mango.image.convolve

mango.image.convolve(input, weights=array([[[ 1.]]]), mode='mirror', cval=None, stride=(1, 1, 1), boffset=(0, 0, 0), eoffset=(0, 0, 0), normWeights=False)

3D convolution filter.

Parameters:

• input (mango.Dds) – Image to be filtered.
• weights (numpy.array.) – 3D convolution kernel/weights.
• mode (str) – String indicating how to handle borders, one of “mirror”, “constant”, “reflect”.
• cval (scalar) – When mode=='constant' use this value to initialise borders.
• stride (3-sequence) – The sub-sampling step-size for each dimension.
• boffset (3-sequence) – The offset ((bz,by,bx) relative to input.origin) at which the filtering starts.
• eoffset (3-sequence) – The offset ((ez,ey,ex) relative to (input.origin + input.shape)) at which the filtering stops.
• normWeights (bool) – If True, weights are normalised before performing the convolution. Also, any partially masked-neighbourhoods are normalised by the sum of the non-masked weights.

Return type:

mango.Dds

Returns:

Convolved image.