NDEVR
API Documentation
PhotogrammetryFeatures

Handles feature extraction (SIFT/AKAZE) and feature matching (cascade hashing / brute force). More...

Static Public Member Functions

static FeatureExtractionResult extractFeatures (const Buffer< File > &original_images, const Buffer< File > &working_images, const Buffer< Vector< 2, uint04 > > &working_sizes, openMVG::sfm::SfM_Data &sfm_data, const PhotogrammetryOptions &options, LogPtr log, ProgressFn progress_fn, const bool &cancel_flag)
 Extract SIFT/AKAZE features from images, parse EXIF GPS, populate sfm_data views.
static uint04 gpsNeighborK (PhotogrammetryQuality quality)
 Number of GPS-nearest neighbours to match per image when a GPS position graph is used.
static std::shared_ptr< InMemoryMatchesProvider > matchFeatures (const std::shared_ptr< InMemoryRegionsProvider > &regions, const std::unordered_map< uint32_t, std::array< double, 3 > > &gps_lla, const openMVG::sfm::SfM_Data &sfm_data, const PhotogrammetryOptions &options, LogPtr log, ProgressFn progress_fn)
 Build pair graph (GPS k-NN or exhaustive) and run feature matching.

Detailed Description

Handles feature extraction (SIFT/AKAZE) and feature matching (cascade hashing / brute force).

All methods are static — no instance state.

Definition at line 67 of file PhotogrammetryFeatures.h.

Member Function Documentation

◆ extractFeatures()

FeatureExtractionResult PhotogrammetryFeatures::extractFeatures ( const Buffer< File > & original_images,
const Buffer< File > & working_images,
const Buffer< Vector< 2, uint04 > > & working_sizes,
openMVG::sfm::SfM_Data & sfm_data,
const PhotogrammetryOptions & options,
LogPtr log,
ProgressFn progress_fn,
const bool & cancel_flag )
static

Extract SIFT/AKAZE features from images, parse EXIF GPS, populate sfm_data views.

Parameters
working_sizesActual pixel dimensions of each working image.

◆ gpsNeighborK()

uint04 PhotogrammetryFeatures::gpsNeighborK ( PhotogrammetryQuality quality)
inlinestatic

Number of GPS-nearest neighbours to match per image when a GPS position graph is used.

Higher values increase match redundancy at the cost of more matching time. High quality uses more neighbours to capture all overlapping views in dense drone flight patterns (typically 70-80% overlap -> ~20-30 actual neighbours).

Definition at line 76 of file PhotogrammetryFeatures.h.

References e_high, and e_normal.

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