MBFStatisticalPixelProfileModel (OpenIMAJ master project 1.3.10 API)

java.lang.Object
- org.openimaj.image.pixel.statistics.MBFStatisticalPixelProfileModel

All Implemented Interfaces:

PixelProfileModel<MBFImage>
```
public class MBFStatisticalPixelProfileModel
extends Object
implements PixelProfileModel<MBFImage>
```
An MBFStatisticalPixelProfileModel is a statistical model of pixels from an MBFImage sampled along a line. The model allows for various sampling strategies (see FLineSampler) and uses the mean and covariance as its internal state. The model is updateable, but does not hold on to previously seen samples to reduce memory usage. Internally, the model is one-dimensional, and is created by stacking the samples from each image band.

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Constructor Summary

Constructors
Constructor and Description
`MBFStatisticalPixelProfileModel(int nsamples, FLineSampler sampler)` Construct a new `MBFStatisticalPixelProfileModel` with the given number of samples per line, and the given sampling strategy.

Method Summary

All Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`float`	`computeCost(MBFImage image, Line2d line)` Compute the cost of a vector of samples extracted along a line in the given image to the internal model.
`float`	`computeMahalanobis(float[] vector)` Compute the Mahalanobis distance of the given vector to the internal model.
`float`	`computeMahalanobis(MBFImage image, Line2d line)` Compute the Mahalanobis distance of a vector of samples extracted along a line in the given image to the internal model.
`float[]`	`computeMahalanobisWindowed(float[][] vector)` Compare this model at each overlapping position of the given vector starting from the first sample and return the distance for each overlap.
`float[]`	`computeMahalanobisWindowed(MBFImage image, Line2d line, int numSamples)` Extract numSamples samples from the line in the image and then compare this model at each overlapping position starting from the first sample at the beginning of the line.
`float`	`computeMovementDistance(MBFImage image, Line2d line, int numSamples, Point2d pt)` Compute the distance between the centre of the given line and the given point, normalised as a function of the length of the sampling line.
`Point2d`	`computeNewBest(MBFImage image, Line2d line, int numSamples)` Extract numSamples samples from the line in the image and then compare this model at each overlapping position starting from the first sample at the beginning of the line.
`Jama.Matrix`	`getCovariance()`
`Jama.Matrix`	`getInverseCovariance()`
`double[]`	`getMean()`
`int`	`getNumberSamples()`
`FLineSampler`	`getSampler()`
`String`	`toString()`
`void`	`updateModel(MBFImage image, Line2d line)` Update the model with a new sample.

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

- Constructor Detail
  - MBFStatisticalPixelProfileModel
```
public MBFStatisticalPixelProfileModel(int nsamples,
                                       FLineSampler sampler)
```
    Construct a new MBFStatisticalPixelProfileModel with the given number of samples per line, and the given sampling strategy.
    
    Parameters:
    
    nsamples - number of samples
    
    sampler - line sampling strategy
- Method Detail
  - updateModel
```
public void updateModel(MBFImage image,
                        Line2d line)
```
    Description copied from interface: PixelProfileModel
    
    Update the model with a new sample.
    
    Specified by:
    
    updateModel in interface PixelProfileModel<MBFImage>
    
    Parameters:
    
    image - the image to extract the sample from
    
    line - the line across with to sample
  - getMean
```
public double[] getMean()
```
    Returns:
    
    the mean of the model
  - getCovariance
```
public Jama.Matrix getCovariance()
```
    Returns:
    
    the covariance of the model
  - getInverseCovariance
```
public Jama.Matrix getInverseCovariance()
```
    Returns:
    
    the inverse of the covariance matrix
  - computeMahalanobis
```
public float computeMahalanobis(float[] vector)
```
    Compute the Mahalanobis distance of the given vector to the internal model. The vector must have the same size as the number of samples given during construction.
    
