Method of providing image feature descriptors

Abstract

A method of providing a set of feature descriptors configured to be used in matching an object in an image of a camera is provided. The method includes: a) providing at least two images of a first object; b) extracting in at least two of the images at least one feature from the respective image, c) providing at least one descriptor for an extracted feature, and storing the descriptors; d) matching descriptors in the first set of descriptors; e) computing a score parameter based on the result of the matching process; f) selecting at least one descriptor based on its score parameter; g) adding the selected descriptor(s) to a second set of descriptors; and h) updating the score parameter of descriptors in the first set based on a selection process and to the result of the matching process.

Description

[0001]This application is entitled to the benefit of, and incorporates by reference essential subject matter disclosed in PCT Application No. PCT / EP2012 / 064422 filed on Jul. 23, 2012.BACKGROUND OF THE INVENTION[0002]1. Technical Field[0003]The invention is related to a method of providing a set of feature descriptors configured to be used in matching at least one feature of an object in an image of a camera, and a corresponding computer program product for performing the method.[0004]2. Background Information[0005]Such method may be used among other applications, for example, in a method of determining the position and orientation of a camera with respect to an object. A common approach to determine the position and orientation of a camera with respect to an object with a known geometry and visual appearance uses 2D-3D correspondences gained by means of local feature descriptors, such as SIFT described in D. G. Lowe. Distinctive image features from scale-invariant keypoints. Int. Jo...

AI Technical Summary

Benefits of technology

The patent describes a method to automatically determine a set of feature descriptors that can be used to match and locate objects under various viewing conditions. This method involves generating a database of image features from different viewpoints and detecting matches between them to find the most representative descriptors. This helps to improve the accuracy of matching and localization without adding much computational load. The method can be applied on mobile devices with limited memory capacity, such as mobile phones. The technical effect of this invention is to reduce the size of the feature database, which makes object recognition faster and more efficient, regardless of changes in viewpoint, illumination, and camera parameters.

Problems solved by technology

Journal Computer Vision, 65:43-72, 2005. that estimate an affine transformation to normalize the neighborhood of a feature exist, but they are currently too expensive for real-time applications on mobile devices.

Invariance to out-of-plane rotations, however, is usually fairly limited and in the responsibility of the description method itself.

However, with an increasing number of reference feature descriptors, the time to match a single current feature descriptor increases, making real-time processing impossible at some point.

Additionally, the amount of reference data, which potentially needs to be transferred via mobile networks, increases which results in longer loading times.

However, the probabilities that need to be stored for feature classifiers require a lot of memory, which makes them unfeasible for a large amount of features in particular on memory-limited mobile devices.

However, the existing methods making use of this result in large amount of descriptor data making them almost unfeasible on mobile devices.

Images