[其它]Curvature Scale Space Corner Detector

Abstract X7tBpyi  
Corners play an important role in object identification L{&Yh|}  
methods used in machine vision and image processing )YwLj&e4tf  
systems. Single-scale feature detection finds it hard to kcMg`pJ4<  
detect both fine and coarse features at the same time. On n+2>jY  
the other hand, multi-scale feature detection is inherently <l eE.hhf.  
able to solve this problem. This paper proposes an Z4gn7 'V  
improved multi-scale corner detector with dynamic region M+lr [,c  
of support, which is based on Curvature Scale Space (CSS) \,lIPA/L  
technique. The proposed detector first uses an adaptive 7fl{<uf  
local curvature threshold instead of a single global y,qn9  
threshold as in the original and enhanced CSS methods. q:vGGK^  
Second, the angles of corner candidates are checked in a /^X)>1)j  
dynamic region of support for eliminating falsely detected $IdU  
corners. The proposed method has been evaluated over a 79h'sp6;  
number of images and compared with some popular zp``e;gY  
corner detectors. The results showed that the proposed .. W"A3$/nq^  
!>GDp>0  

未注册仅能浏览部分内容，查看全部内容及附件请先 登录 或 注册