欢迎访问林业科学,今天是

林业科学 ›› 2011, Vol. 47 ›› Issue (5): 106-111.doi: 10.11707/j.1001-7488.20110517

• 论文 • 上一篇    下一篇

基于AOS的扩展C-V模型及背景填充耦合的单板节子缺陷识别

王阿川, 于琳瑛, 曹军   

  1. 东北林业大学 哈尔滨 150040
  • 收稿日期:2010-01-21 修回日期:2010-04-12 出版日期:2011-05-25 发布日期:2011-05-25

An Inspection for the Veneer Knot Defect Based on Extension of C-V Model and AOS Scheme Coupling with Technique of Painting Background

Wang Achuan, Yu Linying, Cao Jun   

  1. Northeast Forestry University Harbin 150040
  • Received:2010-01-21 Revised:2010-04-12 Online:2011-05-25 Published:2011-05-25

摘要:

分析木材单板节子缺陷图像的特点,提出一种基于AOS的扩展C-V模型及背景填充耦合的单板节子缺陷识别算法。首先,对Chan-Vese提出的基于Mumford-Shah模型的水平集图像分割算法进行改进,使分割速度得到提高; 其次,用AOS算法改进原模型的差分格式,使得差分格式无条件稳定; 最后,结合背景填充技术,使得到的新图像缩减了目标与背景间的特征差别。通过对比试验,表明该分割方法能够快速识别单板单个节子缺陷,充分说明该耦合方法比Chan-Vese方法及其改进方法有更好的分割效果。通过用多水平集作为初始轮廓演化曲线,结果表明该方法也可快速实现对单板多节子缺陷图像的识别,实现对单板节子图像的多目标分割。

关键词: 单板节子缺陷图像, 扩展的C-V模型, AOS算法, 多目标分割, 背景填充技术

Abstract:

The features of the defects image of wood veneer are analyzed, an algorithm of veneer knot defect inspection is proposed which based on extension of C-V model and AOS scheme coupling with filling background. Firstly, improved a new level set approach for image segmentation which was proposed by Chan-Vese based on Mumford-Shah, the algorithm could increase the speed of segmentation. Secondly, the approach used AOS scheme, which is unconditional stable, to discrete the level set function. Finally, coupling with the technique of background make a new image which reduced the different characteristics between the target and background. Contrast experimental results show that, this method can fast inspect single knot defect, reveal this coupling method has better segmentation effect than C-V method and improving method. Using multiple level sets as the initial contours of curve evolution, experimental results show that this method could also quickly inspect the image of veneer multi-knot defect and achieve the mutli-object segmentation of veneer knot.

Key words: veneer knot defect image, extended C-V model, AOS scheme, multi-object segmentation, technique of painting background

中图分类号: