Package

toolkit.filtering

contourlets

Permalink

package contourlets

Visibility

Public
All

Type Members

class Filter extends AnyRef

A filter is a matrix with an origin at some place other than (0, 0).
class ReadMe extends AnyRef

This package implements the nonsubsampled contourlet transform.
This package implements the nonsubsampled contourlet transform.
It is based on the Matlab nonsubsampled contourlet toolbox available here:
http://www.mathworks.com/matlabcentral/fileexchange/10049-nonsubsampled-contourlet-toolbox
Here's a map of Matlab files to corresponding Scala files in this directory:
```
atrousfilters.m   AtrousFilters
dfilters.m        DirectionalFilters
dmaxflat.m        DiamondMaxFlatFilters
mctrans.m         McClellanTransform
modulate2.m       Modulate
nsctdec.m         ContourletTransform
nsdfbdec.m        DirectionalFilterbank
parafilters.m     ParallelogramFilters
resamplz.m        Shear
```
class Shear extends libcog.Operator

An instance of the Shear operator.
An instance of the Shear operator.
Used only by the companion object.

Value Members

object AtrousFilter extends MatlabFunctions

Implements atrousfilters.m from the nonsubsampled contourlet toolbox.
object ContourletTransform extends libcog.Logarithm

Implements a simple Contourlet transform and its inverse.
Implements a simple Contourlet transform and its inverse.
This supports only a single level of decomposition, since that's all that's needed for my application. You may extend this if needed, but it will be tricky since it involves an "a trous" convolution (if you have to ask...).
Implements nsctdec.m and nsctrec.m
object DiamondFilter extends MatlabFunctions

Directional filters.
Directional filters.
Implements dfilters.m (although only one of the filters).
object DiamondMaxFlatFilter extends MatlabFunctions

Diamond max flat filters of various orders.
object DirectionalFilterbank extends libcog.Logarithm with MatlabFunctions

Directional filterbank and its inverse.
Directional filterbank and its inverse.
Implements the filter construction part of nsdfbdec.m and nsdfbrec.m, but does not do the convolution. Although the matlab code supports an arbitrary number of channels (that are a power of 2), I cut this off at 8 channels. This makes the code much more readable. If you need more channels, you may add the code for that.
object McClellanTransform extends MatlabFunctions

Constructs a 2D FIR filter from a 1D filter.
Constructs a 2D FIR filter from a 1D filter.
Implements mctrans.m
object Modulate extends MatlabFunctions

Creates a 2D modulated filter from a 1D filter.
Creates a 2D modulated filter from a 1D filter.
Implements modulate2d.m
object ParallelogramFilters

Generates a set of 4 parallelogram filters from a diamond filter.
Generates a set of 4 parallelogram filters from a diamond filter.
Implements parafilters.m
object Resample

Resamples a filter.
Resamples a filter. See "Multidimensional filter banks and multiscale geometric representations," Do and Lu, 2011.
object SamplingMatrices

The fundamental sampling matrices.
The fundamental sampling matrices. See "Multidimension filter banks and multiscale geometric representations," Do and Lu, Foundations and Trends in Signal Processing, 2011.
Note that the names here correspond to those in the Matlab code, not the paper. Here's the mapping:
```
paper       code (matlab and here)

 Q0           Q1
 Q1           Q2
 R0           R1
 R1           R2
 R2           R3
 R3           R4

 D1           D1
 D2           D2
```
the same as Q1 in the Matlab code).
object Shear

Operator which resamples an image by shearing it in various ways.
Operator which resamples an image by shearing it in various ways. This is not periodic, rather it extends the input where necessary with zeroes. This means the output is always twice as big as the input.
The implements the functionality of resampz.m

Ungrouped