org.gridfour.compress

Class CodecStats

java.lang.Object

org.gridfour.compress.CodecStats

public class CodecStats
extends Object

A simple container for collecting statistics when analyzing compressed data.

This statistics collector is suited to the standard compressors implemented by the Gridfour API. If desired, it may be used for other compressor implementations.

Constructor Summary

Constructors

Constructor and Description
CodecStats(PredictorModelType pcType) Construct a statistics object for the specified predictor model.

Method Summary

Modifier and Type	Method and Description
void	addCountsForM32(int nM32, byte[] m32) Add counts for the M32 symbols derived from the predictor for the tile.
void	addToCounts(int nBytesForTile, int nSymbolsInTile, int nBitsOverhead) Add metadata about the tile to the counts
void	clear() Clear any accumulated counts
double	getAverageLength() Get the average length of encoded tiles, in bytes.
double	getAverageMCodeLength() Gets the average length of the M32 code length across tiles.
double	getAverageObservedMCodes() Gets the average number of unique M32 codes observed in the compressed tile.
double	getAverageOverhead() Get the average overhead per tile.
double	getBitsPerSymbol() Get the number of bits per symbol in the pre-compression encoding.
double	getEntropy() Get the entropy for the data.
String	getLabel() Get a label for the predictor.
long	getTileCount() Get the number of tiles that have been counted.

Methods inherited from class java.lang.Object

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

Constructor Detail

CodecStats

public CodecStats(PredictorModelType pcType)

Construct a statistics object for the specified predictor model.

Parameters:

pcType - a valid predictor model type.

Method Detail

getLabel

public String getLabel()

Get a label for the predictor.

Returns:

a valid string

addToCounts

public void addToCounts(int nBytesForTile,
                        int nSymbolsInTile,
                        int nBitsOverhead)

Add metadata about the tile to the counts

Parameters:

nBytesForTile - the number of byts for the compressed version of the tile.

nSymbolsInTile - the number of values in the tile, typically the number of cells in the grid.

nBitsOverhead - Any compressor-specific overhead that an implementation wishes to track.

addCountsForM32

public void addCountsForM32(int nM32,
                            byte[] m32)

Add counts for the M32 symbols derived from the predictor for the tile.

Parameters:

nM32 - the total number of symbols in the encoded data

m32 - the encoded data

getEntropy

public double getEntropy()

Get the entropy for the data. In information theory, the term "entropy" is essentially an indicator of the number of bits actually needed to encode data.

Returns:

a positive value

clear

public void clear()

Clear any accumulated counts

getBitsPerSymbol

public double getBitsPerSymbol()

Get the number of bits per symbol in the pre-compression encoding.

Returns:

zero or a positive number

getTileCount

public long getTileCount()

Get the number of tiles that have been counted.

Returns:

zero or a positive number

getAverageMCodeLength

public double getAverageMCodeLength()

Gets the average length of the M32 code length across tiles.

Returns:

a positive floating point value, potentially zero.

getAverageObservedMCodes

public double getAverageObservedMCodes()

Gets the average number of unique M32 codes observed in the compressed tile. The maximum number of unique M32 codes is 256 (which would occur if all possible M32 codes were used).

Returns:

a positive floating point value, potentially zero.

getAverageOverhead

public double getAverageOverhead()

Get the average overhead per tile. Not all compressors tabulate this value.

Returns:

zero or a positive value.

getAverageLength

public double getAverageLength()

Get the average length of encoded tiles, in bytes.

Returns:

zero or a positive number.