Difference between revisions of "Example Image Pyramid"

From BoofCV
Jump to navigationJump to search
m (Created page with "= Image Pyramid Example = <center> <gallery widths=300px heights=200px> Image:Example_pyramid.jpg | Discrete Pyramid Visualization </gallery> </center> [http://en.wikipedia.org/...")
 
m
 
(17 intermediate revisions by the same user not shown)
Line 1: Line 1:
= Image Pyramid Example =
<center>
<center>
<gallery widths=300px heights=200px>
<gallery widths=300px heights=200px>
Image:Example_pyramid.jpg | Discrete Pyramid Visualization
Image:Example_pyramid.jpg | Discrete Pyramid Visualization
Image:Example_image_pyramid.png | Cartoon showing a pyramid with 1,2,4 scaling.
</gallery>
</gallery>
</center>
</center>


[http://en.wikipedia.org/wiki/Pyramid_%28image_processing Image pyramids] are a common way to represent multi-resolution image information. In an image pyramid, upper layers are lower resolution versions of the lower layers. BoofCV provides two types of image pyramids built in; PyramidDiscrete and PyramidFloat. PyramidDiscrete only allows the ratio between adjacent to have positive integer values, while PyramidFloat allows any arbitrary positive value. Discrete pyramids are much faster than float pyramids, but much more restrictive.
[http://en.wikipedia.org/wiki/Pyramid_%28image_processing Image pyramids] are a common way to represent multi-resolution image information. In an image pyramid, upper layers are lower resolution versions of the lower layers. BoofCV provides two types of image pyramids built in; PyramidDiscrete and PyramidFloat. PyramidDiscrete only allows the ratio between adjacent to have positive integer values, while PyramidFloat allows any arbitrary positive value. Discrete pyramids are much faster than float pyramids, but much more restrictive.


Two code examples are provided below which demonstrate how to work with each type of pyramid.
Inside of BoofCV several algorithms make use of these two types of pyramids. The KLT feature tracker uses a discrete pyramid and is the fastest feature tracker. Pyramid based scale space feature detectors use the float pyramid. 
 
Two code examples are provided below which demonstrate how to construct and visualize each type of pyramid.


Example Code:
Example Code:
* [https://github.com/lessthanoptimal/BoofCV/blob/master/examples/src/boofcv/examples/ExamplePyramidDiscrete.java ExamplePyramidDiscrete.java ]
* [https://github.com/lessthanoptimal/BoofCV/blob/v0.40/examples/src/main/java/boofcv/examples/imageprocessing/ExamplePyramidDiscrete.java ExamplePyramidDiscrete.java ]
* WRITE FLOAT EXAMPLE
* [https://github.com/lessthanoptimal/BoofCV/blob/v0.40/examples/src/main/java/boofcv/examples/imageprocessing/ExamplePyramidFloat.java ExamplePyramidFloat.java ]


Concepts:
Concepts:
Line 18: Line 20:
* Discrete Vs. Float pyramids
* Discrete Vs. Float pyramids
* Image scaling
* Image scaling
Relevant Applets:
* [[Applet_Pyramid_Discrete| Discrete Pyramid]]
* [[Applet_Pyramid_Float| Float Pyramid]]
* [[Applet_Scale_Space_Point| Scale-Space Feature Detection]]


= Image Pyramid Discrete =
= Image Pyramid Discrete =
<syntaxhighlight lang="java">
<syntaxhighlight lang="java">
/**
/**
  * Demonstrates how to construct and display a {@link PyramidDiscrete}. Discrete pyramids require that
  * Demonstrates how to construct and display a {@link PyramidDiscrete}. Discrete pyramids require that
  * each level has a relative scale with an integer ratio and is updated by sparsely sub-sampling. These
  * each level has a relative scale with an integer ratio and is updated by sparsely sub-sampling. These
  * restrictions allows a very quick update across scale space.
  * restrictions allows a very quick update across scale space.
  *
  *
  * @author Peter Abeles
  * @author Peter Abeles
  */
  */
public class ExamplePyramidDiscrete<T extends ImageBase> {
public class ExamplePyramidDiscrete<T extends ImageGray<T>> {


// specifies the image type
// specifies the image type
Class<T> imageType;
Class<T> imageType;
// The pyramid data structure
PyramidDiscrete<T> pyramid;
// update the pyramid given from
PyramidUpdaterDiscrete<T> updater;


public ExamplePyramidDiscrete(Class<T> imageType) {
public ExamplePyramidDiscrete( Class<T> imageType ) {
this.imageType = imageType;
this.imageType = imageType;
}
}


