基于图像序列期望特征的阈值分割方法  

Threshold segmentation method based on expected characteristic of image queue

作　　者：尹沧涛 何平[2] 戴鹏[2] 张洪健[2] 

机构地区：[1]延边职工大学计算机系,吉林延吉133001 [2]哈尔滨工业大学控制科学与工程系,哈尔滨150001

出　　处：《计算机应用研究》2012年第1期376-378,共3页Application Research of Computers

基　　金：哈尔滨科技攻关项目(2009AA1AG064);广东省产学研结合项目

摘　　要：针对单幅图像自适应阈值分割算法复杂度高、实时性不足的问题,提出了一种基于图像序列期望特征的阈值分割方法。以图像期望背景区域面积特征为例阐述了方法原理和实现流程。通过与传统(Ostu)算法对比,论述了方法在处理具有强一致性特征样本序列时所具有的易调试性、实时性优势。结合列车车厢钩尾螺栓图像特征提取实例进行定量分析,验证了方法的有效性。In order to overcome the high complexity and poor efficiency of the conventional self-adaption threshold segmentation mothod,this paper proposed a fast segmentaiton method based on the expected characteristic of image queue.The principle and process of the method was explained by taking the expected area of image background as an example.Comparing with the Ostu algorithm,it discussed the practicabilty and efficiency of the method.Demonstrated by the application of segmentation of screw bolt image of carriage,the validity of the method was acknowledged.

关 键 词：阈值分割 序列 期望特征 背景区域面积 

分 类 号：U216.3[交通运输工程—道路与铁道工程]