分段点区间的确定及高精度重构方法研究

ON DETERMINATION OF SEGMENTATION POINTS INTERVAL AND HIGH-PRECISION RECONSTRUCTION METHOD

针对截面数据重构过程中,分段点提取精度不高与分段点区间确定中的人机交互行为,提出一种基于数理统计原理的分段点区间确定办法。首先利用均值平滑方法对直线与样条的截面数据进行预处理;其次基于3σ原则确定分段点区间的右端点,然后再根据相关系数方法确定区间的左端点;最后利用黄金分割法在已确定的区间内搜寻最优分段点。在已确定的区间内进行分段点的提取,不仅可以从现存的点中搜寻分段点,而且可在现存数据点之间寻找更精确的分段点,并且区间的确定方法避免了人机交互行为的出现。实例表明该方法有效确定了分段点所在区间并提高了截面数据分段的精度。

For low precision of segmentation points extraction in the process of cross-sectional data reconstruction and for human-computerinteraction behaviour when determining the segmentation points interval,we present a mathematical statistics principle-based method to determinethe segmentation points interval. First,it uses mean smoothing method to pre-process the cross-sectional data of line and spline; Secondly,based on 3ct principle it determines the right endpoints of segment points interval,and then based on correlation coefficient method it determinesthe left endpoints of interval; Finally,it uses golden section method to search the optimal segmentation points within the determined interval.To extract the segmentation points within the determined interval can search the segmentation points from existing points,besides,it canalso find more precise segmentation points among existing data points,and moreover the method of interval determination avoids the occurrenceof human-computer interaction. Examples show that this method is effective in determining the interval of segment points and improves the precisionof cross-sectional data segmentation.