Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (13): 159-160.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Improved method of color image segmentation

JIA Li-juan,ZHANG Guang-nian,GE Qing-ping,JIANG Chang-sheng   

  1. School of Information Engineering,Capital Normal University,Beijing 100037,China
  • Received:2007-07-23 Revised:2007-11-30 Online:2008-05-01 Published:2008-05-01
  • Contact: JIA Li-juan

一种改进的彩色图像分割算法

贾丽娟,张光年,葛庆平,姜长胜   

  1. 首都师范大学 信息工程学院,北京 100037
  • 通讯作者: 贾丽娟

Abstract: This method improves the traditional the approach of color image segmentation based on the fuzzy technology.First,change the color image into gray one.Then paint its gray histogram.For each gray level,we choose one of the pixels,which belong to this gray level,to represent its color.We save the RGB value of each gray level color.Then,according to this color,we build some color fuzzy sets and compute the attachment degree of every color of this image to each color fuzzy set.And based on the principle of maximum degree of membership,we confirm the attachment of every color,and we also build a neural network.This neural network is implemented to segment color images with higher speed and efficiency.

Key words: principle of maximum degree of membership, color image segmentation, gray histogram, neural network

摘要: 文中算法对传统的基于最大隶属度的彩色图像分割算法进行了改进。传统的基于最大隶属度原则的分割方法根据彩色直方图中的色彩矢量来确定目标和背景色模糊集,而绘制彩色直方图时比较繁琐。为了克服这一缺点,先绘制彩色图像对应的灰度直方图。对应于每一个灰度级,取属于该灰度级的像素中的一个像素的颜色来代表这一灰度级所对应的颜色,并保存其RGB值。据此建立一组色彩模糊集。计算图像中的所有色彩在各个模糊集中的隶属度,并基于最大隶属度原则确定色彩的归属。建立一种基于最大隶属度原则的神经网络,实现对彩色图像快速有效的分割。通过灰度直方图判断目标和背景色时的误差很小,并不影响判断。在建立色彩模糊集和计算隶属度实现色彩分类的时候都是根据像素色彩的RGB值来实现的,因而没有颜色损失。实验结果证明,该算法对原有的基于最大隶属度原则的彩色图像分割方法进行了改进,速度快,效果好。

关键词: 最大隶属度原则, 彩色图像分割, 灰度直方图, 神经网络