ResearchBib Share Your Research, Maximize Your Social Impacts
Sign for Notice Everyday Sign up >> Login

RESEARCH ON FEATURE POINTS EXTRACTION METHOD FOR BINARY MULTISCALE AND ROTATION INVARIANT LOCAL FEATURE DESCRIPTOR

Journal: ICTACT Journal on Image and Video Processing (IJIVP) (Vol.5, No. 1)

Publication Date:

Authors : ; ; ; ; ; ;

Page : 873-878

Keywords : Features Extraction; Multiscale; Feature Descriptor; Corner Detection; Rotation Invariant;

Source : Downloadexternal Find it from : Google Scholarexternal

Abstract

An extreme point of scale space extraction method for binary multiscale and rotation invariant local feature descriptor is studied in this paper in order to obtain a robust and fast method for local image feature descriptor. Classic local feature description algorithms often select neighborhood information of feature points which are extremes of image scale space, obtained by constructing the image pyramid using certain signal transform method. But build the image pyramid always consumes a large amount of computing and storage resources, is not conducive to the actual applications development. This paper presents a dual multiscale FAST algorithm, it does not need to build the image pyramid, but can extract feature points of scale extreme quickly. Feature points extracted by proposed method have the characteristic of multiscale and rotation Invariant and are fit to construct the local feature descriptor.

Last modified: 2014-09-02 13:58:23