image-processing

Contains a set of Python tools related to image processing.

Requirements

• NumPy
• SciPy
• pandas (optional)
• rasterio
• fiona
• shapely
• Rtree (optional)
• libspatialindex (optional)
• idl-functions

Segmentation analysis example

import numpy
from scipy import ndimage
from image_processing.segmentation import Segments

dims = (1000, 1000)
class_data = numpy.random.randint(0, 10001, dims)
nlabels = ndimage.label(class_data > 5000, output=class_data)
seg = Segments(class_data)

data = numpy.random.ranf(dims)

# pandas DataFrame example
stats = seg.basic_statistics(data, dataframe=True)
print(stats.head(10))
print(stats["Mean"].head(10))

# compound NumPy array example
stats_c = seg.basic_statistics(data)
print(stats_c[0:10])
print(stats_c["Mean"][0:10])

print("Number of segments: {}".format(seg.n_segments))

# data associated with a given segment
seg_data = seg.data(data, segment_id=seg.ids[0])
print(seg_data)

# remove segments containing < 10 pixels
seg.sieve(10)
print("Number of segments: {}".format(seg.n_segments))

# remove segments containing < 30 pixels
seg.sieve(30)
print("Number of segments: {}".format(seg.n_segments))

Contrast enhancement example

import matplotlib.pyplot as plt
from geoh5 import kea
from image_processing import contrast

with kea.open('LS5-2008-02-25.kea') as src:
    data = src.read([5,4,2])

# alternatively if GDAL is built with support for the KEA format
# with rasterio.open('LS5-2008-02-25.kea') as src:
#     data = src.read([5,4,2])

# transpose to display RGB with matplotlib
data = data.transpose(1,2,0)

lp = contrast.linear_percent(data, percent=3, minv=0)
lg = contrast.log(data, minv=0)
sq = contrast.square_root(data, minv=0)
eq = contrast.equalisation(data, minv=0)

fig = plt.figure()
fig.add_subplot(221)
plt.title('Linear 3%')
plt.imshow(lp)
fig.add_subplot(222)
plt.title('Log')
plt.imshow(lg)
fig.add_subplot(223)
plt.title('Square Root')
plt.imshow(sq)
fig.add_subplot(224)
plt.title('Equalisation')
plt.imshow(eq)
plt.show()

which outputs...