基于最大类间方差法与Mahalanobis距离的IC芯片彩色图像多阈值分割技术研究  

Study on Multi-Threshold Dividing Technology for Chromatic Image based on Maximum Variance between Clusters and Mahalanobis Distance

作　　者：梁忠伟[1] 

机构地区：[1]广州大学机械与电气工程学院,广州510006

出　　处：《组合机床与自动化加工技术》2009年第3期33-39,共7页Modular Machine Tool & Automatic Manufacturing Technique

摘　　要：针对彩色IC芯片图像的分割问题,提出基于最大类间方差法与Mahalanobis距离的多阈值彩色图像分割方法,通过建立最大类间方差数学模型,对彩色图像Mahalanobis距离进行最优分割距离阈值进行求解,并在计算目标点与待处理点的Mahalanobis距离并确定待处理点所从属的分割阈值区间基础上,对像素点赋予伪彩色值,从而进行图像彩色区域的确定和划分。通过编程处理,证实此方法对实际彩色图像进行分割处理后能获得稳定准确的分割效果。In order to solve difficulties in chromatic image＇ s segmentation, a new method of multi-threshold chromatic image＇ s segmenting technology is proposed. By determining maximum variance of clusters and optimum segmentation distance thresholds of chromatic image＇ s M ahalanobis distance, bonding with calculating Mahalanobis distance between object and armed pixels, afterwards labeling each pixel a pseudo color value, each specific image region is identified and projecting displayed. Developed program running result indicates that by this method, clear segmenting image result is gotten.

关 键 词：彩色图像 MAHALANOBIS距离 最大类间方差 多阈值分割 

分 类 号：TG65[金属学及工艺—金属切削加工及机床] TP391.4[自动化与计算机技术—计算机应用技术]