Main Content

featureMatrix

Image scattering feature matrix

Description

smat = featureMatrix(sf,im) returns the scattering feature matrix for the wavelet image scattering network, sf, and the input image, im. im is a real-valued 2-D (M-by-N) or 3-D matrix (M-by-N-by-3). If im is a 3-D matrix, the size of the third dimension must be 3. If im is a 2-D matrix, smat is Np-by-Ms-by-Ns, where Np is the number of scattering paths and Ms-by-Ns is the resolution of the scattering coefficients. If im is a 3-D matrix, smat is Np-by-Ms-by-Ns-by-3.

example

smat = featureMatrix(sf,sc) returns the scattering feature matrix for the cell array of scattering coefficients, sc. sc is obtained from the scatteringTransform method of the wavelet image scattering network.

smat = featureMatrix(___,'Transform',transformtype) applies the transformation specified by transformtype to the scattering coefficients. Valid options for transformtype are 'log' and 'none'. If unspecified, transformtype defaults to 'none'. You can use this syntax with any of the previous syntaxes.

Examples

collapse all

This example shows how to obtain the feature matrix for a wavelet image scattering network.

Load the xbox image. Create an image scattering network suitable for the image.

load xbox
sf = waveletScattering2('ImageSize',size(xbox))
sf = 
  waveletScattering2 with properties:

             ImageSize: [128 128]
       InvarianceScale: 64
          NumRotations: [6 6]
        QualityFactors: [1 1]
             Precision: 'single'
    OversamplingFactor: 0
          OptimizePath: 1

Obtain the feature matrix.

smat = featureMatrix(sf,xbox);

Input Arguments

collapse all

Wavelet image scattering network, specified as a waveletScattering2 object.

Input image, specified as real-valued 2-D matrix or 3-D matrix. If im is 3-D, im is assumed to be a color image in the RGB color space, and the size of the third dimension must equal 3. The row and column sizes of im must match the ImageSize property of sf.

Scattering coefficients, specified as a cell array. sc is obtained from the scatteringTransform method of the image scattering network.

Transformation to apply to the scattering coefficients:

  • 'none': No transformation is applied to the scattering coefficients.

  • 'log': The natural logarithm is applied to the scattering coefficients.

Output Arguments

collapse all

Scattering feature matrix for the 2-D scattering network sf, returned as a real-valued array. If im is a 2-D matrix, smat is Np-by-Ms-by-Ns, where Np is the number of scattering paths and Ms-by-Ns is the resolution of the scattering coefficients. If im is a 3-D matrix, smat is Np-Ms-by-Ns-by-3.

Version History

Introduced in R2019a