osvr::vbtracker::EdgeHoleBasedLedExtractor Class Reference

Public Types
using	MatType = cv::Mat
using	ExternalMatGetterReturn = cv::Mat const &
using	ContourId = std::size_t
using	RejectType = std::tuple< ContourId, RejectReason, cv::Point2d >
using	RejectList = std::vector< RejectType >

Public Member Functions
	EdgeHoleBasedLedExtractor (EdgeHoleParams const &extractorParams=EdgeHoleParams())
LedMeasurementVec const &	operator() (cv::Mat const &gray, BlobParams const &p, bool verboseBlobOutput=false)
	~EdgeHoleBasedLedExtractor ()
	out of line for unique_ptr-based pimpl.
void	reset ()
ExternalMatGetterReturn	getInputGrayImage () const
ExternalMatGetterReturn	getEdgeDetectedImage () const
ExternalMatGetterReturn	getEdgeDetectedBinarizedImage () const
ContourList const &	getContours () const
LedMeasurementVec const &	getMeasurements () const
RejectList const &	getRejectList () const

Member Function Documentation

operator()()

LedMeasurementVec const & osvr::vbtracker::EdgeHoleBasedLedExtractor::operator()	(	cv::Mat const &	gray,
		BlobParams const &	p,
		bool	verboseBlobOutput = false
	)

Set up the threshold parameters

Early out - empty image!

Used to do basic thresholding here first to reduce background noise, but turns out that actually produced worse results at the end of the process (presumably by producing very sharp edges)

Edge detection: apply a laplacian filter to this frame

removal of mjpeg artifacts.

Extract beacons from the edge detection image

---

The documentation for this class was generated from the following files:

• plugins/videotrackershared/EdgeHoleBasedLedExtractor.h
• plugins/videotrackershared/EdgeHoleBasedLedExtractor.cpp