Package boofcv.factory.sfm

Class 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 Description
      FactoryMotion2D()  

    • Method Summary

      All Methods Static Methods Concrete Methods 
      Modifier and Type Method 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.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.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