net.grelf.image
Class Kernel

java.lang.Object
  net.grelf.image.Kernel

public class Kernel
extends java.lang.Object

A square array of values for convolving with an image. Can be set from an image or as a Gaussian hump or loaded from a text (CSV) file.

Constructor Summary

Kernel(int halfWidth, float[][] data)

Kernel(java.awt.Point pt, Image image, int halfWidth)
Create a kernel by sampling a square area symmetrically around the given point in the given accumulator.

Method Summary

boolean	askHalfWidth() Ask the user for the kernel's halfwidth.
static Kernel	createGaussianKernel(Image image, int halfWidth) Create a kernel which is a Gaussian hump.
float[][]	getData() First index = band, second index = samples for given colour band.
int	getHalfWidth() Kernel will comprise (2.halfWidth + 1) x (2.halfWidth + 1) pixels.
static Kernel	load(java.lang.String path, int nBands) Load kernel from a comma-separated text file.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Kernel

public Kernel(int halfWidth,
              float[][] data)

Kernel

public Kernel(java.awt.Point pt,
              Image image,
              int halfWidth)

Create a kernel by sampling a square area symmetrically around the given point in the given accumulator. The width and height of the square kernel will both be 2 * halfWidth + 1. However, if the halfWidth value is negative, the halfWidth will be redetermined from the image. That is done by scanning spirally around the given point first to find a peak sum of levels in all channels and then spirally outwards until the sum of levels in all channels is always below 10% of the peak value found, or the image edge is reached. This latter method is appropriate for picking up a star image as the kernel, for deconvolving astrophotographs. That is because ideally the star image should be a point.

Method Detail

getHalfWidth

public int getHalfWidth()

Kernel will comprise (2.halfWidth + 1) x (2.halfWidth + 1) pixels.

getData

public float[][] getData()

First index = band, second index = samples for given colour band.

load

public static Kernel load(java.lang.String path,
                          int nBands)

Load kernel from a comma-separated text file. Single band. Expects same number of rows as columns.

askHalfWidth

public boolean askHalfWidth()

Ask the user for the kernel's halfwidth.

createGaussianKernel

public static Kernel createGaussianKernel(Image image,
                                          int halfWidth)

Create a kernel which is a Gaussian hump.