boofcv.factory.sfm

Class FactoryMotion2D

java.lang.Object

boofcv.factory.sfm.FactoryMotion2D

public class FactoryMotion2D
extends java.lang.Object

Factory for creating algorithms related to 2D image motion. Typically used for image stabilization, mosaic, and motion detection in video feeds.

Constructor Summary

Constructors

Constructor and Description
FactoryMotion2D()

Method Summary

Modifier and Type	Method and Description
static <I extends boofcv.struct.image.ImageBase<I>,IT extends georegression.struct.InvertibleTransform> ImageMotion2D<I,IT>	createMotion2D(int ransacIterations, double inlierThreshold, int outlierPrune, int absoluteMinimumTracks, double respawnTrackFraction, double respawnCoverageFraction, boolean refineEstimate, boofcv.abst.feature.tracker.PointTracker<I> tracker, IT motionModel) Estimates the 2D motion of an image using different models.
static <I extends boofcv.struct.image.ImageBase<I>,IT extends georegression.struct.InvertibleTransform> StitchingFromMotion2D<I,IT>	createVideoStitch(double maxJumpFraction, ImageMotion2D<I,IT> motion2D, boofcv.struct.image.ImageType<I> imageType) Estimates the image motion then combines images together.

Methods inherited from class java.lang.Object

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

Constructor Detail

FactoryMotion2D

public FactoryMotion2D()

Method Detail

createMotion2D

public static <I extends boofcv.struct.image.ImageBase<I>,IT extends georegression.struct.InvertibleTransform> ImageMotion2D<I,IT> createMotion2D(int ransacIterations,
                                                                                                                                                  double inlierThreshold,
                                                                                                                                                  int outlierPrune,
                                                                                                                                                  int absoluteMinimumTracks,
                                                                                                                                                  double respawnTrackFraction,
                                                                                                                                                  double respawnCoverageFraction,
                                                                                                                                                  boolean refineEstimate,
                                                                                                                                                  boofcv.abst.feature.tracker.PointTracker<I> tracker,
                                                                                                                                                  IT motionModel)

Estimates the 2D motion of an image using different models.

Type Parameters:

I - Image input type.

IT - Model model

Parameters:

ransacIterations - Number of RANSAC iterations

inlierThreshold - Threshold which defines an inlier.

outlierPrune - If a feature is an outlier for this many turns in a row it is dropped. Try 2

absoluteMinimumTracks - New features will be respawned if the number of inliers drop below this number.

respawnTrackFraction - If the fraction of current inliers to the original number of inliers drops below this fraction then new features are spawned. Try 0.3

respawnCoverageFraction - If the area covered drops by this fraction then spawn more features. Try 0.8

refineEstimate - Should it refine the model estimate using all inliers.

tracker - Point feature tracker.

motionModel - Instance of the model model used. Affine2D_F64 or Homography2D_F64

Returns:

ImageMotion2D

createVideoStitch

public static <I extends boofcv.struct.image.ImageBase<I>,IT extends georegression.struct.InvertibleTransform> StitchingFromMotion2D<I,IT> createVideoStitch(double maxJumpFraction,
                                                                                                                                                             ImageMotion2D<I,IT> motion2D,
                                                                                                                                                             boofcv.struct.image.ImageType<I> imageType)

Estimates the image motion then combines images together. Typically used for mosaics and stabilization.

Type Parameters:

I - Image input type.

IT - Model model

Parameters:

maxJumpFraction - If the area changes by this much between two consecuative frames then the transform is reset.

motion2D - Estimates the image motion.

imageType - Type of image processed

Returns:

StitchingFromMotion2D