[其它]Curvature Scale Space Corner Detector

Abstract 0{^@kxV  
Corners play an important role in object identification 9>yLSM,!rS  
methods used in machine vision and image processing '
3TwrY?-  
systems. Single-scale feature detection finds it hard to Ydm0  
detect both fine and coarse features at the same time. On ?oana%  
the other hand, multi-scale feature detection is inherently xP#vAR  
able to solve this problem. This paper proposes an Veo*-sl  
improved multi-scale corner detector with dynamic region H=Scrvfx  
of support, which is based on Curvature Scale Space (CSS) $\M<
gW6  
technique. The proposed detector first uses an adaptive _1Iy/T@1  
local curvature threshold instead of a single global KJn@2x6LP  
threshold as in the original and enhanced CSS methods. #S)]`YW  
Second, the angles of corner candidates are checked in a sL" h  
dynamic region of support for eliminating falsely detected "O{j}QwY  
corners. The proposed method has been evaluated over a 2l]*><q|  
number of images and compared with some popular O#LG$Y n*  
corner detectors. The results showed that the proposed .. =r"-Pm{  
$mdmuUIy-3  

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