摘要
本文提出了一种基于奇异性分析的图像直方图分割算法,由于图像中灰度变化平缓的区域通常和灰度直方图中的尖峰区相对应,该算法根据尖峰的奇异性确定分割出尖峰的左右门限,这种算法还能够有效地消除直方图中的毛刺和小起伏对图像分割的不利影响。模拟实验表明该算法获得了很好的效果,具有一定的推广应用价值。
In this paper, a new method of image segmentation based on singularity analysis is proposed. Because the regions with gentle changes on image is often corresponded to the peaks in gray histgraph. the segmentation thresholds can be determined by analyzing the singularity of those peaks. This method can effectively reduce the side effect of burrs and small heaves. Simulation result, indicates that high performance is reached and this scheme has great value to real applications.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1999年第1期13-18,共6页
Pattern Recognition and Artificial Intelligence
关键词
图像处理
图像分割
奇异性分析
直方图
Iinage Processing, Image Segmentation, Singularity Analysis, Multiscale Analysis
