boofcv.alg.feature.disparity.impl

Class StereoDisparityWtoNaive<I extends boofcv.struct.image.ImageGray>

java.lang.Object

boofcv.alg.feature.disparity.impl.StereoDisparityWtoNaive<I>

public class StereoDisparityWtoNaive<I extends boofcv.struct.image.ImageGray>
extends java.lang.Object

Very basic algorithm for testing stereo disparity algorithms for correctness and employs a "winner takes all" strategy for selecting the solution. No optimization is done to improve performance and minimize cache misses. The advantage is that it can take in any image type.

Constructor Summary

Constructors

Constructor and Description
StereoDisparityWtoNaive(int minDisparity, int maxDisparity, int radiusWidth, int radiusHeight) Configure parameters

Method Summary

Modifier and Type	Method and Description
protected double	computeScore(int leftX, int rightX, int centerY) Compute SAD (Sum of Absolute Difference) error.
void	process(I left, I right, boofcv.struct.image.GrayF32 imageDisparity) Computes the disparity for two stereo images along the image's right axis.
protected double	selectBest(int length) Select best disparity using the inner takes all approach

Methods inherited from class java.lang.Object

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

Constructor Detail

StereoDisparityWtoNaive

public StereoDisparityWtoNaive(int minDisparity,
                               int maxDisparity,
                               int radiusWidth,
                               int radiusHeight)

Configure parameters

Parameters:

minDisparity - Minimum disparity it will consider in pixels.

maxDisparity - Maximum allowed disparity in pixels.

radiusWidth - Radius of the region along x-axis.

radiusHeight - Radius of the region along y-axis.

Method Detail

process

public void process(I left,
                    I right,
                    boofcv.struct.image.GrayF32 imageDisparity)

Computes the disparity for two stereo images along the image's right axis. Both image must be rectified.

Parameters:

left - Left camera image.

right - Right camera image.

selectBest

protected double selectBest(int length)

Select best disparity using the inner takes all approach

Parameters:

length - The max allowed disparity at this pixel

Returns:

The best disparity selected.

computeScore

protected double computeScore(int leftX,
                              int rightX,
                              int centerY)

Compute SAD (Sum of Absolute Difference) error.

Parameters:

leftX - X-axis center left image

rightX - X-axis center left image

centerY - Y-axis center for both images

Returns:

Fit score for both regions.