HISTEQ - Performs an histogram equalisation on an NDF

Description:

This application transforms an NDF  via histogram equalisation. Histogram equalisation is an image-processing technique in which the distribution (between limits) of data values in the input array is adjusted so that in the output array there are approximately equal numbers of elements in each histogram bin. To achieve this the histogram bin size is no longer a constant. This technique is commonly known as histogram equalisation. It is useful for displaying features across a wide dynamic range, sometimes called a maximum-information picture. The transformed array is output to a new NDF.

Usage:

histeq in out [numbin]

Parameters:

IN = NDF (Read)

The NDF structure to be transformed.

NUMBIN = _INTEGER (Read)

The number of histogram bins to be used. This should be a large number, say 2000, to reduce quantisation errors. It must be in the range 100 to 10000. [2048]

OUT = NDF (Write)

The NDF structure to contain the transformed data array.

TITLE = LITERAL (Read)

Title for the output NDF structure. A null value (!) propagates the title from the input NDF to the output NDF. [!]

Examples:

histeq halley maxinf

The data array in the NDF called halley is remapped via histogram equalisation to form the new NDF called maxinf.

histeq halley maxinf 10000 title="Maximum information of Halley"

The data array in the NDF called halley is remapped via histogram equalisation to form the new NDF called maxinf. Ten thousand bins in the histogram are required rather than the default of 2048. The title of NDF maxinf is "Maximum information of Halley".

Notes:

If there are a few outliers in the data and most of the points concentrated about a value it may be wise to truncate the data array via THRESH, or have a large number of histogram bins.

Related Applications :

KAPPA: LAPLACE, LUTABLE, LUTEDIT, SHADOW, THRESH; FIGARO: HOPT.

Implementation Status:

• This routine correctly processes the AXIS, DATA, QUALITY, LABEL, TITLE, WCS, and HISTORY  components of an NDF data structure and propagates all extensions. UNITS  and VARIANCE  become undefined by the transformation, and so are not propagated.

• Processing of bad pixels and automatic quality masking are supported.

• All non-complex numeric data types can be handled.

• Any number of NDF dimensions is supported.

Click here to see an image displayed before (left) and after histogram equalisation (67k). The same scaling limits are used for both images.