Discrete Gaussian kernels are often used for convolution in signal processing, or, in my case, weighting. I used some hardcoded values before, but

here's a recipe for making it on-the-fly.

```
def gauss_kern(size, sizey=None):
""" Returns a normalized 2D gauss kernel array for convolutions """
size = int(size)
if not sizey:
sizey = size
else:
sizey = int(sizey)
x, y = mgrid[-size:size+1, -sizey:sizey+1]
g = exp(-(x**2/float(size)+y**2/float(sizey)))
return g / g.sum()
```

Note that 'size' here refers to (shape[0]/2 - 1), i.e. the resulting kernel side is size*2 + 1 long. I need to use a kernel with a central value of 1, so I adjusted the normalisation accordingly:

```
def gauss_kern(size, sizey=None):
""" Returns a normalized 2D gauss kernel array for convolutions """
size = int(size)
if not sizey:
sizey = size
else:
sizey = int(sizey)
x, y = scipy.mgrid[-size:size+1, -sizey:sizey+1]
g = scipy.exp(-(x**2/float(size)+y**2/float(sizey)))
return g / g.max()
```

## No comments:

## Post a Comment