/**
/**
* Creates a fairly standard pyramid and updater.
* Updates and displays the pyramid.
*/
*/
public void standard() {
public void process( BufferedImage image ) {
// Each level in the pyramid must have a ratio with the previously layer that is an integer value
T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
pyramid = new PyramidDiscrete<T>(imageType,true,1,2,4,8);
PyramidDiscrete<T> pyramid = FactoryPyramid.discreteGaussian(
ConfigDiscreteLevels.levels(4), -1, 2, true, ImageType.single(imageType));
pyramid.process(input);
 
var gui = new DiscretePyramidPanel<T>();
gui.setPyramid(pyramid);
gui.render();
 
ShowImages.showWindow(gui, "Image Pyramid");
 
// To get an image at any of the scales simply call this get function
T imageAtScale = pyramid.getLayer(1);
 
ShowImages.showWindow(ConvertBufferedImage.convertTo(imageAtScale, null, true), "Image at layer 1");
}
 
public static void main( String[] args ) {
BufferedImage image = UtilImageIO.loadImageNotNull(UtilIO.pathExample("standard/barbara.jpg"));
 
var app = new ExamplePyramidDiscrete<>(GrayF32.class);
// var app = new ExamplePyramidDiscrete<>(GrayU8.class);
 
app.process(image);
}
}
</syntaxhighlight>
 
 
= Image Pyramid Float =
 
<syntaxhighlight lang="java">
/**
* Demonstrates how to construct and display a {@link PyramidFloat}. Float pyramids require only require
* that each layer's scale be larger than the scale of the previous layer. Interpolation is used to allow
* sub-sampling at arbitrary scales. All of this additional flexibility comes at the cost of speed
* when compared to a {@link PyramidDiscrete}.
*
* @author Peter Abeles
*/
public class ExamplePyramidFloat<T extends ImageGray<T>> {
 
// specifies the image type
Class<T> imageType;
// The pyramid data structure
PyramidFloat<T> pyramid;


// In most cases sub-sampling with a Gaussian is preferred
public ExamplePyramidFloat( Class<T> imageType ) {
updater = FactoryPyramid.discreteGaussian(imageType,-1,2);
this.imageType = imageType;
}
}


/**
/**
* Creates a more unusual pyramid and updater.
* Creates a fairly standard pyramid and updater.
*/
*/
public void unusual() {
public void standard() {
// Note that the first level does not have to be one
// Scale factory for each layer can be any floating point value which is larger than
pyramid = new PyramidDiscrete<T>(imageType,true,2,6);
// the previous layer's scale.
 
double[] scales = new double[]{1, 1.5, 2, 2.5, 3, 5, 8, 15};
// Other kernels can also be used besides Gaussian
// the amount of blur which is applied to each layer in the pyramid after the previous layer has been sampled
Kernel1D kernel;
double[] sigmas = new double[]{1, 1, 1, 1, 1, 1, 1, 1};
if(GeneralizedImageOps.isFloatingPoint(imageType) ) {
pyramid = FactoryPyramid.floatGaussian(scales, sigmas, imageType);
kernel = FactoryKernel.table1D_F32(2,true);
} else {
kernel = FactoryKernel.table1D_I32(2);
}
 
updater = new PyramidUpdateIntegerDown<T>(kernel,imageType);
}
}


Line 80: Line 110:
*/
*/
public void process( BufferedImage image ) {
public void process( BufferedImage image ) {
T input = ConvertBufferedImage.convertFrom(image,null,imageType);
T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
updater.update(input,pyramid);
pyramid.process(input);


DiscretePyramidPanel gui = new DiscretePyramidPanel();
ImagePyramidPanel<T> gui = new ImagePyramidPanel<>();
gui.setPyramid(pyramid);
gui.set(pyramid, true);
gui.render();
gui.render();


ShowImages.showWindow(gui,"Image Pyramid");
ShowImages.showWindow(gui, "Image Pyramid Float");
 
// To get an image at any of the scales simply call this get function
T imageAtScale = pyramid.getLayer(1);
 
ShowImages.showWindow(ConvertBufferedImage.convertTo(imageAtScale, null, true), "Image at layer 1");
}
}


public static void main( String[] args ) {
public static void main( String[] args ) {
BufferedImage image = UtilImageIO.loadImage("data/standard/barbara.png");
BufferedImage image = UtilImageIO.loadImageNotNull(UtilIO.pathExample("standard/barbara.jpg"));


ExamplePyramidDiscrete<ImageFloat32> app = new ExamplePyramidDiscrete<ImageFloat32>(ImageFloat32.class);
var app = new ExamplePyramidFloat<>(GrayF32.class);
// ExamplePyramidDiscrete<ImageUInt8> app = new ExamplePyramidDiscrete<ImageUInt8>(ImageUInt8.class);
// var app = new ExamplePyramidFloat<>(GrayU8.class);


app.standard();
app.standard();
// app.unusual();
app.process(image);
app.process(image);
}
}
}
}
</syntaxhighlight>
</syntaxhighlight>
= Image Pyramid Float =
NEED TO WRITE THIS

Latest revision as of 15:11, 17 January 2022

Image pyramids are a common way to represent multi-resolution image information. In an image pyramid, upper layers are lower resolution versions of the lower layers. BoofCV provides two types of image pyramids built in; PyramidDiscrete and PyramidFloat. PyramidDiscrete only allows the ratio between adjacent to have positive integer values, while PyramidFloat allows any arbitrary positive value. Discrete pyramids are much faster than float pyramids, but much more restrictive.