    Parameters:
    
    vector - the vector
    
    Returns:
    
    the computed Mahalanobis distance
  - computeMahalanobis
```
public float computeMahalanobis(MBFImage image,
                                Line2d line)
```
    Compute the Mahalanobis distance of a vector of samples extracted along a line in the given image to the internal model.
    
    Parameters:
    
    image - the image to sample
    
    line - the line to sample along
    
    Returns:
    
    the computed Mahalanobis distance
  - computeMahalanobisWindowed
```
public float[] computeMahalanobisWindowed(MBFImage image,
                                          Line2d line,
                                          int numSamples)
```
    Extract numSamples samples from the line in the image and then compare this model at each overlapping position starting from the first sample at the beginning of the line. numSamples must be bigger than the number of samples used to construct the model. In addition, callers are responsible for ensuring the sampling rate between the new samples and the model is equal.
    
    Parameters:
    
    image - the image to sample
    
    line - the line to sample along
    
    numSamples - the number of samples to make
    
    Returns:
    
    an array of the computed Mahalanobis distances at each offset
  - computeNewBest
```
public Point2d computeNewBest(MBFImage image,
                              Line2d line,
                              int numSamples)
```
    Description copied from interface: PixelProfileModel
    
    Extract numSamples samples from the line in the image and then compare this model at each overlapping position starting from the first sample at the beginning of the line. numSamples must be bigger than the number of samples used to construct the model. In addition, callers are responsible for ensuring the sampling rate between the new samples and the model is equal. The point on the line corresponding to the smallest Mahalanobis distance is returned.
    
    Specified by:
    
    computeNewBest in interface PixelProfileModel<MBFImage>
    
    Parameters:
    
    image - the image to sample
    
    line - the line to sample along
    
    numSamples - the number of samples to make
    
    Returns:
    
    the "best" position on the line
  - computeMovementDistance
```
public float computeMovementDistance(MBFImage image,
                                     Line2d line,
                                     int numSamples,
                                     Point2d pt)
```
    Description copied from interface: PixelProfileModel
    
    Compute the distance between the centre of the given line and the given point, normalised as a function of the length of the sampling line.
    
    Specified by:
    
    computeMovementDistance in interface PixelProfileModel<MBFImage>
    
    Parameters:
    
    image - the image to sample
    
    line - the line to sample along
    
    numSamples - the number of samples to make
    
    pt - the point to compare
    
    Returns:
    
    the normalised distance (0 means same point; 1 means on end of line)
  - computeMahalanobisWindowed
```
public float[] computeMahalanobisWindowed(float[][] vector)
```
    Compare this model at each overlapping position of the given vector starting from the first sample and return the distance for each overlap. The length of the vector must be bigger than the number of samples used to construct the model. In addition, callers are responsible for ensuring the sampling rate between the new samples and the model is equal.
    
    Parameters:
    
    vector - array of samples, one vector per band
    
    Returns:
    
    an array of the computed Mahalanobis distances at each offset
  - toString
```
public String toString()
```
    Overrides:
    
    toString in class Object
  - getNumberSamples
```
public int getNumberSamples()
```
    Returns:
    
    the number of samples used along each profile line
  - getSampler
```
public FLineSampler getSampler()
```
    Returns:
    
    the sampler used by the model to extract samples along profiles
  - computeCost
```
public float computeCost(MBFImage image,
                         Line2d line)
```
    Description copied from interface: PixelProfileModel
    
    Compute the cost of a vector of samples extracted along a line in the given image to the internal model.
    
    Specified by:
    
    computeCost in interface PixelProfileModel<MBFImage>
    
    Parameters:
    
    image - the image to sample
    
    line - the line to sample along
    
    Returns:
    
    the computed cost

Class MBFStatisticalPixelProfileModel

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Constructor Detail

MBFStatisticalPixelProfileModel

Method Detail

updateModel

getMean

getCovariance

getInverseCovariance

computeMahalanobis

computeMahalanobis

computeMahalanobisWindowed

computeNewBest

computeMovementDistance

computeMahalanobisWindowed

toString

getNumberSamples

getSampler

computeCost