ecto_opencv.features2d

Feature detection cells.

ecto_opencv.features2d

MatchRefinementPnP

Brief doc

A feature descriptor match refiner, using PnP.

Parameters

• inlier_thresh   type: float    not required   default: 30.0

  The thresh hold on number of inliers to consider pose found.

• min_inliers   type: unsigned int    not required   default: 100

  minimum number of inliers

• n_iters   type: unsigned int    not required   default: 100

  number of ransac iterations

• reprojection_error   type: float    not required   default: 8.0

  error threshold

Inputs

• K   type: cv::Mat   

  Camera model.

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

• test   type: cv::Mat   

  The 3d test points.

• train   type: cv::Mat   

  The 3d training points.

Outputs

• R   type: cv::Mat   

• T   type: cv::Mat   

• found   type: bool   

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The verified matches.

• matches_mask   type: cv::Mat   

  The matches mask, same size as the original matches.

MatchRefinementHSvd

Brief doc

A feature descriptor match refiner, using Homography and svd estimation.

Parameters

• inlier_thresh   type: float    not required   default: 25.0

  The inlier threshold of pose found.

• min_inliers   type: unsigned int    not required   default: 100

  minimum number of inliers

• n_iters   type: unsigned int    not required   default: 200

  number of ransac iterations

• reprojection_error   type: float    not required   default: 43.5

  error threshold

Inputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

• test_2d   type: cv::Mat   

  The 2d test points.

• test_3d   type: cv::Mat   

  The 3d test points.

• train_2d   type: cv::Mat   

  The 2d training points.

• train_3d   type: cv::Mat   

  The 3d training points.

Outputs

• R   type: cv::Mat   

• T   type: cv::Mat   

• found   type: bool   

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The verified matches.

• matches_mask   type: cv::Mat   

  The matches mask, same size as the original matches.

SIFT

Brief doc

An ORB detector. Takes a image and a mask, and computes keypoints and descriptors(32 byte binary).

Parameters

• edgeThreshold   type: float    not required   no default value

• nOctaveLayers   type: int    not required   default: 3

• threshold   type: float    not required   no default value

Inputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

• points   type: cv::Mat   

  2d points.

Outputs

• descriptors   type: cv::Mat   

  The descriptors per keypoints

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

MatchRefinement3d

Brief doc

A feature descriptor match refiner, using an affine 3d to 3d estimator.

Inputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

• test   type: cv::Mat   

  The 3d test points.

• train   type: cv::Mat   

  The 3d training points.

Outputs

• R   type: cv::Mat   

• T   type: cv::Mat   

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The verified matches.

• matches_mask   type: cv::Mat   

  The matches mask, same size as the original matches.

LSHMatcher

Brief doc

Given descriptors, find matches.

Parameters

• key_size   type: unsigned int    not required   default: 8

• multi_probe_level   type: unsigned int    not required   default: 1

• n_tables   type: unsigned int    not required   default: 4

• radius   type: unsigned int    not required   default: 55

Inputs

• test   type: cv::Mat   

  Train descriptors.

• train   type: cv::Mat   

  Test descriptors.

• update   type: bool   

  If set to true, update the descriptors.

Outputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

ORBFeature

Brief doc

An ORB feature detector.

Parameters

• first_level   type: int    not required   default: 0

  The first level of the scales

• n_features   type: int    not required   default: 1000

  The number of desired features

• n_levels   type: int    not required   default: 3

  The number of scales

• scale_factor   type: float    not required   default: 1.20000004768

  The factor between scales

Outputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

Matcher

Brief doc

A feature descriptor matcher.

Inputs

• test   type: cv::Mat   

  Train descriptors.

• train   type: cv::Mat   

  Test descriptors.

Outputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

DrawMatches

Brief doc

Draws matches.

Inputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

• matches_mask   type: cv::Mat   

  The descriptor matches mask.

• test   type: cv::Mat   

  Test keypoints.

• test_image   type: cv::Mat   

  Test image.

• train   type: cv::Mat   

  Train keypoints

• train_image   type: cv::Mat   

  Test image.

Outputs

• output   type: cv::Mat   

  An output image.

