boofcv.alg.feature.disparity.impl

Class UtilDisparityScore

java.lang.Object

boofcv.alg.feature.disparity.impl.UtilDisparityScore

public class UtilDisparityScore
extends Object

Contains common functions for computing disparity scores.

Author:

Peter Abeles

Constructor Summary

Constructors

Constructor and Description
UtilDisparityScore()

Method Summary

Methods

Modifier and Type	Method and Description
static void	computeScoreRow(ImageFloat32 left, ImageFloat32 right, int row, float[] scores, int minDisparity, int maxDisparity, int regionWidth, float[] elementScore) Computes disparity score for an entire row.
static void	computeScoreRow(ImageSInt16 left, ImageSInt16 right, int row, int[] scores, int minDisparity, int maxDisparity, int regionWidth, int[] elementScore) Computes disparity score for an entire row.
static void	computeScoreRow(ImageUInt8 left, ImageUInt8 right, int row, int[] scores, int minDisparity, int maxDisparity, int regionWidth, int[] elementScore) Computes disparity score for an entire row.
static void	computeScoreRowSad(ImageFloat32 left, ImageFloat32 right, int elementMax, int indexLeft, int indexRight, float[] elementScore) compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code.
static void	computeScoreRowSad(ImageSInt16 left, ImageSInt16 right, int elementMax, int indexLeft, int indexRight, int[] elementScore) compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code.
static void	computeScoreRowSad(ImageUInt8 left, ImageUInt8 right, int elementMax, int indexLeft, int indexRight, int[] elementScore) compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code.

Methods inherited from class java.lang.Object

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

Constructor Detail

UtilDisparityScore

public UtilDisparityScore()

Method Detail

computeScoreRow

public static void computeScoreRow(ImageUInt8 left,
                   ImageUInt8 right,
                   int row,
                   int[] scores,
                   int minDisparity,
                   int maxDisparity,
                   int regionWidth,
                   int[] elementScore)

Computes disparity score for an entire row. For a given disparity, the score for each region on the left share many components in common. Because of this the scores are computed with disparity being the outer most loop

Parameters:

left - left image

right - Right image

row - Image row being examined

scores - Storage for disparity scores.

minDisparity - Minimum disparity to consider

maxDisparity - Maximum disparity to consider

regionWidth - Size of the sample region's width

elementScore - Storage for scores of individual pixels

computeScoreRowSad

public static void computeScoreRowSad(ImageUInt8 left,
                      ImageUInt8 right,
                      int elementMax,
                      int indexLeft,
                      int indexRight,
                      int[] elementScore)

compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code. Was original inline, but was actually slightly slower by about 3% consistently, It is in its own function so that it can be overridden and have different cost functions inserted easily.

computeScoreRow

public static void computeScoreRow(ImageSInt16 left,
                   ImageSInt16 right,
                   int row,
                   int[] scores,
                   int minDisparity,
                   int maxDisparity,
                   int regionWidth,
                   int[] elementScore)

Computes disparity score for an entire row. For a given disparity, the score for each region on the left share many components in common. Because of this the scores are computed with disparity being the outer most loop

Parameters:

left - left image

right - Right image

row - Image row being examined

scores - Storage for disparity scores.

minDisparity - Minimum disparity to consider

maxDisparity - Maximum disparity to consider

regionWidth - Size of the sample region's width

elementScore - Storage for scores of individual pixels

computeScoreRowSad

public static void computeScoreRowSad(ImageSInt16 left,
                      ImageSInt16 right,
                      int elementMax,
                      int indexLeft,
                      int indexRight,
                      int[] elementScore)

compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code. Was original inline, but was actually slightly slower by about 3% consistently, It is in its own function so that it can be overridden and have different cost functions inserted easily.

computeScoreRow

public static void computeScoreRow(ImageFloat32 left,
                   ImageFloat32 right,
                   int row,
                   float[] scores,
                   int minDisparity,
                   int maxDisparity,
                   int regionWidth,
                   float[] elementScore)

Computes disparity score for an entire row. For a given disparity, the score for each region on the left share many components in common. Because of this the scores are computed with disparity being the outer most loop

Parameters:

left - left image

right - Right image

row - Image row being examined

scores - Storage for disparity scores.

minDisparity - Minimum disparity to consider

maxDisparity - Maximum disparity to consider

regionWidth - Size of the sample region's width

elementScore - Storage for scores of individual pixels

computeScoreRowSad

public static void computeScoreRowSad(ImageFloat32 left,
                      ImageFloat32 right,
                      int elementMax,
                      int indexLeft,
                      int indexRight,
                      float[] elementScore)

compute the score for each element all at once to encourage the JVM to optimize and encourage the JVM to optimize this section of code. Was original inline, but was actually slightly slower by about 3% consistently, It is in its own function so that it can be overridden and have different cost functions inserted easily.