### Powerpoint from Keith`s presentation

```Ditch ArcGIS and get to know NUMPY
and SCIPY





ArcGIS 10x
Python 2.7.2 or better
Numpy (installed by ArcGIS)
SciPy
Some luck




Check projection consistency
Find a corner point based on a reference raster
Determine number of rows and columns to







Write raster image as masked array.
>>> import numpy.ma as ma
>>> x = [0.,1.,-9999.,3.,4.]
>>> mx = ma.masked_values (x, -9999.)
>>> print mx.mean()
2.0



What is scipy? SciPy (pronounced “Sigh Pie”)
is a Python-based ecosystem of open-source
software for mathematics, science, and
engineering.
Ndimage. Multidimentional image processing.
http://docs.scipy.org/doc/scipy/reference/n
dimage.html
How does one build a custom filter?
Generic_filter





Input – Input array to filter
Function – Function to apply at each element
Footprint - footprint is a boolean array that
specifies (implicitly) a shape, but also which of
the elements within this shape will get passed
to the filter function.
Mode - {‘reflect’, ‘constant’, ‘nearest’, ‘mirror’,
‘wrap’}
Cval - Value to fill past edges of input if mode is
‘constant’. Default is 0.0






Since the output elements are a function of elements in the
neighborhood of the input elements, the borders of the array need
to be dealt with appropriately by providing the values outside the
borders. This is done by assuming that the arrays are extended
beyond their boundaries according certain boundary conditions.
In the functions described below, the boundary conditions can be
selected using the mode parameter which must be a string with the
name of the boundary condition. Following boundary conditions
are currently supported:
“nearest” Use the value at the boundary [1 2 3]->[1 1 2 3 3]
“wrap” Periodically replicate the array [1 2 3]->[3 1 2 3 1]
“reflect” Reflect the array at the boundary [1 2 3]->[1 1 2 3 3]
“constant” Use a constant value, default is 0.0 [1 2 3]->[0 1 2 3 0]
The “constant” mode is special since it needs an additional
parameter to specify the constant value that should be used.


















kernel = np.array([], dtype=bool)
# build kernel from radius and input raster cell size
for row in range(-window, window + 1):
for col in range(-window, window + 1):
Xdist = abs(col * referenceDesc.meanCellHeight)
Ydist = abs(row * referenceDesc.meanCellHeight)
totalDist = pow(Xdist**2 + Ydist**2, 0.5)
kernel = np.append(kernel, [False])
else:
kernel = np.append(kernel, [True])
kernel.shape = (pow(kernel.size,0.5),pow(kernel.size,0.5))
print str(kernel.size) + " pixels in circular kernel"
return(kernel)
OR…
kernelFP = np.array([[False, True, False], [True, True, True], [False, True, False]], dtype=bool)







# function to generate average value in kernel that is not
NODATA
def meanFunc(kernel):
if kernel[int(kernel.size / 2)] == nodataVal:
return nodataVal
else:
# remove backgroud values from array
return np.average(kernel[np.where( kernel != nodataVal )])





























# function to generate edge contrast index in kernel that is not NODATA
def edgeContrastFunc(kernel, edgeDict, argsFilter):
# if center pixel is bigger than 'early' threshold, skip
# nodata pixels are still scanned in case they have large neighbor which would indicate an edge
# for this exercise, nodata is assumed to be open or 'early'
if (kernel[int(kernel.size / 2)] > argsFilter[2]):
return nodataVal
else:
# search early pixel for edges and record in early pixel dictionary
edgeContrastList = []
for adjPixel in (0, 1, 3 ,4):
edgeContrast = 1.0
edgeContrast = (argsFilter[4] * kernel[adjPixel]) + argsFilter[5]
else:
edgeContrast = -1.0
# edge found! add to list that will be inserted into the dictionary and calc stats
if (edgeContrast > -1.0):
edgeContrastList.append(edgeContrast)
argsFilter[0] += 1
if edgeContrast > argsFilter[1]:
argsFilter[1] = edgeContrast
if (len(edgeContrastList) == 0):
return nodataVal
else:
earlyPatchNum = len(edgeDict) + 1
edgeDict[earlyPatchNum] = edgeContrastList
return earlyPatchNum

http://docs.scipy.org/doc/scipy/reference/ge
nerated/scipy.ndimage.filters.generic_filter.ht
ml#scipy.ndimage.filters.generic_filter
```