基于二维灰度直方图最佳一维投影的图像分割方法  

THE GRAY-LEVEL IMAGE THRESHOLDING METHOD BASED ON THE OPTIMAL ONEDIMENSIONAL PROJECTION OF TWODIMENSIONAL HISTOGRAM

作　　者：李立源[1] 龚坚[1] 陈维南[1] 

机构地区：[1]东南大学自动化研究所

出　　处：《自动化学报》1996年第3期315-322,共8页Acta Automatica Sinica

摘　　要：分析了图像二维灰度直方图上聚类集的形状特征，证明了其存在最佳一维投影。利用此最佳一维投影代替一维灰度直方图进行阈值分割，将会大大降低误判的概率。理论分析和实验表明，新方法比现有直接利用一维灰度直方图和直接利用二维庆度直方图进行阈值分割，的方法均优越。The features of the clusters on 2-dimensional histogram of images are analyzed firstly in this paper. It is shown that there is an optimal 1-dimensional projection of the 2-dimensional histogram. If we use this optimal projection to replace the 1dimensional histogram in the threshold division of an image, the probability of error caused by noise can be reduced greatly. Theoretic analysis and experiments show that this new method is much better than the methods that merely make use of histograms of the image.

关 键 词：图象处理 图象分割 阈值化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]