基于曲线拟合的交叉分支判别法  

Intersecting Branches Judging Method Based on Curve Fitting

作　　者：余艳梅[1] 罗代升[1] 刁川[1] 

机构地区：[1]四川大学电子信息学院图像信息研究所,四川成都610064

出　　处：《四川大学学报（工程科学版）》2009年第1期158-161,共4页Journal of Sichuan University (Engineering Science Edition)

基　　金：四川省科技攻关资助项目(05GG021-026-03)

摘　　要：为了能够正确识别在石油地质等行业采用的曲线图中大量存在的交叉曲线,根据图像理解的思想,提出一种交叉分支判别算法,采用组合择优法求得最优拟合组合。选取各分支特征点,对不同分支进行两两组合,并用最小二乘法拟合出各种组合的曲线,计算并比较各种组合情况下拟合曲线与特征点间的偏差值,偏差值最小的分支组合即为所求最优解。实验证明该算法效果良好。To correctly recognize the intersecting curves in curve graphs used in many industries such as petroleum geology,image understanding theory was applied to judge which two intersecting branches belongs to the same curve,and the method of choosing the optimum combination was used to get the best fitting group.First the characteristic points of each branch were chosen.Then the branches were combined as different groups.Next the method of least squares function was used to fit the curve of different combination...

关 键 词：交叉 图像理解 曲线拟合 组合择优 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]