Inside of BoofCV several algorithms make use of these two types of pyramids. The KLT feature tracker uses a discrete pyramid and is the fastest feature tracker. Pyramid based scale space feature detectors use the float pyramid.

Two code examples are provided below which demonstrate how to construct and visualize each type of pyramid.

Example Code:

Concepts:

  • Multi-resolution image processing
  • Discrete Vs. Float pyramids
  • Image scaling

Image Pyramid Discrete

/**
 * Demonstrates how to construct and display a {@link PyramidDiscrete}. Discrete pyramids require that
 * each level has a relative scale with an integer ratio and is updated by sparsely sub-sampling. These
 * restrictions allows a very quick update across scale space.
 *
 * @author Peter Abeles
 */
public class ExamplePyramidDiscrete<T extends ImageGray<T>> {

	// specifies the image type
	Class<T> imageType;

	public ExamplePyramidDiscrete( Class<T> imageType ) {
		this.imageType = imageType;
	}

	/**
	 * Updates and displays the pyramid.
	 */
	public void process( BufferedImage image ) {
		T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
		PyramidDiscrete<T> pyramid = FactoryPyramid.discreteGaussian(
				ConfigDiscreteLevels.levels(4), -1, 2, true, ImageType.single(imageType));
		pyramid.process(input);

		var gui = new DiscretePyramidPanel<T>();
		gui.setPyramid(pyramid);
		gui.render();

		ShowImages.showWindow(gui, "Image Pyramid");

		// To get an image at any of the scales simply call this get function
		T imageAtScale = pyramid.getLayer(1);

		ShowImages.showWindow(ConvertBufferedImage.convertTo(imageAtScale, null, true), "Image at layer 1");
	}

	public static void main( String[] args ) {
		BufferedImage image = UtilImageIO.loadImageNotNull(UtilIO.pathExample("standard/barbara.jpg"));

		var app = new ExamplePyramidDiscrete<>(GrayF32.class);
//		var app = new ExamplePyramidDiscrete<>(GrayU8.class);

		app.process(image);
	}
}


Image Pyramid Float

/**
 * Demonstrates how to construct and display a {@link PyramidFloat}. Float pyramids require only require
 * that each layer's scale be larger than the scale of the previous layer. Interpolation is used to allow
 * sub-sampling at arbitrary scales. All of this additional flexibility comes at the cost of speed
 * when compared to a {@link PyramidDiscrete}.
 *
 * @author Peter Abeles
 */
public class ExamplePyramidFloat<T extends ImageGray<T>> {

	// specifies the image type
	Class<T> imageType;
	// The pyramid data structure
	PyramidFloat<T> pyramid;

	public ExamplePyramidFloat( Class<T> imageType ) {
		this.imageType = imageType;
	}

	/**
	 * Creates a fairly standard pyramid and updater.
	 */
	public void standard() {
		// Scale factory for each layer can be any floating point value which is larger than
		// the previous layer's scale.
		double[] scales = new double[]{1, 1.5, 2, 2.5, 3, 5, 8, 15};
		// the amount of blur which is applied to each layer in the pyramid after the previous layer has been sampled
		double[] sigmas = new double[]{1, 1, 1, 1, 1, 1, 1, 1};
		pyramid = FactoryPyramid.floatGaussian(scales, sigmas, imageType);
	}

	/**
	 * Updates and displays the pyramid.
	 */
	public void process( BufferedImage image ) {
		T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
		pyramid.process(input);

		ImagePyramidPanel<T> gui = new ImagePyramidPanel<>();
		gui.set(pyramid, true);
		gui.render();

		ShowImages.showWindow(gui, "Image Pyramid Float");

		// To get an image at any of the scales simply call this get function
		T imageAtScale = pyramid.getLayer(1);

		ShowImages.showWindow(ConvertBufferedImage.convertTo(imageAtScale, null, true), "Image at layer 1");
	}

	public static void main( String[] args ) {
		BufferedImage image = UtilImageIO.loadImageNotNull(UtilIO.pathExample("standard/barbara.jpg"));

		var app = new ExamplePyramidFloat<>(GrayF32.class);
//		var app = new ExamplePyramidFloat<>(GrayU8.class);

		app.standard();
		app.process(image);
	}
}