首页 | 官方网站   微博 | 高级检索  
     

基于曲率的透视不变性分割
引用本文:颜灿 丁明跃. 基于曲率的透视不变性分割[J]. 红外与激光工程, 1998, 27(3): 22-27,37
作者姓名:颜灿 丁明跃
作者单位:华中理工大学图像识别与人工智能研究所,图像处理与智能控制国家教委开放研究实验室
基金项目:武汉测绘科技大学测绘遥感信息工程国家重点实验室开放基金
摘    要:曲线的透视不变性分割在基于二次曲线的三维重建中非常重要。首先给出了连续曲线零曲率点和角点的提取法,在数字图像中曲经是不连续的,曲率的变化受噪声的干扰很大。在这种情况下为了检测出零曲率点和角点,对曲线采用了高斯滤波然后再提到零曲率点和角点。实验证明这种方法效果优于直接求曲率法。

关 键 词:透视不变性 高斯平滑 零曲率 曲线分割 三维重建

PERSPECTIVE INVARIANCE SEGMENTATIONBASED ON CURVATURE
Yan Can ,Ding Mingyue. PERSPECTIVE INVARIANCE SEGMENTATIONBASED ON CURVATURE[J]. Infrared and Laser Engineering, 1998, 27(3): 22-27,37
Authors:Yan Can   Ding Mingyue
Abstract:Perspective invariance segmentation of curves is very important in quadric-based 3-D re- construction . First an extraction approach of zero curvature and corner points of curves in continuous case is given, In digital images, curvature of curves is not the same with the situation in continuous cases and very sensitive to noise or other disturbance in image. In order to get a robust detection ap- proach of zero curvature and corner points, Gaussian kernel is used to smooth curves firstly and to fur- ther extract zero curvature and corner points in curves . Our experimental results are promising com- pared with the direct detecting of curvature.
Keywords:Perspective invariance Gaussian smoothing Zero curvature Corner points Curve segmentation
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号