edu.mit.sketch.clocksketch.analysis
Class DrawingAnalyzer

java.lang.Object
  edu.mit.sketch.clocksketch.analysis.DrawingAnalyzer

public class DrawingAnalyzer
extends java.lang.Object

This class provides ways to analyze and identify aspects of a PenDrawing. It also organizes PenDrawings by putting Strokes into Symbols

Author:

David Pitman

Nested Class Summary

static class

DrawingAnalyzer.DigitOrder Enums for the order in which the digits were drawn.

Field Summary

static java.lang.String	ARC
static java.lang.String	COMPLEX
static java.lang.String	ELLIPSE These are all simple symbol types that we classify a stroke(s) as
static double	handOutsideRadiusFraction The fraction that determines how far away from the center of the clockface one end point of the hand must be in terms of the clockface's radius Determined experimentally with patient data to be, approximately, a good value.
static double	handWithinRadiusFraction The fraction that determines how close the center of the clockface one end point of the hand must be in terms of the clockface's radius Determined experimentally with patient data to be, approximately, a good value.
static int	k Global variable related to regression analysis.
static java.lang.String	LINE
static java.lang.String	NOTCLASSIFIED
static java.lang.String	POLYGON
static java.lang.String	SPIRAL

Constructor Summary

DrawingAnalyzer()

DrawingAnalyzer(PenDrawing drawingArg)
Constructs a DrawingAnalyzer with a Drawing passed in.

Method Summary

static double	calculateAdjustedMean(double[] valuesArg, int fraction) Produces the equivalent of an interquartile mean, but instead of removing the lower and top fourth of the sorted data, removes 1/fraction from the top and bottom of the data The array passed in is sorted by size within the function.
static double[]	calculateLinearRegression(double[][] points) Calculates the values of linear regression given the set of points.
static void	classifyClockfaceDrawing(PenDrawing drawing) Goes through the strokes in the drawing and attempts to classify all of the strokes into various Clockface symbols.
static double[]	endPointsCorrelation(Point startPoint, Point endPoint, double lengthOfStroke) Returns the x, y, and shortest-distance correlation (between 0 and 1) of how close the endpoints are given the length of the stroke.
static int	findValuesAbove(double[] valuesArg, double threshold, double thresholdmultiplier) Given the list of values, a threshold, and a multiplier for that threshhold, returns the number of values > (threshhold*threshold multiplier)
static double	getDistance(Point firstPoint, Point secondPoint) Finds the Euclidian distance between two points.
static double	getDistanceWithoutHooklets(Stroke stroke)
static double	getLengthWithoutHooklets(Stroke stroke)
static boolean	isCircle(Stroke stroke)
static boolean	isLine(Stroke stroke)
static double	summedValueDistribution(double[] valuesArg, double sigma) Returns the sum of all the Gaussian distribution results of the values.

Methods inherited from class java.lang.Object

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

Field Detail

ELLIPSE

public static final java.lang.String ELLIPSE

These are all simple symbol types that we classify a stroke(s) as

See Also:

Constant Field Values

ARC

public static final java.lang.String ARC

See Also:

Constant Field Values

LINE

public static final java.lang.String LINE

See Also:

Constant Field Values

POLYGON

public static final java.lang.String POLYGON

See Also:

Constant Field Values

COMPLEX

public static final java.lang.String COMPLEX

See Also:

Constant Field Values

SPIRAL

public static final java.lang.String SPIRAL

See Also:

Constant Field Values

NOTCLASSIFIED

public static final java.lang.String NOTCLASSIFIED

See Also:

Constant Field Values

k

public static final int k

Global variable related to regression analysis.

See Also:

Constant Field Values

handWithinRadiusFraction

public static final double handWithinRadiusFraction

The fraction that determines how close the center of the clockface one end point of the hand must be in terms of the clockface's radius Determined experimentally with patient data to be, approximately, a good value.

See Also:

Constant Field Values

handOutsideRadiusFraction

public static final double handOutsideRadiusFraction

The fraction that determines how far away from the center of the clockface one end point of the hand must be in terms of the clockface's radius Determined experimentally with patient data to be, approximately, a good value.

See Also:

Constant Field Values

Constructor Detail

DrawingAnalyzer

public DrawingAnalyzer()

DrawingAnalyzer

public DrawingAnalyzer(PenDrawing drawingArg)

Constructs a DrawingAnalyzer with a Drawing passed in.

Parameters:

drawingArg -

Method Detail

classifyClockfaceDrawing

public static void classifyClockfaceDrawing(PenDrawing drawing)

Goes through the strokes in the drawing and attempts to classify all of the strokes into various Clockface symbols. Will erase all of the current symbols within the drawing and create new symbols within the drawing.

Parameters:

drawing -

findValuesAbove

public static int findValuesAbove(double[] valuesArg,
                                  double threshold,
                                  double thresholdmultiplier)

Given the list of values, a threshold, and a multiplier for that threshhold, returns the number of values > (threshhold*threshold multiplier)

Parameters:

valuesArg -

threshold -

thresholdmultiplier -

summedValueDistribution

public static double summedValueDistribution(double[] valuesArg,
                                             double sigma)

Returns the sum of all the Gaussian distribution results of the values.

Parameters:

valuesArg -

sigma -

calculateAdjustedMean

public static double calculateAdjustedMean(double[] valuesArg,
                                           int fraction)

Produces the equivalent of an interquartile mean, but instead of removing the lower and top fourth of the sorted data, removes 1/fraction from the top and bottom of the data The array passed in is sorted by size within the function.

Parameters:

valuesArg -

fraction -

isLine

public static boolean isLine(Stroke stroke)

isCircle

public static boolean isCircle(Stroke stroke)

getLengthWithoutHooklets

public static double getLengthWithoutHooklets(Stroke stroke)

getDistanceWithoutHooklets

public static double getDistanceWithoutHooklets(Stroke stroke)

getDistance

public static double getDistance(Point firstPoint,
                                 Point secondPoint)

Finds the Euclidian distance between two points.

Parameters:

firstPoint -

secondPoint -

Returns:

if either point is null, returns NaN

endPointsCorrelation

public static double[] endPointsCorrelation(Point startPoint,
                                            Point endPoint,
                                            double lengthOfStroke)

Returns the x, y, and shortest-distance correlation (between 0 and 1) of how close the endpoints are given the length of the stroke. The method uses a simplified gaussian distribution where r=exp(-(foo^2)/(2*sigma^2)) where foo is the distance (x,y,shortest) and sigma=(lengthOfStroke/(distance between points))

Parameters:

startPoint -

endPoint -

lengthOfStroke -

Returns:

{ x Correlation, y Correlation, shortest-distance Correlation }

calculateLinearRegression

public static double[] calculateLinearRegression(double[][] points)

Calculates the values of linear regression given the set of points.

Parameters:

points - [0] = x value, [1] = y value

Returns:

{beta, alpha, r}