org.openimaj.image.feature.dense.gradient.binning

Class FixedHOGStrategy

java.lang.Object

org.openimaj.image.feature.dense.gradient.binning.FixedHOGStrategy

All Implemented Interfaces:

SpatialBinningStrategy

@Reference(type=Inproceedings,
           author={"Dalal, Navneet","Triggs, Bill"},
           title="Histograms of Oriented Gradients for Human Detection",
           year="2005",
           booktitle="Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR\'05) - Volume 1 - Volume 01",
           pages={"886","","893"},
           url="http://dx.doi.org/10.1109/CVPR.2005.177",
           publisher="IEEE Computer Society",
           series="CVPR \'05",
           customData={"isbn","0-7695-2372-2","numpages","8","doi","10.1109/CVPR.2005.177","acmid","1069007","address","Washington, DC, USA"})
public class FixedHOGStrategy
extends Object
implements SpatialBinningStrategy

Implementation of an SpatialBinningStrategy that extracts normalised HOG features in the style defined by Dalal and Triggs. In this scheme, histograms are computed for fixed size cells (size in pixels), and are aggregated over fixed size blocks of cells. Blocks may overlap. The aggregated histogram for each block is then normalised before being appended into the final output histogram.

Author:

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

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	FixedHOGStrategy.BlockNormalisation Block normalisation schemes

Constructor Summary

Constructors

Constructor and Description
FixedHOGStrategy(int cellSize, int cellsPerBlock, FixedHOGStrategy.BlockNormalisation norm) Construct with square cells of the given size (in pixels).
FixedHOGStrategy(int cellSize, int cellsPerBlock, int blockStep, FixedHOGStrategy.BlockNormalisation norm) Construct with square cells of the given size (in pixels).
FixedHOGStrategy(int cellWidth, int cellHeight, int cellsPerBlockX, int cellsPerBlockY, int blockStepX, int blockStepY, FixedHOGStrategy.BlockNormalisation norm) Construct with square cells of the given size (in pixels).

Method Summary

Modifier and Type	Method and Description
Histogram	extract(WindowedHistogramExtractor binnedData, Rectangle region, Histogram output) Extract a histogram describing image content in the given region using the given WindowedHistogramExtractor to extract (sub) histograms from which to build the output.

Methods inherited from class java.lang.Object

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

Constructor Detail

FixedHOGStrategy

public FixedHOGStrategy(int cellSize,
                        int cellsPerBlock,
                        FixedHOGStrategy.BlockNormalisation norm)

Construct with square cells of the given size (in pixels). Square blocks are constructed from the given number of cells in each dimension. Blocks overlap, and shift by 1 cell (i.e. overlap is cellsPerBlock - 1).

Parameters:

cellSize - the size of the cells in pixels

cellsPerBlock - the number of cells per block

norm - the normalisation scheme

FixedHOGStrategy

public FixedHOGStrategy(int cellSize,
                        int cellsPerBlock,
                        int blockStep,
                        FixedHOGStrategy.BlockNormalisation norm)

Construct with square cells of the given size (in pixels). Square blocks are constructed from the given number of cells in each dimension.

Parameters:

cellSize - the size of the cells in pixels

cellsPerBlock - the number of cells per block

blockStep - the amount to shift each block in terms of cells

norm - the normalisation scheme

FixedHOGStrategy

public FixedHOGStrategy(int cellWidth,
                        int cellHeight,
                        int cellsPerBlockX,
                        int cellsPerBlockY,
                        int blockStepX,
                        int blockStepY,
                        FixedHOGStrategy.BlockNormalisation norm)

Construct with square cells of the given size (in pixels). Square blocks are constructed from the given number of cells in each dimension.

Parameters:

cellWidth - the width of the cells in pixels

cellHeight - the height of the cells in pixels

cellsPerBlockX - the number of cells per block in the x direction

cellsPerBlockY - the number of cells per block in the y direction

blockStepX - the amount to shift each block in terms of cells in the x direction

blockStepY - the amount to shift each block in terms of cells in the y direction

norm - the normalisation scheme

Method Detail

extract

public Histogram extract(WindowedHistogramExtractor binnedData,
                         Rectangle region,
                         Histogram output)

Description copied from interface: SpatialBinningStrategy

Extract a histogram describing image content in the given region using the given WindowedHistogramExtractor to extract (sub) histograms from which to build the output.

For efficiency, this method allows the output histogram to be specified as an input. This means that implementors of this interface can attempt to fill the output histogram rather than creating a new instance (although care should be taken to ensure that the porivded output histogram is the correct size and not null).

Users of SpatialBinningStrategys should use the following style for maximum efficiency:

 Histogram h = null;
 ...
 for (Rectangle region : lots_of_regions)
        h = strategy.extract(binnedData, region, h);
 

Specified by:

extract in interface SpatialBinningStrategy

Parameters:

binnedData - the WindowedHistogramExtractor to extract sub-histograms from

region - the region to extract from

output - the output histogram to fill (can be null)

Returns:

the extracted histogram (preferably output)