计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (23): 164-166.

• 图形、图像、模式识别 • 上一篇    下一篇

基于空间最大类间距的像素群分割优化

杨 波1,2,何小海1,左 航3   

  1. 1.四川大学 电子信息学院,成都 610065
    2.西南电子电信技术研究所,成都 610041
    3.四川大学 计算机软件学院,成都 610065
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-08-11 发布日期:2011-08-11

Segment refinement of pixel group based on OTSU

YANG Bo1,2,HE Xiaohai1,ZUO Hang3   

  1. 1.College of Electronic Information,Sichuan University,Chengdu 610065,China
    2.Southwest Institute of Electronic Telecom,Chengdu 610041,China
    3.College of Computer Science and Software Engineering,Sichuan University,Chengdu 610065,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-08-11 Published:2011-08-11

摘要: 目标分割是利用图像的时空先验知识对像素点进行逐点分类的一个过程,目标的固有属性体现在密度及其形状即二维平面的空间分布上。既往的阈值分割大多仅利用了目标的密度信息,而忽略了其空间分布信息。鉴于形状先验知识难以获得,提出利用双阈值分割得到临界区域和目标重心,其后以临界区域像素群距目标重心的距离作为形状的空间分布信息,并由最大类间距算法对该距离进行分割从而实现优化。实验结果显示算法在降低噪声和削弱阴影的同时,有效地改善了目标提取的完整性。

关键词: 像素群, 分割优化, 双阈值, 最大类间距

Abstract: Object segment is essentially a process of classifying pixel with space-time information which is expressed in both density and shape related space distribution.Usually segment with threshold only considers density while ignoring space distribution.While the shape is hard to acquire,this paper proposes an algorithm that employs OTSU to re-segment critical pixels according to their space distribution against object gravity center and thus refine bi-threshold segment result.The experiment shows this method can improve the integrity of object and restrain noise and shade.

Key words: pixel group, segment refinement, bi-threshold, OTSU