一种基于信息融合的图像分割方法  

The testing and discussion of an image segmenting method based on information fusion

作　　者：曾庆贵[1] 宋世军[1] 张佳福[1] 刘胜利[1] 

机构地区：[1]山东建筑大学机电工程学院,山东济南250101

出　　处：《山东建筑大学学报》2008年第2期154-158,共5页Journal of Shandong Jianzhu University

摘　　要：轮廓提取是计算机视觉的基础步骤之一,是图像分析的基础,国内外在此方面都有一定研究,但轮廓的提取由于受到诸如光源、目标形状、成像原理等原因的影响,特别是光源的影响,算法的通用性一般很差。针对以往算法的局限性,本文从色度、纹理入手并结合数学形态学,提出了一种基于信息融合的图像分割方法。经实验证明,在光源变化不大的情况下,该法对不同材质的轮廓提取有较好的通用性。Image segmentation is one of the basic steps in machine vision and the base of image analysis. Many studies have been done in this field in China and abroad. However, due to the influences of light source, target shape, image theory and so on, especially the former factor, the segmenting method has a poor compatibility. Comparing with the limitation of the previous image segmenting methods, this paper, taking hue and texture as well as the method of mathematical morphology into account, proposes a segmenting method based on information fusion. Experiments show that in the condition of no obvious change of light source, the segmenting method has a better compatibility with different homogeneous objects.

关 键 词：信息融合 色度 纹理 图像分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]