摘要
在多波束测深趋势面滤波构建趋势面模型的过程中,其模型变量线性相关时,系数阵存在复共线性,因此将主成分估计的方法应用于趋势面模型系数的求解,从而避免了复共线性的影响,最后通过实测数据进行实验,分别利用最小二乘算法和主成分估计求解二次趋势面模型。比较计算结果,验证了主成分估计算法的有效性。
When the model of trend surface filter in multibeam bathymetry is constructed, the coeffcients matrix is multi-collinearity while variables of the model are linear correlation. For the above problem, the principal components estimation is used in calculating coefficients of the trend surface model in this paper, and it avoids the multi-collinearity. In the end, the experiment is made by the survey data. Least squares method and principal components estimation are used in calculating the model of the quadratic trend surface. The result is compared, and it is proved that principal components estimation is much better than least squares method.
出处
《海洋测绘》
2009年第5期5-7,共3页
Hydrographic Surveying and Charting
基金
中国博士后科学基金(20080431342)
关键词
多波束测深
趋势面滤波
复共线性
主成分估计
最小二乘法
multibeam bathymetry
trend surface filtering
multi-collinearity
principal components estimation
least squares method