SIFTDescriptor

Brief doc

A SIFT feature detector.

Parameters

• edgeThreshold   type: float    not required   no default value

• nOctaveLayers   type: int    not required   default: 3

• threshold   type: float    not required   no default value

Outputs

• descriptors   type: cv::Mat   

  The descriptors per keypoints

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

• points   type: cv::Mat   

  2d points.

ORBstats

Brief doc

Prints stats on ORB descriptors.

Inputs

• descriptors   type: cv::Mat   

  The input descriptors.

Outputs

• distances   type: cv::Mat   

  A histogram of the distances in this set.

FeatureDescriptor

Brief doc

Parameters

• json_descriptor_params   type: boost::python::api::object    not required   default: {"type": "ORB", "module": "ecto_opencv.features2d"}

  Parameters for the descriptor as a JSON string. It should have the format: “{“type”:”ORB/SIFT whatever”, “module”:”where_it_is”, “param_1”:val1, ....}

• json_feature_params   type: boost::python::api::object    not required   default: {"type": "ORB", "module": "ecto_opencv.features2d"}

  Parameters for the feature as a JSON string. It should have the format: “{“type”:”ORB/SIFT whatever”, “module”:”where_it_is”, “param_1”:val1, ....}

Inputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

• points   type: cv::Mat   

  2d points.

Outputs

• descriptors   type: cv::Mat   

  The descriptors per keypoints

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

DrawKeypoints

Brief doc

Draws keypoints.

Inputs

• image   type: cv::Mat   

  The input image, used as the base to draw on.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints to draw.

Outputs

• image   type: cv::Mat   

  The output image.

SIFTFeature

Brief doc

A SIFT feature detector.

Parameters

• edgeThreshold   type: float    not required   no default value

• nOctaveLayers   type: int    not required   default: 3

• threshold   type: float    not required   no default value

Outputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

ORBDescriptor

Brief doc

An ORB descriptor extractor.

Outputs

• descriptors   type: cv::Mat   

  The descriptors per keypoints

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

• points   type: cv::Mat   

  2d points.

FASTFeature

Brief doc

A FAST feature detector.

Parameters

• nonmax   type: bool    not required   default: True

  Use the FAST nonmax suppression.

• thresh   type: int    not required   default: 20

  The FAST threshold. 20 is a decent value.

Outputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

KeypointsToMat

Brief doc

Take key points and return an array of the x,y coordinates.

Inputs

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints

Outputs

• points   type: cv::Mat   

  The array of x,y coordinates

MatchRefinement

Brief doc

A feature descriptor match refiner, using a Homography estimator

Parameters

• match_distance   type: double    not required   default: 120.0

  The match distance threshhold.

Inputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The descriptor matches.

• test   type: cv::Mat   

  The test points.

• train   type: cv::Mat   

  The training points.

Outputs

• H   type: cv::Mat   

  The estimated homography.

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The verified matches.

• matches_mask   type: cv::Mat   

  The matches mask, same size as the original matches.

DescriptorAccumulator

Brief doc

Accumulates descriptors.

Inputs

• descriptors   type: cv::Mat   

  The input descriptors.

Outputs

• descriptors   type: cv::Mat   

  A cumulative view of all descriptors.

ORB

Brief doc

An ORB detector. Takes a image and a mask, and computes keypoints and descriptors(32 byte binary).

Parameters

• n_features   type: int    not required   default: 1000

  The number of desired features

• n_levels   type: int    not required   default: 3

  The number of scales

• scale_factor   type: float    not required   default: 1.20000004768

  The factor between scales

Inputs

• image   type: cv::Mat   

  An input image.

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

• mask   type: cv::Mat   

  An mask, same size as image.

• points   type: cv::Mat   

  2d points.

Outputs

• descriptors   type: cv::Mat   

  The descriptors per keypoints

• keypoints   type: std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >   

  The keypoints.

MatchesToMat

Brief doc

Takes matches and turns them into a cv mat alias..

Inputs

• matches   type: std::vector<cv::DMatch, std::allocator<cv::DMatch> >   

  The matches

Outputs

• matches   type: cv::Mat   

  An nx3 matrix of matched indices with distances in the 3rd column