Our website uses necessary cookies to enable basic functions and optional cookies to help us to enhance your user experience. Learn more about our cookie policy by clicking "Learn More".
Accept All Only Necessary Cookies
Feature Detection using OpenCV icône

1.01 by Robert Woodley


Feb 18, 2016

À propos de Feature Detection using OpenCV

Experiment with various OpenCV Feature Detection and Matching algorithms.

This app demonstrates several of the OpenCV Feature Detection algorithms.

When you first run it, it will prompt you to download opencv from Google Play. This is painless and necessary.

There is a video that explains how to use it here: https://youtu.be/cXuRiw--KH4

Or, read on:

To start, point the camera at an object you'd like to track. Then hit 'Train'. The object should be on a solid background so that no extraneous feature points are detected.

Then try moving the camera around and watch the app try to locate your object on the screen.

There are 3 modes which are accessible from the '...' menu in the upper right:

1. Match mode. This is the default. It shows the original object and draws lines to the feature points it is able to match.

2. Box mode. This uses findHomography() routine to draw a box around the object. Not perfect if the object is too small. Designed for planar objects.

3. Draw key points. Just shows you the detected key points.

You can choose one of several algorithms:

1. ORB Detector, ORB Descriptor Extractor.

2. ORB Detector, FREAK Descriptor Extractor.

3. BRISK Detector, BRIEF Descriptor Extractor.

4. STAR Detector, BRIEF Descriptor Extractor.

There are several matching modes you can experiment with.

1. KNN: K-nearest neighbor matcher. Finds the 15 nearest matches for each feature.

2. Ratio Check: Finds the 2 nearest neighbors and rejects any feature where the 2 nearest matches are too similar.

3. Cross-check: Filters matches to those which agree backwards and forwards.

Ratio Check, Cross-Check filter matches, reduce the total count. You can also filter on 'distance' which is really a quality measure, using the slider.

There are 2 Seek Bars in the app:

The first one controls the max distance beyond which a match will be excluded.

The second one controls the Ransac threshold. It is only used for Homography. (Drawing the box).

Source Code is at: https://github.com/rwoodley/SteamSpace/tree/master/AndroidApps/FeatureDetectorApp

Chargement de la traduction...

Informations Application supplémentaires

Dernière version

Demande Feature Detection using OpenCV mise à jour 1.01

Telechargé par

Mohammad Shoyeb Ansari

Nécessite Android

Android 4.4+

Voir plus

Quoi de neuf dans la dernière version 1.01

Last updated on Feb 18, 2016

Minor bug fixes and improvements. Install or update to the newest version to check it out!

Voir plus

Feature Detection using OpenCV Captures d'écran

Articles populaires dans les dernières 24 heures

Charegement du commentaire...
Abonnez-vous à APKPure
Soyez le premier à avoir accès à la sortie précoce, aux nouvelles et aux guides des meilleurs jeux et applications Android.
Non merci
S'inscrire
Abonné avec succès!
Vous êtes maintenant souscrit à APKPure.
Abonnez-vous à APKPure
Soyez le premier à avoir accès à la sortie précoce, aux nouvelles et aux guides des meilleurs jeux et applications Android.
Non merci
S'inscrire
Succès!
Vous êtes maintenant souscrit à notre newsletter.