Package boofcv.factory.sfm

Class ConfigVisOdomTrackPnP

  • All Implemented Interfaces:
    boofcv.struct.Configuration, java.io.Serializable
    public class ConfigVisOdomTrackPnP
extends java.lang.Object
implements boofcv.struct.Configuration
    Base class for visual odometry algorithms based on PointTracker.
    See Also:
    Serialized Form

    • Field Summary

      Fields 
      Modifier and Type Field Description
      boofcv.factory.geo.ConfigBundleAdjustment bundle
      Configuration for Bundle Adjustment
      boofcv.misc.ConfigConverge bundleConverge
      Convergence criteria for bundle adjustment.
      int bundleMaxFeaturesPerFrame
      Maximum number of features optimized in bundle adjustment per key frame.
      int bundleMinObservations
      Minimum number of observations a track must have before it is included in bundle adjustment.
      int dropOutlierTracks
      Drop tracks if they have been outliers for this many frames in a row
      ConfigKeyFrameManager keyframes
      Specifies when a new key frame is created
      int maxKeyFrames
      Maximum number of key frames it will save.
      boofcv.factory.geo.EnumPNP pnp
      Which PNP solution to use
      boofcv.factory.geo.ConfigRansac ransac
      Configuration for RANSAC.
      int refineIterations
      Number of iterations to perform when refining the initial frame-to-frame motion estimate.

    • Field Detail

      • bundle

        public boofcv.factory.geo.ConfigBundleAdjustment bundle
        Configuration for Bundle Adjustment

      • bundleConverge

        public boofcv.misc.ConfigConverge bundleConverge
        Convergence criteria for bundle adjustment. Set max iterations to ≤ 0 to disable

      • bundleMaxFeaturesPerFrame

        public int bundleMaxFeaturesPerFrame
        Maximum number of features optimized in bundle adjustment per key frame. This is a very good way to limit the amount of CPU used. If not positive then unlimited. ≤ 0 to disable.

      • bundleMinObservations

        public int bundleMinObservations
        Minimum number of observations a track must have before it is included in bundle adjustment. Has to be ≥ 2 and it's strongly recommended that this is set to 3 or higher. Due to ambiguity along epipolar lines there can be lots of false positives with just two views. With three views there is a unique solution and that tends to remove most false positives.

      • dropOutlierTracks

        public int dropOutlierTracks
        Drop tracks if they have been outliers for this many frames in a row

      • maxKeyFrames

        public int maxKeyFrames
        Maximum number of key frames it will save. Must be at least 4

      • ransac

        public boofcv.factory.geo.ConfigRansac ransac
        Configuration for RANSAC. Used to robustly estimate frame-to-frame motion

      • refineIterations

        public int refineIterations
        Number of iterations to perform when refining the initial frame-to-frame motion estimate. Disable ≤ 0

      • pnp

        public boofcv.factory.geo.EnumPNP pnp
        Which PNP solution to use

    • Constructor Detail

      • ConfigVisOdomTrackPnP

        public ConfigVisOdomTrackPnP()

    • Method Detail

      • checkValidity

        public void checkValidity()
        Specified by:
        checkValidity in interface boofcv.struct.Configuration