print('dcb.shape=', dcb.shape)
#print('dcb.dtype=', dcb.dtype)
#print('lambdas=', lambdas)
# Convert type double pixel data to type uint16 for exporting to TIFF.
minval = np.amin(dcb)
maxval = np.amax(dcb)
#print('original dcb: minval=%f, maxval=%f' % (minval, maxval))
new_dcb = np.int16((dcb - minval) * (2**15 - 1) / np.float64(maxval - minval))
#print('new_dcb.dtype=', new_dcb.dtype)
#print('rescaled dcb: minval=%i, maxval=%i' % (amin(new_dcb), amax(new_dcb)))
img = new_dcb[:, :, 0]
def loadBSQ(path='/home/crob/HyperSpec_Data/WBC v ALL/WBC25', debug=False):
d31 = []
d31_norm = []
d25 = []
d25_norm = []
l25 = []
l = []
l3 = []
lam = []
for root, dirs, files in os.walk(path):
print(dirs)
for name in sorted(
files): #os walk iterates arbitrarily, sort fixes it
print(name)
if name.endswith(".png"):
# Import label image
im = np.array(Image.open(os.path.join(root, name)), 'f')
print np.shape(im)
im = im[:, :, 0:3] # > 250
# generate a mask for 3x3 conv layer (probably not needed)
#conv3bw = signal.convolve2d(bw, np.ones([22,22],dtype=np.int), mode='valid') >= 464
print(np.shape(im))
#p = open(name+'_22sqMask.png','wb')
#w = png.Writer(255)
#bw = np.flipud(bw)
im = np.flipud(im)
#l3.append(np.reshape(conv3bw, ))
#l.append(np.reshape(bw, 138659))
l.append(im)
print(np.shape(im))
print("Name = " + name)
if name.endswith(".bsq"):
bs = bsq.readbsq(os.path.join(root, name))
print(np.shape(bs[0]))
print(len(bs[1]))
#separate bsq files by prism
if len(bs[1]) == 31:
print('BSQ is size 31')
print(len(bs[1]))
lam = bs[1]
#d31.append(np.reshape(np.transpose(bs[0], (1, 2, 0)), 4298429))
d31.append(bs[0].astype(np.float32))
d31_norm.append(bs[0].astype(np.float32) / np.amax(bs[0]))
if len(bs[1]) == 25:
print('BSQ is size 25')
print(len(bs[1]))
lam = bs[1]
d25.append(bs[0].astype(np.float32))
d25_norm.append(bs[0].astype(np.float32) / np.amax(bs[0]))
#d25.append(np.reshape(bs[0],[138659,25]).astype(np.float32))
# old don't use #d25.append(np.reshape(np.transpose(bs[0], (1, 2, 0)), 3466475))
out = collections.namedtuple('examples', [
'data31', 'data31_norm', 'data25', 'data25_norm', 'labels', 'lambdas'
])
o = out(data31=np.dstack(d31),
data31_norm=np.dstack(d31_norm),
data25=d25,
data25_norm=d25_norm,
labels=np.dstack(l),
lambdas=lam) #np.vstack(d25), labels=np.hstack(l)
return o
