Constructor

obj = cv.BoostDesc()
obj = cv.BoostDesc('OptionName',optionValue, ...)

Options

• Desc Type of descriptor to use, 'BinBoost256' is default (256 bit long dimension). Available types are:
  • BGM
  • BGMHard
  • BGMBilinear
  • LBGM
  • BinBoost64
  • BinBoost128
  • BinBoost256
• UseScaleOrientation Sample patterns using keypoints orientation. default true
• ScaleFactor Adjust the sampling window of detected keypoints.
  • 6.25 is default and fits for cv.KAZE, cv.SURF detected keypoints window ratio
  • 6.75 should be the scale for cv.SIFT detected keypoints window ratio
  • 5.00 should be the scale for cv.AKAZE, cv.MSDDetector, cv.AgastFeatureDetector, cv.FastFeatureDetector, cv.BRISK keypoints window ratio
  • 0.75 should be the scale for cv.ORB keypoints ratio
  • 1.50 was the default in original implementation

Note

BGM is the base descriptor where each binary dimension is computed as the output of a single weak learner. BGMHard and BGMBilinear refer to same BGM but use different type of gradient binning. In the BGMHard that use ASSIGN_HARD binning type the gradient is assigned to the nearest orientation bin. In the BGMBilinear that use ASSIGN_BILINEAR binning type the gradient is assigned to the two neighbouring bins. In the BGM and all other modes that use ASSIGN_SOFT binning type the gradient is assigned to 8 nearest bins according to the cosine value between the gradient angle and the bin center. LBGM (alias FP-Boost) is the floating point extension where each dimension is computed as a linear combination of the weak learner responses. BinBoost and subvariants are the binary extensions of LBGM where each bit is computed as a thresholded linear combination of a set of weak learners.