In practice, materials science information must be evaluated and compared in a statistical manner. Traditional materials science practice involves pre-processing steps that identify material features; from there statistics of the features are extracted. Examples of feature statistics are orientition distribution functions, volume fraction, variance; the statistics are feature identifiers in materials science.
Spatial statistics provide a powerful objective statistical quantifier for materials science information. Spatial statistics have effective statistical measures embedded in them such as volume fraction and specific surface area.
-
Clone this Repository
-
Compute Spatial Statistics from a URL -
url = 'https://farm3.staticflickr.com/2397/12972389405_223298503d_z.jpg'; [F,xx] = SpatialStatsFFT(url);
You will recieve an alert that the raw image has been stored in your Matlab workspace.
SpatialStatsFFT- Compute the Spatial Statistics using Fast Fourier Transform algorithms for speak.PairCorrelationFFT- Compute the Pair Correlation by computing the vector-resolved Spatial Statistics and integrating over angle.FindPeaksSSFFT- Find the peaks (or valleys) in the vector-resolved Spatial StatisticsPlotSlice- A requested visualization tool to plot individual slices in volumetric spatial statistics.
Automated code documentation can be found here.
Spatial statistics have shown diverse applications in
- Metallic feature identification
- Homogenization relations in fuel cells
- Electronic potential identification in Molecular Dynamics
I suggest using Mark Tygert's randomized Principal Component Analysis function that can be downloaded from http://cims.nyu.edu/~tygert/pca.m .