• CN:11-2187/TH
  • ISSN:0577-6686

›› 2006, Vol. 42 ›› Issue (增刊): 170-174.

• 论文 • 上一篇    下一篇

扫码分享

基于彩色图像的运动目标分割方法

杨杰;张铭钧;徐建安   

  1. 哈尔滨工程大学机电工程学院
  • 发布日期:2006-12-30

MOVING OBJECT SEGMENTATION METHOD BASED ON COLOR IMAGE

YANG Jie;ZHANG Mingjun;XU Jianan   

  1. School of Mechatronics Engineering, Harbin Engineering University
  • Published:2006-12-30

摘要: 针对彩色图像Hue-saturation-intensity颜色模型各分量所携带的颜色信息不同的特点,结合区域分割算法在彩色图像中提取区域颜色特征,并将区域颜色特征作为局部阈值分割方法的阈值特征,提出一种基于图像区域颜色特征的运动目标分割方法。分割试验显示了该运动目标分割方法的有效性和可行性。提出融合梯度算子和子区合并技术进行区域分割的方法,子区合并技术能够在一定程度上防止两个独立的区域被错误的合并成一个区域,这种区域分割方法和常用的Region growing算法的对比分割试验显示该方法对灰度变化不均匀的区域进行分割的效果优于Region growing算法。

关键词: 彩色图像, 区域颜色特征, 图像分割

Abstract: Region color features are extracted from color image according to the different color information of the different component in Hue-Saturation-Intensity color model, combined with region segmentation algorithm. Region color features act as the threshold properties which are used in local threshold segmentation method. So a moving object segmentation method according to the image region color features is presented. Segmentation experiments show the effectivity and feasibility of the segmentation method. A region segmentation method, which integrates the subregion merged technique into the gradient algorithm, is presented. Subregion merged technique can prevent two separate regions from incorrect combination to some extent. The contrast experiments of segmentation show the region segmentation method is better than the region growing algorithm on segmenting the region whose gray scale is distributed roughly.

Key words: Color image, Image segmentation, Region color features

中图分类号: