boofcv.alg.shapes.corner

Class RefineCornerLinesToImage<T extends boofcv.struct.image.ImageSingleBand>

java.lang.Object

boofcv.alg.shapes.corner.RefineCornerLinesToImage<T>

public class RefineCornerLinesToImage<T extends boofcv.struct.image.ImageSingleBand>
extends java.lang.Object

Refines the estimate of a corner. A corner is defined as the intersection of two straight edges. The inside edge is either all darker or lighter than the background. The optimization function when refining the edge seeks to maximize the difference between the inside and outside of the edge. Internally SnapToEdge is used to perform this optimization.

The corner is defined using three points. One corner and two line end points. One is to the left of the right line. Left is defined as counter-clockwise. Optimization is done by fitting a line with that initial seed. It is then iteratily updated by finding new end points with the new line that are the same distance away. Pixels near the corner are excluded from the optimization because the edge is less clear at that point.

For input polygons which are in undistorted coordinates by with a distorted image call getSnapToEdge() and invoke BaseIntegralEdge.setTransform(PixelTransform_F32)}.

Field Summary

Fields

Modifier and Type	Field and Description
protected T	image

Constructor Summary

Constructors

Constructor and Description
RefineCornerLinesToImage(boolean fitBlack, java.lang.Class<T> imageType) Simplified constructor which uses reasonable default values for most variables
RefineCornerLinesToImage(double cornerOffset, int maxLineSamples, int sampleRadius, int maxIterations, double convergeTolPixels, boolean fitBlack, java.lang.Class<T> imageType) Constructor which provides full access to all parameters.

Method Summary

Modifier and Type	Method and Description
georegression.struct.point.Point2D_F64	getRefinedCorner()
georegression.struct.point.Point2D_F64	getRefinedEndLeft()
georegression.struct.point.Point2D_F64	getRefinedEndRight()
SnapToEdge<T>	getSnapToEdge()
boolean	isFitBlack() True if it is fitting black to the image
protected boolean	optimize(georegression.struct.point.Point2D_F64 a, georegression.struct.point.Point2D_F64 b, georegression.struct.line.LineGeneral2D_F64 found, boolean darkRight) Fits a line defined by the two points.
boolean	refine(georegression.struct.point.Point2D_F64 corner, georegression.struct.point.Point2D_F64 endLeft, georegression.struct.point.Point2D_F64 endRight) Refines the fit a polygon by snapping it to the edges.
void	setImage(T image) Sets the image which is going to be processed.

Methods inherited from class java.lang.Object

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

Field Detail

image

protected T extends boofcv.struct.image.ImageSingleBand image

Constructor Detail

RefineCornerLinesToImage

public RefineCornerLinesToImage(double cornerOffset,
                                int maxLineSamples,
                                int sampleRadius,
                                int maxIterations,
                                double convergeTolPixels,
                                boolean fitBlack,
                                java.lang.Class<T> imageType)

Constructor which provides full access to all parameters. See code documents value a description of these variables.

Parameters:

cornerOffset - pixels this close to the corner will be ignored. Try 2

maxLineSamples - Number of points along the line which will be sampled. try 10

sampleRadius - How far away from the line will it sample pixels. >= 1

maxIterations - Maximum number of iterations it will perform. Try 10

convergeTolPixels - When the corner changes less than this amount it will stop iterating. Try 1e-5

fitBlack - If true it's fitting a black shape with a white background. false is the inverse.

RefineCornerLinesToImage

public RefineCornerLinesToImage(boolean fitBlack,
                                java.lang.Class<T> imageType)

Simplified constructor which uses reasonable default values for most variables

Parameters:

fitBlack - If true it's fitting a black shape with a white background. false is the inverse.

imageType - Type of input image it processes

Method Detail

setImage

public void setImage(T image)

Sets the image which is going to be processed. If a transform is to be used BaseIntegralEdge.setTransform(boofcv.struct.distort.PixelTransform_F32) should be called before this.

refine

public boolean refine(georegression.struct.point.Point2D_F64 corner,
                      georegression.struct.point.Point2D_F64 endLeft,
                      georegression.struct.point.Point2D_F64 endRight)

Refines the fit a polygon by snapping it to the edges.

Parameters:

corner - (input) Initial estimate of polygon corner. Not modified.

endLeft - (input) End point of left line. Not modified.

endRight - (input) End point of right line. Not modified.

optimize

protected boolean optimize(georegression.struct.point.Point2D_F64 a,
                           georegression.struct.point.Point2D_F64 b,
                           georegression.struct.line.LineGeneral2D_F64 found,
                           boolean darkRight)

Fits a line defined by the two points. When fitting the line the weight of the edge is used to determine how influential the point is

Parameters:

a - Corner point in image coordinates.

b - Corner point in image coordinates.

found - (output) Line in image coordinates

Returns:

true if successful or false if it failed

getRefinedCorner

public georegression.struct.point.Point2D_F64 getRefinedCorner()

getRefinedEndLeft

public georegression.struct.point.Point2D_F64 getRefinedEndLeft()

getRefinedEndRight

public georegression.struct.point.Point2D_F64 getRefinedEndRight()

getSnapToEdge

public SnapToEdge<T> getSnapToEdge()

isFitBlack

public boolean isFitBlack()

True if it is fitting black to the image